Michel Biezunski
Our brain works by associating thoughts. A thought can be expressed as a set of subjects. We communicate with others by talking about subjects in conversations. Our mental representation of what these subjects mean does not uniquely relate on the name we use to designate them. The distinction between the "signified" – the meaning of a concept – and the "signifier" – the symbols used to express it – is at the foundation of modern structural linguistics and semiotics. The name is purely a label that we use as a shortcut but names are different depending on the language we speak, a name can be used to mean things that have nothing to do with each other (homonyms, context-dependent names), and a thing can be labeled with different, equivalent names (synonyms).
The way we understand what a subject means is complex and is a major challenge to overcome when using computers to manage knowledge-based systems. It happens during conversations that the most important things we communicate is not something that we say, but something that is implied to be understood either by the tone we use to say it, or some gesture that communicates something important not conveyed by the words themselves. Computers are not well equipped to deal with ambiguity, misunderstandings, context-dependent information, or non-verbal communication. They can handle correctly the low-hanging fruits, which represent a significant part of information, but they do not cover the entire spectrum of human communication.
Knowledge systems, including computerized taxonomies and ontologies, usually require an agreement on a name used to identify a concept. These names are supplemented by metadata that describe the inherent characteristics of the object under consideration. These representations work as long as the concepts, and the categories to they belong, are well-identified, but they may fail to grasp the concepts described, as time goes by, as when new subjects are introduced and the distinction between previously well-described objects becomes blurry.
Topic maps provide independence between the data that are being described and the subjects that qualify them. For example in a book index, there is no requirement that the term in the index be identical to a word used in the page. What matters is the fact that the abstract concept used in the index corresponds to a subject that can be derived from reading the content located at the corresponding page. A topic map is an overlay which provides a perspective on the knowledge that is described. The Topic maps paradigm consists of the ability to point to any kind of data from outside, without the constraint of having to rely on what is already there. In fact, several perspectives can be created on top of the same data set, that can provide filtered information aimed at different audiences.
The Topic Maps paradigm aims at providing a model that enables these ideas to be processable by computers. The abstract subjects are represented by proxies, which are compound objects with properties. These proxies are called topics , to indicate that they represent a subject as a location (from the greek word topos ) in a semantic space called a topic map. They are nodes —vertices—, in a graph that can be navigated following the connections — edges— between them or to the documents serving as sources.
The Topic Maps international standard provides a way to interchange topic maps between applications. By design, it does not prescribe any way to specify topics nor any particular semantic for their relations.
The Topic Maps model is based on establishing one location per subject. A subject is any subject of conversation, which can represent an actual thing, or an abstraction, a relation, a point of view, etc. The subject exists independently of the names used to designate it.
A topic is a computer representation for a subject. A subject is defined for human consumption by a "subject indicator" which contains an unambiguous definition. It is identified by a computer as a unique object by a "subject identifier".
When an information resource is considered a subject, it is referenced by a "subject locator", indicating where it can be found. For example, if the resource is available on the Web, its subject locator is a Uniform Resource Identifier, or more accurately an IRI (Internationalized Resource Identifier), using Unicode characters instead of just the ASCII character set.
A topic has inherent properties, including its names, occurrences and associations.
There can be multiple names for a topic, and, since the name is not uniquely identifying a topic, one given name can be used in different topics. The "scope" attribute is used to delimit the domain in which certain properties are applicable. For example, a scope can be used to specify a scientific domain in which a name is valid, the language for that name, or to disambiguate homonyms. The scope can also be used to define contexts for relationships and occurrences.
When a topic applies to an information resource, the relation between the topic and the resource is called an occurrence. In other words, the subject for this topic occurs in this particular resource. An occurrence can be typed, and scoped, if needed.
Topics can be interconnected, and the relations between topics are called "associations". Each association belongs to a type, and the role that every topic plays in this association is described by a role property. The role property has a player (the topic) and a role type (the nature of the role played).
Since the purpose of topic maps is to provide a single location per subject, regardless of the name(s) used to describe it, the procedure for merging topics is precisely defined by the data model. When two topics merge, their properties are added, except when they are redundant.
In order to improve interoperability between Topic Maps-enabled applications, some relationships are defined, but are not mandatory. These relationships include the type-instance relationship, the supertype-subtype relationship. In addition, a special variant of name called "sort name" is used to indicate strings that are used when sorting the names.
The Topic Maps international standard does not prescribe how a subject should be defined. This empowers the creators of a topic map with the ability to choose the conceptual basis on which they establish subjects.
A subject can be anything whatsoever, regardless of whether it exists or has any other specific characteristics, about which anything whatsoever may be asserted by any means whatsoever. In particular, it is anything about which the creator of a topic map chooses to discourse." http://www.isotopicmaps.org/sam/sam-model/#d0e746
The rationale for this definition, which in practice amounts to the absence of a definition, is to make topic maps the ultimate merging space for information coming from a variety of sources. And the reference to the process of creation of a topic map also indicates that topic maps can be used to capture information that was created by humans as well as by machines.
The Topic Maps model is a knowledge representation aimed at describing units of meaning and their connections. It has been designed to capture the traditional navigational aids, such as indexes, thesaurii, glossaries, catalogs, dictionaries, tables of contents, cross-references, bibliographical references. Indexes are a list of terms–topic names–presented in alphabetic order together with the locations where they are relevant–occurrences: in printed books, occurrence indicators of topics are page numbers. Thesaurii are topics connected by a predefined set of relationships: generic (related), or hierarchical (broader/narrower). Glossaries are a list of topics followed by an explanatory, which is an occurrence of the topic which plays the role of definition. Tables of contents are list of topics whose occurrences are the chapters, or sections, in which they occur. References are links to other occurrences of the same topic: they are called cross-references if they point to another location in the same text, or bibliographic references if they point to another text which needs to contain a way to locate it, i.e. its author, title, date of publication, or URL.
Taxonomies are classification systems, based on assigning categories for each item. Categories are often organized into a hierarchy. For example, a taxonomy of living creatures could consider animals as living creatures, mammals as animals, cats as mammals, tigers as cats, etc. A hierarchical relation is a relation that can be expressed with a statement in which the middle term is "is a": an animal is a living creature, a mammal is an animal, a cat is a mammal, etc. This relationship is sometimes considered to be a "class instance relationship".
The description of the ways categories are organized is called a schema. A database schema, or an XML schema, does not only describe the containment rules, but also the data types, and additional constraints. For example, a valid date could be defined as a positive integer not higher than the current year. Schemas complemented with rules and constraints are referred to as ontologies.
Hierarchical relations can be expressed as associations between topics representing items and topics representing categories. Therefore topic maps encompass taxonomies and ontologies, with the addition of rules and constraints in a separate layer. The Topic Maps model is itself an ontology, in the sense that it is made of constructs, such as "topic", "name", "association", "occurrence" that have rules attached. But it is an ontology capable of representing other ontologies. The exact nature of the constraints in the Topic Maps model, and the levels to which the rules applied has given rise to extensive discussions, and there are several possible interpretations of how to use the Topic Maps ontology. For example, there are contexts where topic names should be considered topics, associations should be considered occurrences. An extensive practice of topic map implementations shows that the power of the Topic Maps ontology is its flexibility, whereas other ontologies are adopted because of the constraints that they provide.
A relation type can itself be considered a "topic". The ability to consider relation types as topics enables integration between datasets coming from heterogeneous sources in which the schemas use different representations for certain types of relationships, but are all considered equivalent in an integrated environment. For example, relations expressed as follows: "is a", "instance of", "belongs to", "narrower than", "component", "child of" could all be seen as a generic hierarchical relation type. But this can be fine-tuned depending whether it makes sense in a given application environment.
Furthermore, an instance of a given relationship can be reified, i.e. it can be considered a topic per se: the assertion "London is a city" which contains relations between three topics ("London", "is a" and "city"), is itself a topic. It can therefore serve as a node in an association with other topics. The fact that London is a city has already been attested during the first century before AD is an association between the topics "London is a city" and "First Century before AD" whose semantic is represented by the role "attested in" (itself a topic).
The Topic Maps paradigm obliterates the traditional distinction between data and metadata. The classic distinction is based on the view that data represents the content of information, while metadata adds contextual information about the information. In a taxonomy, metadata is the category to which data belongs. Metadata also serves to add information outside of the domain of the content. A book on the history of painting is about art, but it also has an author, a publisher, a date of publication, all considered metadata, because they do not refer to the subjects in the content of the book. The distinction between data and metadata has become even more acute when information is stored on computers using relational databases, where tables have a header (or a field name) that establishes the field and cells containing the data.
The principle of separation between structure and content used by markup languages such as XML or HTML, separates the data content from the metadata encoded as tags. In a library, a catalog is metadata, while the books and other source materials are considered data. But this distinction is not as clear as it seems. For example, a library catalog notice which is technically part of metadata includes the name of the author, considered data content inside the book. Inside the book, though, the author name is considered to be part of the data, since it is part of the title page which is part of the content. This illustrates the fact that someone's data is someone's metadata. The same information seen in the perspective of Topic maps is described as a graph of interconnected topics. The fact that there are no constrinats on what the semantic of the relationship represents, results in an unlimited number of possible connections between information items. Flattening is of paramount importance for data integration. It can be used not only to collect information from a variety of resources, but also to keep track of the provenance information, by creating topics and connections to the original repositories characteristics from which integrated information comes from.
A topic map can be construed as an added information layer, which, instead of being aimed at replacing a data structure, points at it in order to shed a new light on its components for usages that were unforeseen at the time the data models were created. It can be used also for forensic analysis, containing not only the provenance information, but also the processes through which the information came into existence. Once the information exists as a topic map graph, it can be displayed in a variety of ways to provide custom perspectives, and it can be recalculated in a variety of ways to provide new visualizations including new combinations of existing elements, as in a kaleidoscope.
The essential distinction can even be pushed one step further by describing processes as relationships, using the same underlying structure. Robert Glushko writes:
The vanishing difference between data and metadata, and the equivalence between a semantic relationship and a process, opens a whole range of applications for integration of information coming from very diverse sources, as well as the ability to conduct forensic analysis by describing processes that explain the provenance of potentially any information item.[Robert J. Glushko, The Discipline of Organizing: Professional Edition, O'Reilly, The MIT Press, 2013, section 1.3: The Concept of "Resource"].
The equivalence between a semantic relationship and a process is not something has been explicitly part of the Topic Maps paradigm, but it is a consequence of it that has been developed further in a model called the "Data Projection Model" . In this model, the arithmetic expression "2 + 3 = 5" is seen as a relationship between two topics, "2" and "3", with the relationship being expressed as an addition, abbreviated with the "+" sign. The equal sign can be expressed in two different ways in a topic map context, either as an association between the reified relation "2 + 3" considered a topic and the topic 5, associated through a relationship whose semantic is the equality. It can also be interpreted by saying that the topic "2 + 3" has an alternate name, i.e. "5".
Information content is inherently complex and ambiguous. The assertions "London is a city" and "the City is in London" have a completely different meaning, which is expressed by the capitalization of the word "city" that represents respectively a generic term to describe a group of people living together in a dense area, and a unique term coined to designate the Financial District in London. In addition to the fact that humans speak different languages, the polysemic structure of a human language, results in differences in understanding depending on where people come from, their education level, and even the mood in which we are or the tone in which we speak, that can alter the meaning of what we say. In writing, ambiguities still subsist, and misunderstandings happen. The use of computers to handle information leads to denying that ambiguity exists. As computers are not inherently able to cope with these kinds of complex or ambiguous situations, the way information is stored tends to deny these multiple levels of complexity, forcing to a simplification of interpretation. Although most of times, especially in a professional context, there is an implicit agreement on what things mean, there is always a chance that the meaning is missed, especially when a considerable amount of information is collected.
By going Semantic, the World Wide Web Consortium has moved in a direction of trying to connect knowledge based on common understandings of concepts, best exemplified with the Linked Data architecture, where every imaginable topic gets an addressable web address (URL), and therefore can be referenced by others. The main problem with this approach is that by setting the application span so large, it diminishes the value of the information retrieved, because it's often taken out of the context in which it is queried.
The potential inherent to the paradigm represents a long-term, enduring, applicable functionality that may well persist long into the future, regardless of the formats used to interchange information.
A database is a storage system in which data are structured in order to facilitate queries based on common criteria. A relational database model has a structure similar to rows in a table, with well-defined column headers. Each column header represents the type to which a specific data (cell) belongs. A database schema contains the definitions of the fields allowed as well as the data type to which they are required to belong, in order to provide validation. In addition, relational databases provide the ability to join tables by fields. For example, an invoice table can be related to a customer name, which is an entry in another table.
Object-oriented databases consider that data are stored as objects with properties, for example the data type: a "person object" may be assigned a "date of birth" property that is constrained to be an integer with four digits. The advantage of object-oriented databases is that objects can be reused across several higher level objects.
Graph databases store data as nodes and edges (links). The edges have their own properties, for example bidirectionality. The main difference between these three kinds of databases — relational, object, and graph — is the interface they provide to their users. In a relational database, the records are data corresponding to predefined fields, optionally joined to other records through foreign keys. In object databases, a related object could be described as a property of the source object, and in a graph database, the relationships themselves are considered first class objects and can be addressed as such. Performance issues, query languages, reusability and familiarity with the paradigm are factors that are taken into account while deciding which data storage is best fitted for a given environment.
Topic maps can be implemented using either of these database technologies. A topic can be defined as a record in a relational database object, as an object in an object-oriented database, and as a node in a graph-oriented database. The way a topic maps system is implemented does not necessarily impact the user experience, except for constraints that are imposed by an implementation's specific requirements. When a topic maps-based system contains export functionalities, it is posssible to use the data coming from another system regardless of the internal database on which it is built.
Managing knowledge has been performed by human beings, long before computers were used. Therefore, a subject matter experts have accumulated lore to do this, and the way they operate is somewhat different from what computer systems require. Recently, subject matter experts have been learning how to use technology to do their work. And technologists still have to learn about the specific requirements needed by knowledge experts.
