1. Domain specification
5.2 Additional remarks on Topic Maps in LIS
5.2.3 Semantic Web
The literature on Topic Maps and LIS seems to agree on the fact that Topic Maps is one of the enabler technologies to achieve the idea of the “Semantic Web”. For instance Adams (2002), Bokman (2006), and many others, give Topic Maps such a role. Fith (2002) for example states that “One of the core ideas behind the Semantic Web is the creation of machine-
processable relationships between resource identifiers (URI’s). Two often discussed ways of representing those relationships are RDF and Topic Maps.”
However, this identification is problematic. On the one side, the Semantic Web –SW– with capital letters is one of the activities of the W3C, an international consortium that develops standards and guidelines for the Web. On the other hand