Since a long time, subjects are used to describe contents, either in classification systems such as library catalogs, or at a finer granularity level, in book indexes: a book record in a library catalog is about one subject, while an index entry refers to a subject in a given location in the book, usually indicated by the page number. The way subjects are created may also greatly differ, on a conceptual level. Isolating a concept by naming it and establishing its subject are not easy tasks. This is at the core of philosophical and epistemological theoretical world views. The library and information science curricula consider this at the core of their teachings. Practically speaking, building agreements within a community that has vested interests in sharing certain well-defined concepts or products is a complex task. And it not guaranteed to be stable over an extended time period, as new information may break the previous consensus with the introduction of new, unhearded concepts. A comprehensive review of these complex matters is presented by Birger Hjørland in the article "Subject" of the encyclopedia http://www.isko.org/cyclo/subject , who introduces a distinction between the conceptual analysis and the translation stages, i.e. the assignment of the applicability of subjects to indexing and classification.
The transition from print-based technologies to digital-based technologies has modified the scale of knowledge management, and has triggered profound changes in the way knowledge management is handled.
Several attempts have been done, and continue to be done, to circumvent the inherent difficulty of qualifying the semantic of information items by using algorithms to replace human determination of subjects by automatic processes. The availability of the Internet as a global network and the World Wide Web as a universally accessible knowledge repository offers an opportunity for connecting concepts at a scale that changes the qualitative nature of the landscape. At the scale of one book, recording all words into an index and pretending that the value of such an index is equivalent to a humanly crafted index has been ridiculed. Just getting a concordance table of all words used is interesting for statistical purposes, and is useful in comparative literary studies, to establish the frequence of usage of certain words by certain authors. However it doesn't provide a way for most readers to view and access the major concepts in this particular corpus.
It is a completely different story when the amount of information is so massive that no human being will ever be able to analyze and process the trove of available data. Because of the availability of massive data, it is possible to create algorithms that can scan all of it, and be refined with multiple criteria to yield to results that are much more meaningful, and therefore much more usable than algorithms that were applied on a much smaller amount of data. Artificial intelligence techniques, including the fact that machines contain "learning" abilities that take into account new data to dynamically reconfigure themselves are changing the landscape. The search algorithms used to browse the World Wide Web are based on such techniques, although they are also supplemented by human work to solve some of the issues that were bypassed by the algorithms.
The philosophical and epistemological preconceptions used to determine how knowledge "presents itself" still exist, but they are now hidden into algorithms that are not accessible directly by human users. As knowledge consumers, we are left with using the results of processing, with no ways to understand why and how the information we see is what it is. But it is not because this information is not visible that it doesn't exist. It only means that if we want to understand what we see, we need to dig deeper into how why this information has become what it is. In a book index, we rely on one individual's choices, and we may disagree with the choices that were made, and we may think we would have done it differently. Nevertheless, we give credit to the author of the index. On a massive scale, unless we can see the algorithms used to produce the information, we lose our ability to offer a different perspective. Worse, we will never know what we do not know. There may be some important information available that was missed by the search algorithm, but we won't know it. However, these are cases where we do know, if either we are experts in a domain or if we are in charge of managing a well-known data repository.
The Topic maps paradigm can be used as the conceptual basis on which forensic analyses can be conducted and provide methods for auditing content.
Topic Maps originate from the community involved in promoting generic markup in the publishing industry. When computers started to be used to publish books and other printed materials, the idea of describing the semantic nature of a textual fragment rather than the way it should be formatted became the founding principle of generic markup. Furthermore, instead of providing a fixed set of predefined tags, the idea of letting users define themselves the elements they needed to use in a given application context was considered an important advantage. Naturally, communities of users could agree on a set of shared elements, so that each individual member or company in a domain would not have to reinvent the wheel. The precedent of handling a set of data (database) using a model with pre-defined fields and characteristics was extended to handing textual content (document) that complies with a pre-defined structure.
The Standard Generalized Markup Language (SGML) originated from the Generalized Markup Language (GML) used by IBM for its technical documents, and was published in 1986 by the International Organization for Standardization (ISO). It defines document structuring, and validation mechanisms to parse document instances against a document structure. Element names appear in a document surrounded by angle brackets. Although this notation can be changed, the angle bracket notation is the visual symbol for SGML documents.
As the World Wide Web was in its inception, hyperlinks became more important, with the emergence of the Hypertext Markup Language (HTML), a library of tags using angle brackets like SGML.
The work continued to extend the methodology emanating from standard markup to time-dependent information and hyperlinks. The HyTime standard (Hypermedia/Time-based Structuring Language) was published in 1992 and was considered at the time as a potential successor to SGML.
The rapid rise of HTML and the associated browser technologies showed that, more than the strict compliance with generic markup standards, success came from the ability to create a gigantic hyperlink-based network that connected documents accessible through the Internet via a simple HyperText Transfer Protocol, known as "http". As the World Wide Web started to expand and change everything, one major feature was missing, i.e. the ability to qualify the nature of the relations. This requirement turned out to become the basis for what would a few years later emerge as the "Semantic Web". Two different approaches were taken that would eventually become the Resource Description Framework and Topic Maps.
In contrast with the web hyperlinks — unidirectional links pointing to a specific location — the publishing industry was looking for ways to describe the structure of richer connections. A consortium of Unix vendors was under pressure by its customers to harmonize the vocabulary used by the various tools providers, and initiated a study group looking for ways to design interoperable indexes in the technical documentation of their products. This group, called Davenport, was trying to match the requirements of the publishers with the new features available in the newly established standards.
A traditional index, generally located at the back of a book, contains a list of entries pointing to locations in the content of the book relevant to the concepts expressed by the entry terms. The Davenport Group was concerned by the applicability of its findings. A description of an index, as part of a document, should be able to describe entries as structured paragraphs, made of a term, and a locator (usually a page number). In addition, internal links inside the index, known as "see" and "see also" could also be described by regular link structures.
Animated discussions started in the group, as it was joined by members of the standards community who were promoting the use of complex hyperlink models which were part of the HyTime standard. HyTime has a hyperlink module made of several models. One, called "independent link", was of particular interest, because, rather than being embedded in a document which is the origin of the link as in HTML, this link has multiple targets pointing to various locations. Therefore, it is created and maintained independently of its anchors. The idea that it was possible to manage links independently served as a starting point to define a topic as an object independent from its source, and more broadly to define an architecture in which information can be qualified as a superimposed layer. It opened the possibility to create an architecture that was independent from the sources. In this perspective, an index could be seen as a set of terms, created independently of what the text contains, that qualified the subject of text fragments (for example, pages), and this approach could be generalized to other situations where the semantic qualification of information needs a certain degree of freedom towards the sources to which it applies.
These two approaches on how to handle indexes had their own merits, but were technically very different. Acknowledging this situation, members of the Davenport decided to split the work into two groups. One group created a structured markup schema specifically adjusted for handling technical documentation, which included a whole model for documents, not only indexes, and they created what became the Docbook document type definition, which has been widely used since in the world of technical documentation. The other group worked on what would become known as topic maps.
The concept of independent linking was evaluated as a possible abstract model to capture the essence of indexing. After a few years of intense brainstorming, we arrive to the idea of a topic, which was an abstract concept, that could be represented by multiple names (even by no name at all), pointing to various locations in external resources that were relevant to it. Whether the string of characters that represented the term was present or not was irrelevant. A portion of a document could be about a given subject without mentioning the name under which the subject was qualified. This characteristic is also present in the book indexes. The indexer is creating a qualifying name for a concept that applies to a particular section of the book without requiring that this name would be present.
In 1995, after having tried several possible models, we came to realize that a topic could be a computerized instantiation of a subject, represented by a HyTime-based independent link construct. In addition, we wanted to be able to navigate between topics that could be related together, and created the notion of "associated topics" that collectively comprise a network graph.
As this model was applicable far beyond interoperable indexes for technical documentation, we proposed to the ISO technical group that was responsible for SGML and related standards to make it a generic standard.
ISO accepted the project, and we started the standardization process under the name "Topic Navigation Maps" (TNM). The name was later simplified and the standard became simply known as "Topic Maps". Its first edition was published in 2000.
As we were working on developing the Topic Maps standard as an application of HyTime, the markup community was working on simplifying SGML and removing many of the features that were used only rarely, and were particularly challenging to implementers. A new standard, the eXtended Markup Language (XML) was published in 1996 by the World Wide Web consortium and became widely adopted. Many SGML applications were converted to XML applications. Soon after the publication of the first version of the Topic Maps standard, we worked on an XML version, which was published in 2001 under the name XML Topic Maps (XTM). We dropped the reference to HyTime, and used the XLink specification instead, which came out as part of the XML family of new standards. The way it was used was not significantly different from the simple hypertext links in HTML.
Concurrently to the work done on Topic Maps, the World Wide Web consortium was working on a graph representation of hyperlinks between web resources. The Resource Description Framework (RDF) was published practically simultaneously as the first version of Topic Maps and become the core of the Semantic Web. The data represented with RDF was stored as "triples" (subject-predicate-object), with the distinct feature that the resources were represented by a unique Uniform Resource Identifier (URI). RDF is used to represent metadata in several reference initiatives, including the Dublin Core, which has been adopted by the community of librarians to represent online resources. Topic Maps and RDF have in common the fact that data is represented as a graph. But the focus of the RDF team was on ontology processes, and the ability to automate the creation of knowledge representations, while the Topic Maps team was focused on providing ways to create human-created index-like graphs.
While the standards were in their inception, there were contacts between the teams and a willingness to cooperate, but they were not frequent enough to have resulted in a common, or unified approach.
An analysis of guidelines for RDF and Topic Maps interoperability shows that there are several approaches being considered: semantic mapping, object mapping, and hybrid. Subjects in Topic Maps are considered equivalent to RDF resources. Topic Maps allow for n-ary associations whereas RDF triples amount to binary relationships. But it is possible to envision a set of mapping rules to translate automatically one notation to the other (Presutti V., Garshol L.M., Vitali F., Pepper S., and Gessa, N. 2005).
At the same time, the Semantic Web project, based on RDF, and Linked Data, was also showing signs of decrease in interest. The agreement on a universally valid URL-based concepts is still very much alive, and serves as the foundation for libraries and open data exchange in general. But the learning curve has become a deterrent, as more and more technical layers were added.
In the industry, the willingness to share information was not a priority. Most of the times, the purpose is to share information internally. When only one topic map application is present, there is no need for an interchange format. As companies look at organizing their internal information repositories in the most efficient way, Since most of the topic map tools were designed to facilitate the XTM format, they fell out of grace, and many built-in topic maps applications were created.
The origin of topic maps was an attempt to bring the connected information, then called "hypermedia" to the world of publishing, by leveraging the traditional navigation systems, such as book indexes, library catalogs, thesaurii, dictionaries, etc. The analysis of these interconnections as a graph structure was the major breakthrough accomplished by the topic maps designers. Then came the Web and the Semantic Web, which aimed at relating online resources by their semantic properties, in order to facilitate access and automated processes. That was the origin of RDF, the Resource Description Framework, and the Dublin Core, a metadata set aimed at library cataloging, and later the Linked Data project. Every topic is represented by a unique Web address (its URI) and is accessible by many applications.
The "Linked Data" project, that is part of the Semantic Web project by the World Wide Web Consortium, aims at connecting data available on the Web through relationships. The Open Linked Data format is used by many libraries in an effort to create cross-library accesses to materials. It is based on the Dublin Core, which is a metadata format itself based on RDF that was created in 1995 in a meeting in Dublin, Ohio which was attended by members of the RDF team as well as the Topic Maps team.
The need for a more flexible organization of data has persisted, and even increased. The gradual rise of graph databases and the relatively new interest around knowledge graphs has revived the need for concepts similar to those of the Topic Maps model, bypassing the need for a common XML-based syntax. Other standards for exchanging graph-based information exist or have emerged. One of them is RDF ^[Resource Description Framework, a W3C Recommendation (1999)], also on the decline, due to its verbosity and complexity. The standard that is looking most promising today to exchange graph data is called "Property Graphs" and it is based on key/value pairs for nodes and edges in the graph. Because of its generic nature, it can serve to transfer data from an environment to another. It doesn't contain the distinctions made in the Topic Maps about multiple names, nor does it consider the relation semantics as topics, but it can be used for interchange purposes between graph databases.
The Knowledge Graph, which appears on many Google search pages, comes from Freebase, a knowledge base created by a company called Metaweb, that was bought by Google in 2010, and was inspired by Topic Maps ^[Private conversation with Veda Hlubinka-Cook, co-founder of Metaweb].
In 2010, Google coined the term "knowledge graph" to describe a knowledge base they acquired, Freebase, from a company called Metaweb, which has designed this base using the concepts from topic maps, to semantically connect topics together.
ISO/IEC 13250:2003 — SGML applications — Topic maps
The initial version of ISO/IEC 13250, Topic Maps, was published in 2000. The standard is defined as a set of architectural forms, a set of templates from which a document type definition can be written. Architectural forms are a feature of HyTime, the Hypermedia/Time-based Structuring Language (ISO/IEC 10744:1997) containing far-reaching hyperlinking and addressing facilities. HyTime has been used as a source for other standards or concepts, such as the Document Object Model used in HTML and XML.
A simplified version of the interchange syntax, written for XML and called XTM — for XML Topic Maps —, was published in 2001 by an organization called topicmaps.org . It was submitted to ISO for integration into the standard.
The second edition of the Topic Maps standard was published in 2003. It reproduces the content of the first edition, and adds the XML representation of the topic maps architecture, in a document type called "XML DTD for Web-oriented Topic Maps", which is identical to the previously mentioned XTM (XML Topic Maps) and was already started to gain significant momentum. The HyTime-based notation for Topic Maps, now called HyTM, is still present. As Topic Maps were applied and implemented, the HyTM model was practically never used.
The usual process of transforming an SGML document type definition into XML is straightforward and doesn't require any major change. However, in this case, HyTM was not a document type definition but a set of architectural forms, i.e. a set of templates that would serve as a basis to create document type definitions. Consequently, significant changes between the two versions are worth noting: XTM introduces a distinction between two kinds of subjects: those which are addressable online and those which are not addressable. The links used in XTM are based on "simple xlink" instead of Hytime "varlink". These are links that are very similar to the universally used "href" attribute in HTML, and therefore facilitate implementation of topic maps on the Web. Finally, the notion of facets, which were designed to qualify properties for a subject, have been discarded, because they could be easily represented by using topic associations.
The new version of the standard facilitated the adoption of Topic Maps through the XTM notation.It is more concise, straightforward, easier to understand by implementers. In 2013, a new version of XTM, 2.0, was introduced (see below).
After the publication of the second edition of the Topic Maps standard, work continued. The standard was supplemented with new parts, that were intended to clarify how to interpret the existing standard (data model) and provided new languages to either express or process topic maps.
The definitions of the concepts were later reorganized into part 2 of the standard, under the name "Data Model" and the XTM representation of Topic Maps was later reorganized into part 3 of the standard, under the name "XML Syntax".
ISO/IEC 13250-2: 2006 — Topic maps — Part 2: Data model
The data model describes the properties of the constructs in the XML version of the Topic Maps standard. It specifies data types for the constructs present in the XTM document type definition, and documents them with UML diagrams. The merging operation procedure is described in detail. Core subject identifiers are defined. These subjects identifiers are given URLs and are available as "public subject identifiers".
ISO/IEC 13250-3: 2013 — Topic Maps — Part 3: XML Syntax
The XTM structure for topic maps was revised in 2013 and is now known as XTM 2.0. The main motivations for this revision were to remove features that were practically never used and added complexity, mainly without loss of functionalities. For example, the reference to XML Base was removed, and support for XLink was removed. ^[The differences are documented in http://www.garshol.priv.no/blog/85.html .] Other differences are changes in the names of the elements and attributes, removal of wrappers. Several new functionalities were added, such as the support for typed data, the ability to declare types for topic names, and a change in the way reification is indicated.
ISO/IEC 13250-4: 2009 — Topic Maps — Part 4: Canonicalization
The purpose of the Canonical XTM format, or CXTM, is to ensure that two instances of topic maps which conform to the Data Model are serialized identically byte-by-byte, in order to allow the creation of test suites to test topic maps products. Its features include the order in which the various elements should appear, and the requirement that empty lists are used, wherever applicable, even when they contain no item. In addition, the encoding of the documents should conform to Unicode Normalization Form C.
ISO/IEC 13250-5: 2015 — Topic Maps — Part 5: Reference Model
The Topic Maps Reference Model adds one level of abstraction above the data model. It considers the Topic Maps Data Model as just one instance of a more general model of a "subject map". A subject is any abstract item of conversation that is represented by a computer proxy, which consists in a declared set of key/value properties. A subject map is a finite set of proxies. The key/value properties are more general than the defined properties for a topic in the Topic Maps Data Model, which is one instance of what can be represented with the reference model.
However, the Reference Model defines two types of relationship, "is a" (instance of) and "sub" (subclass of) that can be used for inferencing. A number of primitive navigation operators are defined, that return a set of keys relative to a given proxy. Maps are defined by sets of constraints. Subject merging and clues for interpreting a subject are also defined.
ISO/IEC 13250-6: 2010 — Topic Maps — Part 6: Compact syntax
The Topic Maps Compact Syntax is a text-based notation that provides a light-weight alternative to the XML Topic Maps (XTM) notation. Its purpose is to be useful for manually authoring topic maps, provide human-readable examples in documents, and service as a syntactic basis for the Topic Maps Constraint Language.
ISO/IEC 19756: 2011. Topic Maps Constraint Language
As the topic maps defines a structure of data which is generic, there is a need to further constrain specific implementations. This opens the ability to validate a topic map according to specific semantic constraints valid within a given environment. For example,
works-for isa tmcl:association-type
expresses the fact that "works-for" is an association type. The Topic Maps Constraint Language contains features that enable the ability to follow an association, optionally with roles of a certain type, establish whether a topic is a subtype or an instance of another topic, and locate occurrence values. There are two types of validation rules, those that apply to an individual topic, and those which apply globally to the whole topic map.
Several implementations of topic maps software contain notations for querying topic maps. Queries play an important role in what users expect from their topic maps-based system. A substantive amount of work went on to harmonize these languages into a standard. But the work has not been completed yet.
The proposed query language contains a notation to query topic maps and defines how processors should behave. The latest draft available at the time of publication of this article lists the navigation axes that are allowed in a topic map environment (types, players, roles, etc.) as well as traversal results (the result of a forward traversal), reification (which expresses the topic emerging from a construct such as an association between topics), and atomification (which describes a way to convert a complex construct in integer or strings). Furthermore, the Query Language defines comparison and ordering of tuples. It defines a notation for querying a topic map. For example:
select \$p / name where \$p isa person & lives-in-city (being : \$p , city : \$\_)
finds all person names living in any city.
An attempt was made to define a graphical notation, inspired by UML (Unified Modeling Language), for Topic Maps.
Topic maps have been adopted in commercial applications as well as in the academia. In the initial phase, topic maps applications have been mostly developed in Europe and in North America, before taking off in Asia, most notably in Japan, Korea and in the recent years, in China.
The interest for topic maps took off in 2000, the year when the standard was initially published. It gave rise to an increasing number of works in the academic world, where it continued to grow until 2008. The number of publications on topic maps decreased since then, but the visibility of topic maps has decreased since then and in recent years a renewed interest for topic maps has emerged, most notably in Asia. However, beyond the publicly available sources, several projects were developed, and still are, that refer to topic maps. Some of them are confidential, and others have simply not publicized enough to be visible.
One of the biggest knowledge bases using concepts inspired by the Topic Maps paradigm was Freebase, developed within a company called Metaweb, which was acquired by Google in 2010, and now appears on the search pages as the "Google Knowledge Graph". The US government made uses of Topic Maps, including by the Department of Energy, and the Internal Revenue Service. In Europe, city portals were created using Topic Maps. Norway has been an active center of development for various topic maps applications. Topic maps applications continue to be developed in East Asia, including Japan, South Korea and China.
In the academic world, many research projects have been conducted using Topic Maps, included those funded within the European Community supporting the Semantic Web initiative. Topic maps have been a recurring subject during markup conferences, especially between 2005 and 2010. A research laboratory dedicated to topic maps was created in Leipzig, Germany, and the Topic Maps Lab organized annual conferences dedicated to research and applications in the domain.
When topic maps were created, the web was still in its infancy, and the amount of information available in search was far smaller than it is now. Search technologies were for a large part based on string recognition, and information owners had an idea of the extent of the data they were dealing with. In that period, the need to organize information was considered a high priority. As the amount of information available started to grow, and the concept of "big data" started to emerge, the science of data analytics became prominent, and the use of techniques based on automated processes, labeled under the term "artificial intelligence" took the lead on more traditional information management techniques. Refinements in the algorithms, machine learning features, and improvements in the quality of search results have provided an alternative to a fully human-driven classification of information.
Many applications are not strictly based on topic maps, but address a similar goal, which is to organize information into a network of topics.
Wikidata, which is the knowledge base that feeds Wikipedia, has a structure which resembles a topic map, although it doesn't directly reference it as such.
Many applications have been created by direct reference to Topic Maps at Columbia University (New York), New York University, the American Geophysical Union, RILM (Répertoire international de littérature musicale), the US Department of Energy, the US Internal Revenue Service, the European Community, the Norwegian government, publishing companies in Netherlands and the United States, by industry manufacturers in Germany, and in many other countries.
Presentations were delivered in conference series specifically devoted to Topic Maps, in Oslo, Norway and Leipzig, Germany, and a conference called the "Extreme Markup Conference" (now called "Balisage"), devoted to leading edge technologies in the XML universe, including topic maps. The table below contains the titles of conferences where topic maps-related presentations have been delivered.
This table is not exhaustive, but it gives an idea of the context in which topic maps applications have been developed, at least the ones which were publicly exposed. It shows that the peak of topic maps activity is in the year 2008. We also can see an evolution from mostly research-related activities to an increasing number of production-related activities in the later years.
Between 2017 and 2019, several conferences have been devoted to knowledge graphs, in China, Singapore, Cuba, Italy, United States (Texas, New York), Australia, United Kingdom, Netherlands, Germany, Austria. The kind of applications that are described are very similar to the topic maps applications from the previous decades.
| Year | Conference Subject |
|---|---|
| 1999 | - intelligent data analysis, XML |
| 2000 | - XML |
| 2001 |
- web information systems
- database and expert systems - groupware - interactive information and processing - meteorology, oceanography and hydrology - astronomical data analysis - computer assister radiology and surgery |
| 2002 |
- extreme markup languages 2002 (XML)
- semantic web - information resources management - engineering education - autotest conference - information visualization - systemics - cybernetics - virtual observatories - natural language processing for biomedical applications |
| 2003 |
- semantic web
- intelligent systems - Emnekart (Norwegian conference on topic maps) - extreme markup languages - web services and e-business - advanced learning technologies - knowledge management - data warehousing and knowledge discovery - electronic publishing - information reuse and integration - information and knowledge engineering - intelligent data acquisition |
| 2004 |
- Emnekart (Norwegian conference on topic maps)
- extreme markup languages 2004 - database and expert systems - web engineering - Asian digital libraries - extending database technology - concept mapping - TeX XML and digital typography - meaningful Internet systems - educational multimedia - hypermedia and telecommunications |
| 2005 |
- First TMRA workshop on topic maps research and applications
- Emnekart (Norwegian conference on topic maps) - Extreme markup languages 2005 - artificial intelligence in education - advanced information networking - medical informatics - reasoning web - geoscience and remote sensing - digital libraries - advanced learning technologies - information reuse and integration - computational intelligence in robotics - intelligent computing - programming languages and compilers - web and mobile information systems - topic maps research. |
| 2006 |
- leveraging the semantics (Topic Maps research and Applications)
- Emnekart 2006, 4th Norwegian Topic Maps conference Asian topic maps summit 2006 - Extreme Markup Languages 2006 - applied intelligent systems - digital information management - e-learning and digital entertainment - topic maps research - ontologies-based databases and informational systems. artificial intelligence applications and innovation - service systems and service management - adaptative hypermedia and adaptative web-based systems - principles and practice of semantic web reasoning - enterprise information system - web intelligence and intelligent agent technology. |
| 2007 |
- scaling topic maps (3rd international conference, Leipzig)
- international topic maps conference, Oslo, Norway - Extreme Markup Languages 2007 - Asian topic maps summit 2007 - knowledge-based intelligent informational and engineering systems - applications of natural language to information systems - artificial intelligence in education - public administration - topic maps research - environmental engineering - intelligent data acquisition and advanced computing systems - intellectual capital knowledge - Atlantic Web Intelligence conference - databases in networked information systems - e-learning - web-based education - advanced learning technologies - computational science - strategic management - knowledge management - intelligent agent technology - complex systems and applications - distributed computing and applications to business - engineering and science - intelligent information technology application. |
| 2008 |
- subject-centric computing (4th conference on topic maps research and applications)
- topic maps 2008 (Oslo) - balisage 2008 - e-commerce technology - enterprise information systems - design - knowledge-based intelligent information and engineering systems - conceptual structures - database and expert systems - model-based software and data integration - visualization, imaging and simulation - advanced information networking - design of communication - Chinese Control conference - blended learning - technology enhanced learning - convergence and hybrid information - intelligent systems - archiving - e-learning - grid and cooperative computing - e-business - intelligent systems design and applications - learning to live in the knowledge society - evolutionary computation - smart manufacturing applications - knowledge acquisition and modeling - algorithms for large-scale information processing in knowledge discovery |
| 2009 |
-linked topic maps (5th TMRA conference)
- topic maps 2009 (Oslo, Norway) - balisage 2009 - intelligent data engineering and automated learning - computer-supported cooperative work in design - application of natural languages to information systems - conceptual structures - methodologies for intelligent systems - database and expert systems - hybrid learning and education - autonomous infrastructure management and security - metadata and semantic research - intelligent and distributed computing - intellectual capital knowledge management and organizational learning - IT in medecine and education - distance learning - simulation and communication - environmental science and information application technology. |
| 2010 |
- topic maps research and applications (6th and last international conference on topic maps)
- topic maps 2010 (Oslo, Norway) - international society for knowledge organization - advances in product development and reliability - education and new learning technologies - information management and evaluation - computing technologies in agriculture - advances in semantic processing - e-learning - knowledge engineering and ontology development. |
| 2011 |
- computer aided systems theory
- business information management - model and data engineering - information modeling and knowledge - frontiers of manufacturing and design - complexity informatics and cybernetics - software services and semantic technologies - technology education and development - ecoinformatics and education - digital information and communication - knowledge engineering and ontology development - knowledge management and information sharing. |
| 2012 |
- information modeling and knowledge bases
- advanced information system engineering - medical informatics in Europe - soft computing applications - knowledge information and creativity support systems - electrical engineering - computing science and automatic control - technology for education - systems man and cybernetics - engineering and computer science. |
| 2013 |
- high performance computing and communications
- healthcare informatics - electronics telecommunications and computers - integrated information - information science and management engineering - military communications and information systems. |
| 2014 |
- semantic web
- advanced learning technologies - research challenges in information science - Pacific visualization - broadband in wireless computing communications and applications - fuzzy systems knowledge discovery and natural computation. |
| 2015 |
- balisage 2015
- intelligent systems and informatics - web engineering - knowledge-based and intelligent information and engineering systems - multimedia interaction design and innovation - computational collective intelligence - distance learning simulation and communication - intelligent transportation big data and smart city - military technologies. |
| 2016 |
- computer science and education
- natural computation fuzzy systems and knowledge discovery |
| 2017 | - computer science and intelligent controls |
| 2018 | - indexing |
| 2019 | - digital experience |
Topic maps have also been the subject of many academic works and publications, a Topic Maps conference was organized in Oslo, Norway, and a Topic Maps was created in the University of Leipzig, Germany. The Extreme Markup Conference, which has become the Balisage conference, which groups advanced uses and breakthroughs in the XML world, has also featured many topic maps-related presentations.
The interest for topic maps has culminated, especially in the Academic World, in 2008. Then, the center of gravity for the activity, which was high in Europe, followed by North America, has shifted to China, where a significant number of application projects have - and are still - designed, many of which have been published in various scientific publications.
In the first period, the focus has been on the interchange of topic maps, and free tools have been designed in order to facilitate the creation and editing of documents compliant with the standard format. The most well-known tool is Omnigator, made by Ontopia, a company based in Norway.
Topic maps are used both in business and government applications and also as a subject for research in the academic world.
Delimiting the extent to which topic maps are used is not as straightforward as it seems.
Several topic maps-based applications are not advertising that they are using the standard, and therefore they will not appear in a survey of topic maps usages. Other topic maps implementations use the concepts in the standard, but not the interchange syntax. For example, there are knowledge networks using the topic maps applications that import and export information in another format, such as RDF. Other implementations use constructs which are similar, albeit not identical, to the ones provided in the standard.
Another factor that makes the situation blurrier is the fact that the offer for pure topic maps software is rather scarse. Sometimes the software consists of a user interface that transforms user input into a topic map in the XTM notation. These implementations have a limited interest in terms of usability, as the demand for interchanging topic maps produced in various environments is not high.
For more than a decade after the topic map standard was published, free software was made available by topic maps companies which help its users get on board with the concepts in the standard. These products play an important rule in creating interest for the standard, but the transition between this learning step and the implementation of working applications turned out to be more difficult for potential users who realized that they had to either build their own tools, or add topic maps-enabling features to other software.
The openness of the topic maps architecture therefore proved to be challenging in terms of adoption.
It is possible that the full potential of topic maps will take longer to achieve that it was initially thought. Many implementations of taxonomies for example are quasi-topic map applications, and they work well as long as they remain self-contained. When the need arises to connect these knowledge bases with others that are maintained by different techniques, topic maps may become more relevant than ever to provide the missing linking fabric between them.
The emergence of graph databases and their increasing share of the market also signals a possible new avenue for a new generation of topic maps-based applications. The future will depend on how much of the conceptual model contained in the Topic Maps architecture will be considered a useful addition to the design of these new approaches.
This section summarizes work on topic maps in various application domains.
Research institutions working on agriculture in China have created multiple modeling approaches on a variety of platforms (Jiang, Haiyan and Fu, Bing and Zhang, Mei and Zhu, Yan and Cao, Weixin. 2011). However, many complex aspects of the submodels show some similarities, including their interdependencies. For example, the field scale crop growth models are influenced by the weather, soil, varieties and cultivation conditions. The Agricultural Model Component Library is composed of a set of faceted terms, and resource files. This faceted model lacks associations between its objects. A descriptive model describing association between those resources was created. A topic map model was designed to integrate these approaches, and was divided into three steps: the metadata relationships, the relationships between abstract and concrete properties, and relationships that use components from the models. The research team conducting this project has focused on the wheat growing model components in the agricultural model component database.
Relying on merely transposing existing structured information, which predominates in the domain of aircraft maintenance is not sufficient, especially when the use of multiple devices. Using topic maps facilitates a semantic redesign. This process is achieved mainly through associations between topics. For example the distinctions between "note", "warning", or "caution" can be further amplified by adding a scope of "important" and "very important". The ability to refine the level of semantic abstraction opens the possibility to refine interfaces on various devices, present or future (Kadner, Kay and Roussel, David. 2007).
The goal of a topic-maps based approach is to provide more flexibility and a better user interface to engineers who can populate information in the repository system without having to dig into all details of the multi-standards compliance required in the documentation. A network of associations is used to navigate who has been involved in a given activity, and how components, documents and states are related together. Another network is being maintained between documents and the various activities they describe (Brown, D and Leal, D and McMahon, C and Crossland, R and Devlukia, J. 2004).
Disclosures which are essential to decision making appear in multiple places.
Topic maps are being considered in the area of cultural heritage for the "Archivio di Stato di Pavia". Harmonization between different heritage areas comprises three levels: the entity level, that can be expressed as an authority file mapping between multiple ontologies, the structure level to describe relations between topics, and the semantic level that can be used to create and maintain a semantic network.
Topic maps were used to set up a repository of visual resources and to foster new paths in image cataloguing and retrieval. An existing thesaurus was imported as a topic map (Leuenberger, M and Grossmann, S and Stettler, N and Herget, J. 2006). Topics maps are described as an appropriate solution for digital archives on the Web.
Military Universities Cooperation
The goal of the CEFME portal is to provide access to information about the structure and activities of military universities in ten Central European countries, from Austria to the Baltic republics, including their departments, research and conference activities. The portal is derived from MilUNI, and adds new classes to the Knowledge Management system, such as meeting, action, workgroup and function. The portal content is maintained by national administrator teams, using the ATOM2 data editor.
The knowledge management system built for military universities is based on Topic Maps. The ontology consists of classes with characteristics, including university, organization, person, conference, collection, article, study programme, project, product, venue, web article, occurrences of these classes (for example "University of Defence, Czech Republic", and associations between occurrences of classes, for example a hierarchical relationship connecting a university with its faculties. Topic maps were chosen because the model is intuitive, and conforms better to human thinking than any other representation. A topic map software (AToM) has been implemented, extending the Topic Maps model to account for specific requirements including processing, namely changes in the processing of occurrences of classes, work with associations. Several data types have been introduced, including code and ident, for identification, group tree for a simple built-in taxonomy, selection for a one-level code list, text, picture, and file. Topic associations have been complemented with ordering, a parameter indicating the power of the relationship, and hierarchy, a special type of relationship expressing the Parent-Child association. Three built-in taxonomies have been incorporated: activity profile, domain tree, geographical tree. User interfaces have been implemented to use the taxonomy terms as query filters.
The next step of this project is to get users to use it and obtain their feedback.
Folklore resources, made of handcraft objects, clothing, written texts, music records, photographs, video recordings, etc., require a metadata model that combines elements from various metadata standards. Furthermore, an access policy is required to protect the authenticity of information without preventing retrieval of resources. A metadata schema was created including three kinds of data: descriptive, structural, and administrative. The Topic Maps tool was used to generate the mapping between the various protocols and standard representations used. Every metadata element is considered a topic, and three topic types have been created to assign each topic to the kind of data it represents. Associations between topics are qualifies as "equivalence", "refinement", "hierarchical".(Lourdi, Irene and Papatheodorou, Christos and Nikolaidou, Mara. 2007)
Business games are used for training, and discuss tools for the development or modifications of workflows, simulate management decisions and project management. Their success comes from a holistic approach, which has progressively been dissolved due to the number and diversity of products available. Most business games are created in isolation from each other. Various integration methods have been used in an effort to integrate the different approaches: classical approaches such as Object-Role Modeling, System Analysis, Decision Tables, and UML have been augmented by using Topic Maps, Mind Mapping, Influence Graphs. Topic Maps have been specifically used to create a web portal for the whole simulation model, extracted from the Mind Maps, and can also be embedded in e-learning systems supporting the training of simulation (Reusch, Peter J. A. and Bozguney, Emine and Reusch, Pascal. 2007).
Topic maps are used to unify classification systems for products and services in order to limit redundancies, reduce stocks and costs, to support international trade and tariff systems, e-commerce. This approach is particularly relevant in Europe, using the German system "Eclass", but also in America, which uses primarily the United Nations Standard Products and Services Code (UNSPSC). In Europe, if a buyer using the German system wants to send an order to a Spanish company, they can't switch the product descriptions without any mapping. The classification principles of the products are quite different depending on the classification system, and designing a topic map to perform the mapping has proven a fruitful approach to this problem. The topic map approach can be applied to any classification system used, and is appropriate for describing complex topics and associations with many roles and types (Reusch, PJA and Reusch, P. 2003).
Topic maps are used to unify classification systems for products and services in order to limit redundancies, reduce stocks and costs, to support international trade and tariff systems, e-commerce. This approach is particularly relevant in Europe, using the German system "Eclass", but also in America, which uses primarily the United Nations Standard Products and Services Code (UNSPSC). In Europe, if a buyer using the German system wants to send an order to a Spanish company, they can't switch the product descriptions without any mapping. The classification principles of the products are quite different depending on the classification system, and designing a topic map to perform the mapping has proven a fruitful approach to this problem. The topic map approach can be applied to any classification system used, and is appropriate for describing complex topics and associations with many roles and types (Reusch, PJA and Reusch, P. 2003).
By using Topic Maps, the city of Bergen was able to switch from a portal organized according to their administrative structure to a new architecture which integrated these data and was organized by subjects, more in tune with what its citizens were looking for. This transformation took place in 2007 (Garshol, Lars Marius. 2008).
The topic maps navigation system includes four layers: information resources, knowledge, information navigation, and application. The requirements for a city information portal include a unified information channel, a powerful content management capability, an individual application service, and integration with the existing information system. The "topic map ontology" is a knowledge layer built above information resources, which includes diverse information sets organized as databases: statistical data, education, traffic, tour, environment. This project was focused on integrating the pieces together, and the following step is about automating the construction of topic maps from evolving information resources.
Topic maps have been presented as a viable solution for designing a urban traffic information portal in China, in 2008 (Jun, Zhai and Wang Qinglian and Miao, Lv. 2008).
Topic maps are used in a training tool for the network as a whole. There are several topic maps modules, on scientific writing and open access. The technology was selected for its modular structure and adaptability to different local training requirements. When tests were conducted about the relevance of the training materials, the responses were overwhelmingly positive. 94% of the participants or more express satisfaction on the usefulness of the course and efficiency to learn new concepts. 88% agreed with the new methodology.
Medical data relevant to a patient are spread across multiple web locations. The need for integrating heterogeneous data sources integration has led to select topic maps in combination with description logics. The system proposed is a navigation map resulting from merging all medical data about a patient enabling a doctor to rebuild a comprehensive medical record (Ouziri, Mourad and Verdier, Christine. 2008). Topic maps representing data source are enriched by inserting common knowledge in DAML+OIL. Topic map index resources and captures semantics at the same time. Multiple topic maps are merged into a single global topic map. One of the limitation of topic maps, according to the authors (Ouziri, Mourad and Verdier, Christine. 2008), is the fact that the constraint language was not mature. They are therefore proposing their own representation for constraints based on Description Logics. This project was based on a small sample of data, and therefore their prototype could not be validated over real data.
A major challenge in the diagnosis systems used for melanoma detection is the ability to assure interoperability and reusability of the multiple images created by dermatoscopy analysis and related diagnostic data. The Topic maps approach offers the flexibility necessary to manage an ontology in complex datasets, where new properties and relationships can emerge which were not part of a pre-defined schema. AJAX technology is used for increased efficiency between the servers and the clients and a new system was created to connect to the various sources available on multiple servers. The creation of a topic maps-based ontology has been done through several steps, starting with the identification of the basic components of the knowledge infrastructure of digital images and clinical data. The separation between the knowledge layer and the physical information resources is considered a major asset in this context. Once data is acquired, an application was designed for automatic detection of the ABCD parameters. The ABCD rule is used in dermatoscopy analysis to classify images and establish whether they lead to a benign, suspicious or melanoma diagnosis by analyzing the following properties: A symmetry, B order, C olor, and D iameter or Differential structures (Papastergiou, A. and Hatzigaidas, A. and Tryfon, G. and Ioannidis, D. and Grammatikopoulos, G.. 2007).
A topic maps-based platform has been developed to assist in the recruitment of clinical trial participants(Damen, David and Luyckx, Kim and Hellebaut, Geert and Van den Bulcke, Tim. 2013). Eligibility screening is a labour-intensive and long process that requires access to lab results, clinical notes, medical codes, imaging reports, stored in multiple places and sometimes proprietary databases. The semantic representation was build using Topic maps, chosen over RDF because of its flexibility and the fact that associations are always reified, i.e. they can be treated as topics when needed. The design includes the following topic types: Concept, Cell, Group, Clinical Trial and Institution. The concept is the most granular building block, the cell is an arbitrary complex aggregation of the values retrieved from a concept to a ternary logic value: true, false or unknown. A group combines cells and other groups into a logical structure that can be used for inclusion/exclusion criteria.
Topic maps have been used in a historical project of Spanish civil engineering. Personal fonds of the most important Spanish civil engineers were made accessible online and connected to additional information resources containing metadata from thesauruses. These metadata records are collected as RDF fragments containing descriptive metadata the document, such as creators, dates, abstracts and keywords and applied to the data present in the documents, such as subjects, persons, corporations, geographic locations, etc., and are aggregated as a topic map, which is considered the actual metadata registry (Eito-Brun, Ricardo. 2014).
A set of predetermined concepts for concepts (e.g., question, problem, definition, example) and relationships (e.g. references, supports, contradicts) has been developed during the initial phase to address the complexity of conceptual design during the requirement analysis phase. Concept graphs were used, because of their flexibility and the ability to create annotations. Then, during the design phase, a software architecture was developed using UML, addressing all the concepts and relationships created during the initial phase. But, due to their weak expressiveness, they had to be supplemented by additional sets of templates and constraints. Topic maps were used to establish the interrelations between concepts and the evolutionary changes. The relationships are reified. Both roles and process steps are expandable because of the flexibility of Topic Maps. The "scope" operation and the reification of relationships are among the most essential features that were used.(Ueberall, Markus and Drobnik, Oswald. 2006)
Digital assistants such as Siri for IOS or Cortana for Windows assume patterns as input commands. But the rigid language structure which it presupposes doesn't always correspond to non-English languages. Sentences can be expressed in more flexible ways, for example in Japanese. A method for selecting answers to users' queries was developed using dependency analysis to create triples. Each noun is represented by a topic. Each sentence representing an answer is also a topic. The fact that an answer corresponds to a noun is an association between topics. The relationships between the words and appropriate answers uses a topic map.(Kimura, Masaomi. 2015) Although the evaluation of that method on the sample studies proved incomplete, the method for preventing irrelevant answers can be refined, in order to increase the accuracy.
Topic maps based on the strict data model only support topic navigation and do not reflect the relevance of the various elements contributing to the overall knowledge (Lu, Huimin and Feng, Boqin and Zhao, Yingliang and Zheng, Qinghua and Liu, Jun. 2008).
The principles of faceted thesauruses provide the construction rules for topic maps used in classificatory structures. Hierarchical relationships and controlled vocabularies, associative relationships can be defined as strict rules to apply by subdomain (Buchel, O and Coleman, A. 2003).
Topic Maps are used to address the heterogeneous nature of metadata. However, it does not consider constraints. Description Logics is used to perform reasoning on complex hierarchical data structures. The combination of both could indicate a fruitful approach towards interoperability of distributed learner profile (Ouziri, Mourad. 2008).
With the popularization of network technology and higher level educational information, education in the field of knowledge management has become a research hotspot. The knowledge point is an essential cell which transferred information in teaching. Automatically extracted from the network a lot of teaching resources in knowledge and information, and then translates the information into the corresponding elements of topic maps. A model of educational resources knowledge management based on topic maps is put forward. The method of semantic relatedness of topic maps evaluation among topics and between topic and resources to be studied, which can be applied to knowledge services. For example, knowledge navigation system, constitute the new navigation system, the precise concept of association, narrow the search space, returns the sorted result set, enhance knowledge retrieval, efficiency and accuracy of navigation.
Topic maps is considered important for managing educational resources at a national level in China. The goal of this initiative is to improve the efficiency of educational resources, for teacher preparation, search , and personalized services for learning services. In 2003, China started becoming involved with topic maps, mostly on a theoretical level. On a practical level, domains that were explored including knowledge organization, management and navigation, applied to digital libraries, government, business, education. Topic maps enable not only to create a clear hierarchy in a teaching resources architecture, but also enable constant renewal of resources, and providing personalized content (Wang Xiaobo and Liu Jinglong and Li Xiaop. 2016).
Topic maps are used to describe the role that is played by people carrying up specific tasks in a work environment (Seo, Wonchul and Choi, Sungchul and Kim, Kwangsoo and Lee, Jae Yeol. 2011). For knowledge workers, skills are related to the roles. A role is related to a goal and a set of responsibilities. A human is related to a set of capabilities through a relationship "is capable of" and to a set of roles by "is experienced by". Responsibilities are related to capabilities by a relationship of the type "supported by". The topic maps approach has been chosen because the description of goals, capabilities and responsibilities can be fully captured using associations. Also, the comparison with the Web Ontology Language (OWL) is in favor of Topic Maps because OWL uses attributes for relationships that can be changed, and eventually may lead to a less stable model. The data model brought by Topic maps is therefore considered more stable.
The Knowledge Management System of the Army of the Czech Republic uses Topic Maps to select the concepts and relationships between this concepts. The research team discovered a number of duplicates between documents and their parts. The initial category of "Document" was further divided into groups according to their focus. A similar process occurred with the "Process" class that was divided into groups according to the type, purpose and nature of the operation. The knowledge layer instances were connected to articles. Navigation through the various of knowledge was facilitated by the ability to obtain information in many different ways (Burita, Ladislav and Lunacek, O. and Maly, V. and Ondryhal, V. and Hruza, Petr. 2013).
A software tool was built to visualize graphically a topic map produced from the National Library of Medicine's MeSH thesaurus, more specifically from the disease and drug categories. This tool includes the ability to query semantically the medical database from the topic map, and from the content. The topic map serves to help users navigating by a subject of interest, and overcomes the obstacle of requiring familiarity with the structure of the database (Stanescu, Liana and Burdescu, Dumitru. 2009).
The visible advantage of using topic maps is to reduce the learning curve to access the medical database.
The accumulation on online learning materials gives rise to shareable content objects. A topic map-based ontological model has been developed to connect the materials through the knowledge domains across sets of objects (Adorni, Giovanni and Coccoli, Mauro and Vercelli, Gianni and Vivanet, Giuliano. 2008).
General article. No specific insight.
Topic maps is used to transfer topics and associations from a humanoid robot's speech. This robot is built on a technology that contains multi-lingual speech recognition. There are different types of topics, that emanate from learning resources, web design, discernment —which provides a structure for an avatar and conversation system—and subjects of knowledge, organized in 13 domains. Essential expressions distinguish between the introduction of a concept, its deepening, and its completion. That can be applied to a movie plot, with a beginning, a development, and an ending. Using this structure, it becomes possible to connect the robot to the topic map and to interact with it through a dialogs. When the robot recognized the name of a topic type, it can retrieve the dialog occurrence to narrate (Matsuura, Shu and Naito, Motomu. 2016).
Topic maps have been chosen to build an ontology-based document driven memory, because it removes the distinction between types and instances, by the unconstrained semantic in the relations between topics, the ability to attach resources to concepts through occurrences, and to express annotations using relations and concept labels inside scopes (Abel, MH and Lenne, D and Moulin, C and Benayache, A. 2004).
Topic Maps for e-Learning (TM4L) contains an editor and a viewer. The editor handles the part-whole relationship, and builds a topic partonomy instead of a taxonomy, due to the important role played by the part-whole relationship type in e-learning. The viewer offers three views: a graph view, a text view, and a tree view, to navigate the topic map. Only topics that are most immediately related are shown (Dicheva, D and Dichev, C and Wang, DD. 2005).
Topic Maps provide a ways to assist and enhance specific types of thinking and learning. Students can write down what they have learned as a set of topics. These topics can be connected with an existing map of existing concepts that constitute the core of knowledge and each student can realize the gap between whey they learn and the core of knowledge in which the new concept resides. By connecting with new concepts, students have a way to master comprehensive concepts. These benefits are compared with those provided by other technology paradigms, such as those used by data mining, case-based reasoning, information retrieval, weblogs and portfolios (Wu Kebao and Dai Junxun. 2008).
Assistors at the call centers at the Internal Revenue Service in the US are facing multiple questions from taxpayers, and need to give them precise answers on the spot, sometimes about complex questions, using the very last information available, which is constantly being updated. The need to improve the quality of their answers has been the incentive for the IRS to develop a topic-map based system, called Tax Map, where all information available on a given subject at a given topic is represented by a topic, which is associated to a number of closely related topics as well as references to a number of documents, forms, and publications, which are relevant to that subject. After several years where Tax Map was reserved for internal use, the IRS has released a web version available on the Internet (https://taxmap.irs.gov), which is updated on a weekly or bi-weekly basis, depending on the availability of new documents. This topic map is used by accountants and tax professionals.
Tax Map is a topic map project that was used at the Internal Revenue Service (IRS), the tax administration agency in the US, from 2001 to 2019. Before the project started, the IRS call centers whose mission is to answer taxpayers' questions over the phone underwent an investigation by the US Congress after taxpayers complained about the inadequacy of the answers they received. The investigation showed that, although 70% of the answers were correct, the rest was incomplete at best, and sometime misleading. The call center assistors were overwhelmed with a constant flux of updates, delivered to them in printed form, and couldn't follow up with the new materials. In order to address this issue, the IRS decided to change the way information was delivered, and to regroup, in electronic form, all information relevant to a given topic. However, IRS decided that the production of Tax Map should not interfere with the existing workflow, and that it should be created as an external layer, pointing to existing information, rather than starting anew with a different way to author the source documents.
The Tax Map project consists in ingesting a number of document sources and publish a web site where links are made from topics created dynamically to the occurrences of the documents where relevant information appears. For a few years, the use of this web site was restricted to IRS employees. Then IRS published a Cd-Rom containing Taxmap pages. Later the Taxmap web site was released to the public. A more comprehensive version was produced at the same time for internal use only.
As the document sources were in SGML then in XML, extracting topics from section headers was straightforward. The document sources were the IRS publications, forms and instructions, and the Frequently Asked Qestions and Teletax topics. As the documents were properly structured, this process was relatively straightforward. But the result of the integration of extracted topics was not entirely satisfying, as, despite the efforts made to author the information in a consistent way, variants of the same topics appear to slightly different names in different contexts. Sometimes a name was used to designate what could be interpreted as two different topics.
A persistent knowledge base was created to intercept these issues on the fly, while running the scripts aimed at producing the Tax Map web pages. This knowledge base was maintained by IRS as a configuration file, and contained information enabling the creation of relations between topics, declaring which topics should be considered synonyms, or renamed. Some headers were deleted when they lost meaning while being extracted (for example, a header named "Table 1" was excluded from the topic map). Automatic rules were designed to relate automatically topics when their names contained three words or more in common with other names.
The batch process that resulted in the creation of Tax Map was fully automated. When IRS issued new versions of the documents, or was retiring some documents, the whole process was restarted from scratch and a brand new topic map was produced. This process was repeated every week, or every other week depending on the pace of publication. In exceptional circumstances (for example a tax reform), the production of new versions of Tax Map was even more frequent. The knowledge base was persistent and was applied to every iteration of the production process. The terms that were not hits as extracted headers were ignored, but they were kept in case they would reappear later.
The maintenance of the knowledge base was done internally by IRS. For several years, workshops were organized to address a specific domain, such as retirement, or the Affordable Care Act, or foreign-related tax issues. The knowledge base was maintained by the Tax Map Editor at IRS.
Using Tax Map was straightforward and could be used by IRS call centers assistors as well as tax professionals without any preliminary training and learning curve. ***
Topic maps are used by defining mappings between existing resources, designing cross-lingual resources and guaranteeing reliable distributed knowledge exchange. Topic maps can help navigate between multiple existing faceted classifications. It facilitates the integration of different ontologies, classifications, thesauri (Hopmans, G and Kruijsen, PP and Oud, L and Verhoeff, J and Kuster, MW and Clews, J. 2006).
Key Messages Implications for Practice
• NEtwork of COllaboration Between Europe and Latin American Caribbean countries (NECOBELAC) supports cooperation between European and Latin American countries to make it easier to spread valuable heath information online.
• Modular training material on scientific writing and open access (OA) publishing is available on the Project website to be used to ‘train the trainers’ and then for local training activities.
• Course replication at local level contributes to create major awareness on health information dissemination embedding best practices at workplace.
Implications for Policy
NEtwork of COllaboration Between Europe and Latin American Caribbean countries networking and cooperation among European and Latin American public health institutions favours the development of shared advocacy initiatives.
The Enhanced Networked Monographs project is a project developed with the New York University Library and three academic publishers, which consists in integrating back-of-the-book indexes for a variety of books on humanities: history, religion, gender studies, sociology, etc. The resulting index contains the exact entries as found in each book. It is contained in a topic map that provides access to every page of every book present in the indexes. Since every index was conceived in isolation, there are a number of variations in the way the terms appear, and a tool has been developed to curate the results and superimpose a more accurate integrated vision of the topics. Automatic integration rules combined with manual curation has been used. The project is currently under completion.
A book recommendation system based on topic maps is not restricted by the syntactic constraints imposed on existing recommenders. SOLE-R, a "Semantic, Ontological and Linguistic Enhanced Recommender", is a project developed specifically for book recommendation using semantic and linguistic features using topic maps to describe books as well as users(Garrido, Angel L. and Pera, Maria Soledad and Ilarri, Sergio. 2014). The resulting, unique, topic map emanates from topic maps creating by aggregating book descriptions scraped from various web resources, and using a semantic/linguistic engine to extract the significant topics, and user profiles containing their interests and preferences.
Beomap is a topic map used to explore Twitter data. Users can use an arbitrary query to define an ad-hoc topic map, opening the way to explore the underlying social media topics space. Beomap also helps discovering how a topic relevant to the main query may also match additional aspects that pertain to another query. This approach leads to serendipitous discoveries. (Leginus, Martin and Zhai, ChengXiang and Dolog, Peter. 2015). Furthermore, Beomap enables a user to explore and navigate the space through user-chosen visualization metrics.
Online reputation has become increasingly important, and it becomes important to check how it comes into existence. Web data is aggregated as fuzzy sets, which serve to populate a topic map. Logic based ontologies are preferred to topic maps for general axiomatic specifications, and topic maps are used here as a visualization technique, used to help locate information by subject, and therefore filter according to specific conversations. The grassroots ontology built from a folksonomy is handled as a topic map, which enables characterization of information and navigation between items. Navigating the topic map helps find similar terms clustered around a topic.
The knowledge for organizational memory is represented as a topic map. Organizational memory is based on the same processes as knowledge management in general, with an emphasis on organizational settings and business goals. Topic maps are used to operationalize the trio of logic, ontology and computation. A research framework has been proposed that comprises the organization structure and culture, the work practices, the technology and the individuals. Documents have been annotated withe meta-information supporting query and retrieval. The knowledge representation constructs have been developed as three ontologies: an information ontology, an enterprise ontology, and a domain ontology. Organizational memory is implemented by cross-referencing these three ontologies.
The XTM format has been chosen as the interchange syntax for inference rules. The methodology was supported by the China Credit Information Service, Ltd. (CCIS), which maintains knowledge since 45 years about businesses in Taiwan and Mainland China and has been used by its Asset Management Division on computer-aided valuation of real estate. The experiment was successful but needs further results to be further expanded. (Ju, Teresa L.. 2006)
The K-Discovery project proposes using Topic Maps to address the challenges of finding relevant knowledge across continuously growing and distributed organizational memories. The case study was conducted with data from the Groupware Competence Center of the University of Paderborn. Their organizational memories contain more than 30 databases and around 60,000 documents. Relying on a combination of database structures and full-text search has proven insufficient for searching information across sources. Topic Maps enable creating link networks above heterogeneous resources. The basic design elements of groupware-based applications, especially forms, is used to identify the topics and how they are associated. The fields in forms are being analyzed as a basis for topic typing (person, city, etc.) and for populating the relationships between those topics (a person "x" lives in the city "y").
The K-Discovery project was focused on working on a visualization technique, known as "hyperbolic tree" to represent the topics and their neighbors with information on the nature of their relationships. (Smolnik, S and Erdmann, I. 2002)
Topic maps are used to create a content-based recommendation system that combines users' interests with news which are relevant to their current location. In (Luis Garrido, Angel and Buey, Maria G. and Ilarri, Sergio and Furstner, Igor and Szedmina, Livia. 2015), the authors propose to add geolocation capabilities to content management systems, and to identify places among the words used in textual content. An elaborated geolocation process is combined with natural language process techniques to create knowledge that is usable in this context. This knowledge is represented as a topic map. The topic map is built using a general categorization module and a geographical categorization module.
The philosophy ontology described contains concepts originating from classical philosophy and the semantic relations that connect them. An ontology management system was developed based on Topic maps containing a semi-automated concept extractor and tools to store and retrieve ontologies based on topic maps. Building the ontology involved planning, conceptualization and implementation.
Topic maps have been created to capture data from a study on the evolution of the research literature on Gender Studies in Informatics. Topics are acquired by extracting keywords through text mining, and creation of semantic relations between them. Text mining involves segmentation, phrase detection, collocations, and integration of external knowledge. The semantic relations are derived from collocations and integration of external knowledge. The topic map creation involves the formation of a co-occurrence matrix, cluster formation, and creation of maps. Methods for the creation of topic maps include "Leximap", based on co-word pairs in a co-word matrix, the "Inclusion Index" to determine the spatial occurrences of the concepts in the topic maps, and the "Joint Conditional Probability Index (JCP)", which also determines the strength of their relations. The result of the study helps trace out the interactions among the subdomains in Gender Studies Informatics (Suriya, M and Nagarajan, R and Babu, RS and Kumaresan, V. 2004).
While many topic maps applications have been developed in many different environments, we have only witnessed few application cases involving the merging of existing topic maps, which has been the main motivation for the standardization of the Topic Maps model. There are many "non-standard" topic maps, which rely on connecting subjects together, and with the resources in which they occur, which implement the same principles.
The Topic Maps interchange format (mainly XTM) is used, but the offer for topic maps-compliant software has become rather sparse, at least as standalone products. Many applications have been built implementing topic maps but they are not necessarily exportable to usage environments other than the ones they were designed for. Furthermore, the use of XML as a data interchange format seems to be declining, as technology moves towards the cloud, and the interchange of information is turning toward JSON which makes data automatically browser-enabled.
However, many new web applications based on concepts similar to topic maps are being built, using graph database technologies as their backend. In this new landscape, it is possible that the concepts such as the interchangeability of interconnected topics is going to be perceived as an increasingly useful "new" requirement.
The notation for topic maps, first based on SGML, was soon converted into XML, as a way to simplify the work of the implementers. RDF also was using XML as its primary syntax in its first iterations. At that time (early 2000s) XML became the lingua franca of the Web, and served as the privileged way to exchange not only structured documents, but also data emanating from various software tools. However, XML was not as easy as it was initially planned. Numerous difficulties came from using namespaces, and also from the necessity of parsing the documents against a schema. Any document that was not 100% compliant was rejected by the browsers and other widely used technologies. So progressively XML became supplanted by a format that the web browsers could natively understand, the Javascript Object Notation (JSON), which requires a much lighter background to be processed and exchange. Most straightforward data structures were therefore more easily exchanged using this latter format. XML remains useful for encoding documents, and in a way returned where it belonged. As Topic maps were using an XML-based interchange format, they are perceived as belonging exclusively to that universe. And XML became the lingua franca for data exchange, getting more traction than just in the publishing world.
In the last decade, things started to change. The availability of new technologies, based on web platforms, has favored the use of Javascript for many applications, and has resulted in shifting the preference for data exchange to a format directly processable by browsers, the JavaScript Object Notation (JSON), and became more liked than XML, which added extra complexities due to the necessity of parsing. Furthermore, the ability to add external entities to an XML application has created a vulnerability. It turned out that XML was more and more perceived as adding unnecessary complexities to data exchange, and that JSON became more popular.
RDF can be - and is - used to encode these knowledge graph, but RDF suffers two limitations that prevented it to become the most used language to exchange knowledge graphs. RDF is constrained to declare a URI for each resource (aka topic), which is appropriate on the Web, but limits its use for applications which are not web-centric. Its semantic for triples (subject-object-predicate) can be used for almost anything, but it had the effect to limit its usage to a certain kind of relationship. In addition, the ecosystem around RDF, including the Web Ontology Language (OWL), the query language (SPARQL) creates expectations that users need to learn these technologies in order to benefit from all the usages that are possible. This explains why RDF has mostly been developed in research and academic applications, and is considered as a deterrent by industry users who are looking for a quicker time to deploy applications.
The "property graph" exchange format is a simple format that enables to exchange graph information based on the properties of their nodes and relations between nodes (edges). In a way it is a subset of the topic maps format, with nodes representing topics and edges representing relations. The internal properties of the topics (multiple names, scopes) can be added as properties to the nodes and interchanged as property graphs. Property graphs enable a similar reinterpretation to occur with RDF-based resources. Therefore it can be considered as the least common denominator that enables the integration between knowledge bases using either one of these formats. Furthermore, other formats can be converted into Property graphs.
Therefore it is possible to continue using topic maps even when using an alternative format for interchanging data. The main concept of topic maps, which is to isolate topics by pointing at information sources from outside, and organize a graph of topics and user-defined semantics for their relationships, is still achievable. New technologies are emerging to make this possible, including native graph databases, processing tools to validate and query the graph, curation tools to interact with the graph.
Using topic maps in real-world applications also have demonstrated some of the deficiencies of the original topic map model. The distinction between an association (a link between two topics) and an occurrence (a link between a topic and an external resource) is somewhat artificial as a resource can itself be considered a topic. In sophisticated topic maps applications, there can even be a distinction between two topics for the resource: one for the resource itself (for example, a book) and one for the subject of the resource (the subject about which this book is). Another example of limitation is the notion of topic type, which is just a shortcut for a relationship between two topics, one representing the item, the other representing the category. There are applications in which there is a need to distinguish between different semantics for types, for example, if it would make sense to distinguish type, class, category, broader term, etc. Of course, it is possible to distinguish all these nuances using regular associations for topic types, making the notion of "topic type" irrelevant. Another weakness of the model comes from the notion of scope. A scope is used to disambiguate several topics using the same name by declaring its domain of validity. The Topic Maps standard recommends using scope to qualify the language for names. But it may be easier to create a new property for language isolated from scope, in a context where multilingual disambiguation is required. The result of these limitations is that the standard interchange syntax recommended to use topic maps may flatten distinctions that could have been made by the various component topic maps involved. A freely defined set of properties to attach on nodes, such as what is enabled by the property graph notation, yields more accurate results.
Therefore, the topic maps model can be reinterpreted not as a constraint, but as a guiding template to facilitate the creation of knowledge graphs, because it provides well-defined structural components that help to get on board, knowing that it is up to the users to add to that environment specific features that are not included, and of course discard any feature that is not adequate in their context of use.
Abel, MH and Benayache, A and Lenne, D and Moulin, C and Barry, C and Chaput, B. 2004. "Ontology-based organizational memory for e-learning", EDUCATIONAL TECHNOLOGY & SOCIETY 7: 98-111.
Abel, MH and Lenne, D and Moulin, C and Benayache, A. 2004. "Using topic maps in an e-learning context", WEB ENGINEERING, PROCEEDINGS , LECTURE NOTES IN COMPUTER SCIENCE , 4th International Conference on Web Engineering (ICWE 2004), JUL 26-30, 2004, Munich, GERMANY 3140: 581-582.
Adorni, Giovanni and Coccoli, Mauro and Vercelli, Gianni and Vivanet, Giuliano. 2008. "An Ontological Model for Learning Content Design", 7TH EUROPEAN CONFERENCE ON E-LEARNING, VOL 1 , 7th European Conference on e-Learning, NOV 06-07, 2008, Ayia Napa, CYPRUS 1-7.
Adorni, Giovanni and Coccoli, Mauro and Vercelli, Gianni and Vivanet, Giuliano. 2008. "An Ontological Model for Learning Content Design", 7TH EUROPEAN CONFERENCE ON E-LEARNING, VOL 1 , 7th European Conference on e-Learning, NOV 06-07, 2008, Ayia Napa, CYPRUS 1-7.
Amrouche, Hakim and Abel, Marie-Helene and Balla, Amar and Moulin, Claude. 2007. "Interrogation of ontologies formalized in topic maps with TMQL in E-learning context", ADVANCES IN INTELLIGENT WEB MASTERING , ADVANCES IN SOFT COMPUTING , 5th International Atlantic Web Intelligence Conference (AWIC 2007), JUN 25-27, 2007, Fontainebleau, FRANCE 43: 29-+.
Arndt, Hans-Knud and Graubitz, Henner and Klesinski, Rene. 2007. "Using topic maps for sustainability reporting", INFORMATION TECHNOLOGIES IN ENVIRONMENTAL ENGINEERING , Environmental Science and Engineering , 3rd International ICSC Symposium on Information Technologies in, MAR 29-30, 2007, Carl VonOssietzky Univ Oldenburg, Oldenburg, GERMANY 47-+.
Barta, R and Zangerl, A. 2003. "Slide shows meet topic maps", IKE'03: PROCEEDINGS OF THE INTERNATIONAL CONFERENCE ON INFORMATION AND KNOWLEDGE ENGINEERING, VOLS 1 AND 2 , International Conference on Information and Knowledge Engineering (IKE, JUN 23-26, 2003, LAS VEGAS, NV 379-383.
Benayache, A and Abel, MH. 2005. "Using knowledge management method for e-learning", Methods and Technologies for Learning , 1st International on Methods and Technologies for Learning, MAR 09-11, 2005, Palermo, ITALY 447-453.
Biemann, C and Quasthoff, U and Bohm, K and Wolff, C. 2003. "Automatic discovery and aggregation of compound names for the use in knowledge representations", JOURNAL OF UNIVERSAL COMPUTER SCIENCE , 3rd International Conference on Knowledge Management (I-Know 03), JUL 02-04, 2003, Graz, AUSTRIA 9: 530-541.
Brown, D and Leal, D and McMahon, C and Crossland, R and Devlukia, J. 2004. "A Web-enabled virtual repository for supporting distributed automotive component development", ADVANCED ENGINEERING INFORMATICS 18: 173-190.
Buchel, O and Coleman, A. 2003. "How can classificatory structures be used to improve science education", LIBRARY RESOURCES & TECHNICAL SERVICES 47: 4-15.
Burgess, Mark. 2009. "Knowledge Management and Promises", SCALABILITY OF NETWORKS AND SERVICES, PROCEEDINGS , Lecture Notes in Computer Science , 3rd International Conference on Autonomous Infrastructure, Management, JUN 30-JUL 02, 2009, Enschede, NETHERLANDS 5637: 95-107.
Burita, Ladislav and Lunacek, O. and Maly, V. and Ondryhal, V. and Hruza, Petr. 2013. "Knowledge Management System Based on Lessons Learned Documents", 2013 MILITARY COMMUNICATIONS AND INFORMATION SYSTEMS CONFERENCE (MCC) , Military Communications and Information Systems Conference (MCC), OCT 07-09, 2013, Saint Malo, FRANCE -.
Burita, Ladislav and Lunacek, O. and Maly, V. and Ondryhal, V. and Hruza, Petr. 2013. "Knowledge Management System Based on Lessons Learned Documents", 2013 MILITARY COMMUNICATIONS AND INFORMATION SYSTEMS CONFERENCE (MCC) , Military Communications and Information Systems Conference (MCC), OCT 07-09, 2013, Saint Malo, FRANCE -.
Burita, Ladislav. 2011. "Ontology Development as a Software Engineering Procedure", DIGITAL INFORMATION AND COMMUNICATION TECHNOLOGY AND ITS APPLICATIONS, PT II , Communications in Computer and Information Science , International Conference on Digital Information and Communication, JUN 21-23, 2011, Univ Bourgogne, Dijon, FRANCE 167: 1-8.
Burita, Ladislav. 2014. "Knowledge Management System for Military Universities Cooperation", 2014 10TH INTERNATIONAL CONFERENCE ON COMMUNICATIONS (COMM) , 10th International Conference on Communications (COMM), MAY 29-31, 2014, Bucharest, ROMANIA -.
Campillo Torres, Irima and Raquel Hernandez, Thais and Cabrera Morales, Idalmis and Matos Matos, Mercedes and Palomino Palomino, Magaly. 2016. "To a better representation of the information and the knowledge at libraries", REVISTA PUBLICANDO 3: 3-13.
Carloni, Olivier and Leclere, Michel and Mugnier, Marie-Laure. 2006. "Introducing graph-based reasoning into a knowledge management tool: An industrial case study", ADVANCES IN APPLIED ARTICIAL INTELLIGENCE, PROCEEDINGS , LECTURE NOTES IN COMPUTER SCIENCE , 19th International Conference on Industrial, Engineering and Other, JUN 27-30, 2006, Annecy, FRANCE 4031: 590-599.
Cheever, Erik and Bielenberg, Ames. 2011. "TOPIC MAPS USED TO PRESENT INTERRELATIONSHIPS IN DYNAMIC LINEAR SYSTEMS", 2011 ASEE ANNUAL CONFERENCE & EXPOSITION , ASEE Annual Conference & Exposition , ASEE Annual Conference and Exposition, JUN 26-29, 2011, Vancouver, CANADA -.
Chu, Ding and Sheets, David A. and Zhao, Ye and Wu, Yingyu and Yang, Jing and Zheng, Maogong and Chen, George. 2014. "Visualizing Hidden Themes of Taxi Movement with Semantic Transformation", 2014 IEEE Pacific Visualization Symposium (PacificVis) , IEEE Pacific Visualization Symposium , IEEE Pacific Visualization Symposium (PacificVis), MAR 04-07, 2014, Keio Univ, Yokohama, JAPAN 137-144.
Chuprina, Svetlana and Statsenko, Natalya. 2010. "Using Ontology and Metadata to Integrate eLearning Resources and Administrative Information System of University", PROCEEDINGS OF THE 9TH EUROPEAN CONFERENCE ON E-LEARNING, VOL 1 , 9th European Conference on e-Learning, NOV 04-05, 2010, Instituto Superior Engenharia Porto, PORTUGAL 171-178.
Damen, David and Luyckx, Kim and Hellebaut, Geert and Van den Bulcke, Tim. 2013. "PASTEL: A Semantic Platform for Assisted Clinical Trial Patient Recruitment", 2013 IEEE INTERNATIONAL CONFERENCE ON HEALTHCARE INFORMATICS (ICHI 2013) , 1st IEEE International Conference on Healthcare Informatics (ICHI), SEP 09-11, 2013, Philadelphia, PA 269-276.
De Castro, Paola and Marsili, Daniela and Poltronieri, Elisabetta and Agudelo Calderon, Carlos. 2012. "Dissemination of public health information: key tools utilised by the NECOBELAC network in Europe and Latin America", HEALTH INFORMATION AND LIBRARIES JOURNAL 29: 119-130.
Dicheva, D and Dichev, C and Wang, DD. 2005. "Visualizing topic maps for e-learning", 5th IEEE International Conference on Advanced Learning Technologies, Proceedings , 5th IEEE International Conference on Advanced Learning Technologies, JUL 05-08, 2005, Kaohsiung, TAIWAN 950-951.
Dong, Y and Li, MS. 2004. "HyO-XTM: a set of hyper-graph operations on XML Topic Map toward knowledge management", FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE 20: 81-100.
Dudycz, Helena. 2012. "An Attempt of the Heuristic Evaluation of Visualization in Searching Economic Information in Topic Maps", ADVANCED INFORMATION SYSTEMS ENGINEERING WORKSHOPS, CAISE 2012 , Lecture Notes in Business Information Processing , 24th International Conference on Advanced Information System Engineering, JUN 25-26, 2012, Gdansk, POLAND 112: 130-142.
Eito-Brun, Ricardo. 2014. "XTM-based Metadata Exploration and Retrieval: The History of Spanish Civil Engineering Project", 3RD INTERNATIONAL CONFERENCE ON INTEGRATED INFORMATION (IC-ININFO) , Procedia Social and Behavioral Sciences , 3rd International Conference on Integrated Information (IC-ININFO), SEP 05-09, 2013, Prague, CZECH REPUBLIC 147: 345-348.
Garrido, Angel L. and Pera, Maria Soledad and Ilarri, Sergio. 2014. "SOLE-R, a Semantic and Linguistic Approach for Book Recommendations", 2014 14TH IEEE INTERNATIONAL CONFERENCE ON ADVANCED LEARNING TECHNOLOGIES (ICALT) , IEEE International Conference on Advanced Learning Technologies , 14th IEEE International Conference on Advanced Learning Technologies, JUL 07-10, 2014, Athens, GREECE 524-+.
Garrido, Piedad and Naranjo, Fernando and Tramullas, Jesus and Esteban, Miguel A. and Lopez, Ana and Azuara, Guillermo and Salinas, Ana and Ramos, Pedro and Hervas, Eva and Pascual, Eduardo. 2010. "Free Traceability Management using RFID and Topic Maps", PROCEEDINGS OF THE 4TH EUROPEAN CONFERENCE ON INFORMATION MANAGEMENT AND EVALUATION , 4th European Conference on Information Management and Evaluation, SEP 09-10, 2010, Univ Nova Lisboa, Lisbon, PORTUGAL 93-102.
Garrido, Piedad and Naranjo, Fernando and Tramullas, Jesus and Esteban, Miguel A. and Lopez, Ana and Azuara, Guillermo and Salinas, Ana and Ramos, Pedro and Hervas, Eva and Pascual, Eduardo. 2010. "Free Traceability Management using RFID and Topic Maps", PROCEEDINGS OF THE 4TH EUROPEAN CONFERENCE ON INFORMATION MANAGEMENT AND EVALUATION , 4th European Conference on Information Management and Evaluation, SEP 09-10, 2010, Univ Nova Lisboa, Lisbon, PORTUGAL 93-102.
Garshol, Lars Marius. 2008. "A citizen's portal for the city of Bergen", SCALING TOPIC MAPS , LECTURE NOTES IN ARTIFICIAL INTELLIGENCE , 3rd International Conference on Topic Maps Research and Applications, OCT 11-12, 2007, Leipzig, GERMANY 4999: 25-35.
Gui, Weixia and Li, Xucheng and Zhou, Songru. 2014. "A knowledge management model of virtual enterprise based on topic maps", INFORMATION SCIENCE AND MANAGEMENT ENGINEERING, VOLS 1-3 , WIT Transactions on Information and Communication Technologies , International Conference of Information Science and Management, 2013, Wuhan, PEOPLES R CHINA 46: 1569-1574.
Hierl, Sonja and Bauer, Lydia and Boeller, Nadja and Herget, Josef. 2007. "DIAMOND: Didactical Approach for Media Competence Development in a blended learning environment", 6TH EUROPEAN CONFERENCE ON E-LEARNING , 6th European Conference on e-Learning (ECEL 2007), OCT 04-05, 2007, Copenhagen Business Sch, Copenhagen, DENMARK 327-339.
Hofmann, T. 1999. "Probabilistic topic maps: Navigating through large text collections", ADVANCES IN INTELLIGENT DATA ANALYSIS, PROCEEDINGS , LECTURE NOTES IN COMPUTER SCIENCE , 3rd International Symposium on Advances in Intelligent Data Analysis, AUG 09-11, 1999, AMSTERDAM, NETHERLANDS 1642: 161-172.
Hopmans, G and Kruijsen, PP and Oud, L and Verhoeff, J and Kuster, MW and Clews, J. 2006. "Topic maps for European Administrative Nomenclature", CHARTING THE TOPIC MAPS RESEARCH AND APPLICATIONS LANDSCAPE , LECTURE NOTES IN ARTIFICIAL INTELLIGENCE , 1st International Workshop on Topic Maps Research and Application, OCT 06-07, 2005, Leipzig, GERMANY 3873: 177-182.
Hopmans, G and Kruijsen, PP and Oud, L and Verhoeff, J and Kuster, MW and Clews, J. 2006. "Topic maps for European Administrative Nomenclature", CHARTING THE TOPIC MAPS RESEARCH AND APPLICATIONS LANDSCAPE , LECTURE NOTES IN ARTIFICIAL INTELLIGENCE , 1st International Workshop on Topic Maps Research and Application, OCT 06-07, 2005, Leipzig, GERMANY 3873: 177-182.
Huth, C and Smolnik, S and Nastansky, L. 2001. "Applying topic maps to ad hoc workflows for semantic associative navigation in process networks", CRIWG 2001: SEVENTH INTERNATIONAL WORKSHOP ON GROUPWARE, PROCEEDINGS , 7th International Workshop on Groupware (CRIWG 2001), SEP 06-08, 2001, DARMSTADT, GERMANY 44-49.
Jiang, Haiyan and Fu, Bing and Zhang, Mei and Zhu, Yan and Cao, Weixin. 2011. "The Study on the Organization Approach of Agricultural Model Components Library Based on Topic Map", COMPUTER AND COMPUTING TECHNOLOGIES IN AGRICULTURE IV, PT 3 , IFIP Advances in Information and Communication Technology , 4th IFIP TC 12 Conference on Computer and Computing Technologies in, OCT 22-25, 2010, China Agr Univ, EU China Ctr Informat & Commun Technol, Nanchang, 346: 186-+.
Ju, Teresa L.. 2006. "An open architecture for codifying organizational memory", EISTA '06: 4TH INT CONF ON EDUCATION AND INFORMATION SYSTEMS: TECHNOLOGIES AND APPLICAT/SOIC'06: 2ND INT CONF ON SOCIAL AND ORGANIZATIONAL INFORMATICS AND CYBERNETICS, VOL I , 4th Int Conf on Education and Information Syst/2nd Int Conf on Social, JUL 20-23, 2006, Orlando, FL 256-261.
Ju, Teresa L.. 2006. "Representing organizational memory for computer-aided utilization", JOURNAL OF INFORMATION SCIENCE 32: 420-433.
Jun, Zhai and Wang Qinglian and Miao, Lv. 2008. "Application of XML Topic Maps to Knowledge Navigation and Information Retrieval for Urban Traffic Information Portal", Proceedings of the 27th Chinese Control Conference, Vol 6 , 27th Chinese Control Conference, JUL 16-18, 2008, Kunming, PEOPLES R CHINA 458-462.
Kadner, Kay and Roussel, David. 2007. "Documentation for aircraft maintenance based on topic maps", LEVERAGING THE SEMANTICS OF TOPICS MAPS , Lecture Notes in Artificial Intelligence , 2nd International Conference on Topic Maps Research and Applications, OCT 11-12, 2006, Leipzig, GERMANY 4438: 56-+.
Kim, Jung-Min and Choi, Byoung-Il and Shin, Hyo-Phil and Kim, Hyoung-Joo. 2007. "A methodology for constructing of philosophy ontology based on philosophical texts", COMPUTER STANDARDS & INTERFACES 29: 302-315.
Kimura, Masaomi. 2015. "A proposal of topic map based Chatterbot for non-English natural language input", KNOWLEDGE-BASED AND INTELLIGENT INFORMATION & ENGINEERING SYSTEMS 19TH ANNUAL CONFERENCE, KES-2015 , Procedia Computer Science , 19th International Conference on Knowledge Based and Intelligent, SEP 07-09, 2015, Singapore, SINGAPORE 60: 841-849.
Leginus, Martin and Zhai, ChengXiang and Dolog, Peter. 2015. "Beomap: Ad Hoc Topic Maps for Enhanced Exploration of Social Media Data", ENGINEERING THE WEB IN THE BIG DATA ERA , Lecture Notes in Computer Science , 15th International Conference on Web Engineering (ICWE), JUN 23-25, 2015, Rotterdam, NETHERLANDS 9114: 200-218.
Leuenberger, M and Grossmann, S and Stettler, N and Herget, J. 2006. "Topic maps for image collections", CHARTING THE TOPIC MAPS RESEARCH AND APPLICATIONS LANDSCAPE , LECTURE NOTES IN ARTIFICIAL INTELLIGENCE , 1st International Workshop on Topic Maps Research and Application, OCT 06-07, 2005, Leipzig, GERMANY 3873: 107-111.
Li, Guangzheng and Ren, Wei. 2011. "E-Learning Resources Organization Based on Topic Maps", COMPUTING AND INTELLIGENT SYSTEMS, PT III , Communications in Computer and Information Science , International Conference on Computing, Information and Control (ICCIC, SEP 17-18, 2011, Wuhan, PEOPLES R CHINA 233: 81-88.
Li, Jie and Hale, Andrew. 2016. "Output distributions and topic maps of safety related journals", SAFETY SCIENCE 82: 236-244.
Long Mengmeng and Cheng Jinhua. 2009. "A Study on Knowledge Navigation System-based XTM for City Information Resource", 2009 INTERNATIONAL FORUM ON INFORMATION TECHNOLOGY AND APPLICATIONS, VOL 2, PROCEEDINGS , International Forum on Information Technology and Applications (IFITA, MAY 15-17, 2009, Chengdu, PEOPLES R CHINA 63-67.
Lourdi, Irene and Papatheodorou, Christos and Nikolaidou, Mara. 2007. "A multi-layer metadata schema for digital folklore collections", JOURNAL OF INFORMATION SCIENCE 33: 197-213.
Lu, Huimin and Feng, Boqin and Zhao, Yingliang and Zheng, Qinghua and Liu, Jun. 2008. "A New Model for Distributed Knowledge Organization Management", GCC 2008: Seventh International Conference on Grid and Cooperative Computing, Proceedings , 7th International Conference on Grid and Cooperative Computing, OCT 24-26, 2008, Shenzhen, PEOPLES R CHINA 261-265.
Luis Garrido, Angel and Buey, Maria G. and Ilarri, Sergio and Furstner, Igor and Szedmina, Livia. 2015. "KGNR: A Knowledge-based Geographical News Recommender", IEEE 13TH INTERNATIONAL SYMPOSIUM ON INTELLIGENT SYSTEMS AND INFORMATICS (SISY) , International Symposium on Intelligent Systems and Informatics , 13th IEEE International Symposium on Intelligent Systems and Informatics, SEP 17-19, 2015, Subotica, SERBIA 195-198.
Matsuura, Shu and Naito, Motomu. 2016. "Shaping Dialogues with a Humanoid Robot Based on an E-Learning System", 2016 11TH INTERNATIONAL CONFERENCE ON COMPUTER SCIENCE & EDUCATION (ICCSE) , International Conference on Computer Science & Education , 11th International Conference on Computer Science & Education (ICCSE), AUG 23-25, 2016, Nagoya Univ, Nagoya, JAPAN 7-12.
Mueller, Jenny and Bleier, Arnim and Bock, Benjamin and Kirchgeorg, Manfred and Maicher, Lutz. 2009. "Enhancing Customer Relationship Management using Enterprise Information Integration with Topic Maps", AIC '09: PROCEEDINGS OF THE 9TH WSEAS INTERNATIONAL CONFERENCE ON APPLIED INFORMATICS AND COMMUNICATIONS: RECENT ADVANCES IN APPLIED INFORMAT AND COMMUNICATIONS , Recent Advances in Computer Engineering, AUG 20-22, 2009, Moscow, RUSSIA 151-+.
Ouziri, Mourad and Verdier, Christine. 2008. "Domed: Semantic data integration and navigation in Web-based medical records", INFORMATICS FOR HEALTH & SOCIAL CARE 33: 191-210.
Ouziri, Mourad. 2007. "A Semantic-Based Navigation Approach for Information Retrieval in the Semantic Web", EMERGING ARTIFICIAL INTELLIGENCE APPLICATIONS IN COMPUTER ENGINEERING: REAL WORD AI SYSTEMS WITH APPLICATIONS IN EHEALTH, HCI, INFORMATION RETRIEVAL AND PERVASIVE TECHNOLOGIES , Frontiers in Artificial Intelligence and Applications 160: 90-105.
Ouziri, Mourad. 2008. "Accessing the distributed learner profile in the semantic web", KNOWLEDGE - BASED INTELLIGENT INFORMATION AND ENGINEERING SYSTEMS, PT 1, PROCEEDINGS , LECTURE NOTES IN ARTIFICIAL INTELLIGENCE , 12th International Conference on Knowledge-Based Intelligent Information, SEP 03-05, 2008, Zagreb, CROATIA 5177: 464-472.
Papastergiou, A. and Hatzigaidas, A. and Tryfon, G. and Ioannidis, D. and Grammatikopoulos, G.. 2007. "Employing melanoma diagnosis support using Ajax and topic maps", PROCEEDING OF THE 11TH WSEAS INTERNATIONAL CONFERENCE ON COMPUTERS: COMPUTER SCIENCE AND TECHNOLOGY, VOL 4 , Electrical and Computer Engineering Series , 11th WSEAS International Conference on Computers, JUL 26-28, 2007, Crete, GREECE 317-+.
Popa, Emil M. and Hunyadi, Daniel and Pah, Iullan. 2008. "Ontology-based document-driven memory for e-Learning", PROCEEDINGS OF THE 8TH WSEAS INTERNATIONAL CONFERENCE ON APPLIED INFORMATICS AND COMMUNICATIONS, PTS I AND II: NEW ASPECTS OF APPLIED INFORMATICS AND COMMUNICATIONS , Recent Advances in Computer Engineering , 8th WSEAS International Conference on Applied Informatics and, AUG 20-22, 2008, Rhodes, GREECE 326-+.
Portmann Edy, Meier Andreas, Cudre-Mauroux Philippe and Pedrycz Witold. 2015. "FORA - A fuzzy set based framework for online reputation management", FUZZY SETS AND SYSTEMS 269: 90-114.
Portmann, Edy and Tam Nguyen and Sepulveda, Jose and Cheok, Adrian David and BA Meier, A and Donze, L andBF Meier, A and Donze, L. 2012. "Fuzzy Online Reputation Analysis Framework", FUZZY METHODS FOR CUSTOMER RELATIONSHIP MANAGEMENT AND MARKETING: APPLICATIONS AND CLASSIFICATIONS 139-167.
Presutti V., Garshol L.M., Vitali F., Pepper S., and Gessa, N. 2005. "Towards the definition of guidelines for RDF and Topic Maps interoperability", http://isotopicmaps.org/tmcl/tmcl.html#normative-references, In Proceedings of the 5th International Workshop on Knowledge Markup and Semantic annotation (SemAnnot 2005) located at the 4rd International Semantic Web Conference (ISWC 2005) .
Qi, Huang and Sun, Shouqian. 2006. "Knowledge management system based on semantic Web in e-learning community", TECHNOLOGIES FOR E-LEARNING AND DIGITAL ENTERTAINMENT, PROCEEDINGS , LECTURE NOTES IN COMPUTER SCIENCE , 1st International Conference on Technologies for E-Learning and Digital, APR 16-19, 2006, Hangzhou, PEOPLES R CHINA 3942: 188-192.
Quan, Lu and Jing, Chen. 2009. "A Grid Agent Model for Information Retrieval based on Topic Map and Knowledge Elements Mining", 2009 INTERNATIONAL CONFERENCE ON ENVIRONMENTAL SCIENCE AND INFORMATION APPLICATION TECHNOLOGY, VOL III, PROCEEDINGS, , International Conference on Environmental Science and Information, JUL 04-05, 2009, Wuhan, PEOPLES R CHINA 347-+.
Rauschmayer, A and Renner, P. 2005. "TUBE - Structure-orientation in a prototype-based programming environment", PLC '05: Proceedings of the 2005 International Conference on Programming Languages and Compilers , International Conference on Programming Languages and Compilers, JUN 27-30, 2005, Las Vegas, NV 194-200.
Reusch, Peter J. A. and Bozguney, Emine and Reusch, Pascal. 2007. "Integrated tool sets for business games and simulation", IDAACS 2007: PROCEEDINGS OF THE 4TH IEEE WORKSHOP ON INTELLIGENT DATA ACQUISITION AND ADVANCED COMPUTING SYSTEMS: TECHNOLOGY AND APPLICATIONS , IEEE International Workshop on Intelligent Data Acquisition and Advanced , 4th IEEE International Workshop on Intelligent Data Acquisition and, SEP 06-08, 2007, Dortmund, GERMANY 485-+.
Reusch, PJA and Reusch, P. 2003. "Classification of products and services to support business process engineering and e-commerce", IDAACS'2003: PROCEEDINGS OF THE SECOND IEEE INTERNATIONAL WORKSHOP ON INTELLIGENT DATA ACQUISITION AND ADVANCED COMPUTING SYSTEMS: TECHNOLOGY AND APPLICATIONS , IEEE International Workshop on Intelligent Data Acquisition and Advanced , 2nd IEEE International Workshop on Intelligent Data Acquisition and, SEP 08-10, 2003, LVIV, UKRAINE 452-457.
Reusch, PJA and Reusch, P. 2003. "Classification of products and services to support business process engineering and e-commerce", IDAACS'2003: PROCEEDINGS OF THE SECOND IEEE INTERNATIONAL WORKSHOP ON INTELLIGENT DATA ACQUISITION AND ADVANCED COMPUTING SYSTEMS: TECHNOLOGY AND APPLICATIONS , IEEE International Workshop on Intelligent Data Acquisition and Advanced , 2nd IEEE International Workshop on Intelligent Data Acquisition and, SEP 08-10, 2003, LVIV, UKRAINE 452-457.
Rittershofer, A. 2005. "Supporting self-regulated e-learning with visual topic-map-navigation", KNOWLEDGE AND INFROMATION VISUALIZATION: SEARCHING FOR SYNERGIES , LECTURE NOTES IN COMPUTER SCIENCE , International Workshop on Visual Artefacts for the Organization of, MAY, 2004, Knowledge Media Res Ctr, Tuebingen, GERMANY 3426: 355-363.
Rittershofer, A. 2005. "Supporting self-regulated e-learning with visual topic-map-navigation", KNOWLEDGE AND INFROMATION VISUALIZATION: SEARCHING FOR SYNERGIES , LECTURE NOTES IN COMPUTER SCIENCE , International Workshop on Visual Artefacts for the Organization of, MAY, 2004, Knowledge Media Res Ctr, Tuebingen, GERMANY 3426: 355-363.
Rittershofer, A. 2005. "Supporting self-regulated e-learning with visual topic-map-navigation", KNOWLEDGE AND INFROMATION VISUALIZATION: SEARCHING FOR SYNERGIES , LECTURE NOTES IN COMPUTER SCIENCE , International Workshop on Visual Artefacts for the Organization of, MAY, 2004, Knowledge Media Res Ctr, Tuebingen, GERMANY 3426: 355-363.
Saint-Marc, Cecile and Davoine, Paule-Annick and Villanova-Oliver, Marlene. 2014. "Methods for mapping volcanic events overlaid across time", JOURNAL OF MAPS 10: 249-256.
Schweiger, R and Dudeck, J. 2006. "Improving information retrieval using XML and topic maps", CHARTING THE TOPIC MAPS RESEARCH AND APPLICATIONS LANDSCAPE , LECTURE NOTES IN ARTIFICIAL INTELLIGENCE , 1st International Workshop on Topic Maps Research and Application, OCT 06-07, 2005, Leipzig, GERMANY 3873: 253-262.
Schweiger, Ralf and Hoelzer, Simon and Dudeck, Joachim. 2005. "Advanced Information Retrieval Using XML Standards", CONNECTING MEDICAL INFORMATICS AND BIO-INFORMATICS , Studies in Health Technology and Informatics , 19th International Congress of the, AUG 28-SEP 01, 2005, Geneva, SWITZERLAND 116: 677-682.
Seo, Wonchul and Choi, Sungchul and Kim, Kwangsoo and Lee, Jae Yeol. 2011. "An ontology-based approach to role-human assignment in human processes for knowledge intensive services", EXPERT SYSTEMS WITH APPLICATIONS 38: 8352-8369.
Smolnik, S and Erdmann, I. 2002. "Visual navigation of distributed knowledge structures in groupware-based organizational memories", SIXTH INTERNATIONAL CONFERENCE ON INFORMATION VISUALISATION, PROCEEDINGS , 6th International Conference on Information Visualisation, JUL 10-12, 2002, LONDON, ENGLAND 353-360.
Stanescu, Liana and Burdescu, Dumitru. 2009. "Information Structuring and Retrieval with Topic Maps for Medical E-Learning", ADVANCES IN ELECTRICAL AND COMPUTER ENGINEERING 9: 27-33.
Suriya, M and Nagarajan, R and Babu, RS and Kumaresan, V. 2004. "Constrains in building domain-specific topic maps for the discipline "Gender studies in Informatics (GSI)"", DIGITAL LIBRARIES: INTERNATIONAL COLLABORATION AND CROSS-FERTILIZATION, PROCEEDINGS , LECTURE NOTES IN COMPUTER SCIENCE , 7th International Conference on Asian Digital Libraries (ICADL 2004), DEC 13-17, 2004, Shanghai, PEOPLES R CHINA 3334: 290-299.
Ueberall, Markus and Drobnik, Oswald. 2006. "Collaborative software development and topic maps", CHARTING THE TOPIC MAPS RESEARCH AND APPLICATIONS LANDSCAPE , Lecture Notes in Artificial Intelligence , 1st International Workshop on Topic Maps Research and Application, OCT 06-07, 2005, Leipzig, GERMANY 3873: 169-176.
Van Hemel, S and Paepen, B and Engelen, J. 2003. "Smart search in newspaper archives using topic maps", FROM INFORMATION TO KNOWLEDGE , 7th International Conference on Electronic Publishing, JUN 25-28, 2003, UNIV MINHO, GUIMARAES, PORTUGAL 251-259.
Vassallo, S. 2006. "Navigating through archives, libraries and museums: Topic maps as a harmonizing instrument", CHARTING THE TOPIC MAPS RESEARCH AND APPLICATIONS LANDSCAPE , LECTURE NOTES IN ARTIFICIAL INTELLIGENCE , 1st International Workshop on Topic Maps Research and Application, OCT 06-07, 2005, Leipzig, GERMANY 3873: 231-240.
Wang Xiaobo and Liu Jinglong and Li Xiaop. 2016. "Education Resource Knowledge Management Model Base on Topic Maps and Knowledge Relevance Algorithm", 2015 INTERNATIONAL CONFERENCE ON INTELLIGENT TRANSPORTATION, BIG DATA AND SMART CITY (ICITBS) , International Conference on Intelligent Transportation, Big Data and Smart City (ICITBS), DEC 19-20, 2015, Vietnam, AUSTRIA 351-357.
Weichhart, Georg and Neubauer, Matthias and Stary, Christian. 2010. "Learning Interoperability in Emerging Supply Networks", ENTERPRISE ARCHITECTURE, INTEGRATION AND INTEROPERABILITY , IFIP Advances in Information and Communication Technology , 21st IFIP World Computer Congress (WCC), SEP 20-23, 2010, Australian Comp Soc (ACS), Brisbane, AUSTRALIA 326: 90-100.
Weippl, Edgar R. and Klemen, Markus and Fenz, Stefan and Ekelhart, Andreas and Tjoa, A. Min. 2007. "The semantic desktop: A semantic personal information management system based on RDF and topic maps", ONTOLOGIES-BASED DATABASES AND INFORMATION SYSTEMS , Lecture Notes in Computer Science , 2nd VLDB Workshop on Ontologies-Based Databases and Informational, SEP 11, 2006, Seoul, SOUTH KOREA 4623: 135-+.
Wu Kebao and Dai Junxun. 2008. "Knowledge Management Technologies in Education", KAM: 2008 INTERNATIONAL SYMPOSIUM ON KNOWLEDGE ACQUISITION AND MODELING, PROCEEDINGS , International Symposium on Knowledge Acquisition and Modeling, DEC 21-22, 2008, Wuhan, PEOPLES R CHINA 93-97.
Wu, Linjing and Liu, Qingtang and Liu, Man andBA Anonymous and BF Anonymous. 2014. "Topic Map Construction of Text Resources", INTERNATIONAL SYMPOSIUM ON FUZZY SYSTEMS, KNOWLEDGE DISCOVERY AND NATURAL COMPUTATION (FSKDNC 2014) , International Symposium on Fuzzy Systems, Knowledge Discovery and, AUG 19-21, 2014, Xiamen, PEOPLES R CHINA 120-126.
Yu, Jeong-Youn and Hwang, Yun-Young and Lee, Kyu-Chul. 2006. "Representing process-based ontology using topic maps for Web Services", CIC 2006: 15TH INTERNATIONAL CONFERENCE ON COMPUTING, PROCEEDINGS , 15th International Conference on Computing, NOV 21-24, 2006, Mexico City, MEXICO 373-+.
Yu, Jeong-Youn and Hwang, Yun-Young and Lee, Kyu-Chul. 2006. "Representing process-based ontology using topic maps for Web Services", CIC 2006: 15TH INTERNATIONAL CONFERENCE ON COMPUTING, PROCEEDINGS , 15th International Conference on Computing, NOV 21-24, 2006, Mexico City, MEXICO 373-+.
Zaher, L'Hedi and Cahier, Jean-Pierre and Zacklad, Manuel. 2006. "Towards open information retrieval", 2006 INTERNATIONAL CONFERENCE ON SERVICE SYSTEMS AND SERVICE MANAGEMENT, VOLS 1 AND 2, PROCEEDINGS , 3rd International Conference on Service Systems and Service Management, OCT 25-27, 2006, Troyes, FRANCE 41-+.
Zhai Jun and Bao Hong-Yang and Liu Wei-Bin. 2007. "Using XML topic maps to achieve knowledge navigation for city information portal", PROCEEDINGS OF 2007 INTERNATIONAL CONFERENCE ON PUBLIC ADMINISTRATION (3RD), VOL I , 3rd International Conference on Public Administration, OCT 21-22, 2007, Univ Elect Sci & Technol, Sch Polit Sci & Public Adm, Chengdu, PEOPLES R 389-395.
Zhai, Jun and Liu, Weibein and Wang, Qinglian and Lv, Miao. 2008. "Using XML Topic Maps to Achieve Knowledge Navigation for E-Government Information Portal", SEVENTH WUHAN INTERNATIONAL CONFERENCE ON E-BUSINESS, VOLS I-III , 7th Wuhan International Conference on E-Business, MAY 31-JUN 01, 2008, Wuhan, PEOPLES R CHINA 908-913.
Zhai, Jun and Shi, Zhiman and Zhou, Zhou and Chen, Yan. 2007. "Knowledge navigation for digital city based on topic maps", DCABES 2007 Proceedings, Vols I and II , International Symposium on Distributed Computing and Applications to, AUG 14-17, 2007, Yichang, PEOPLES R CHINA 130-133.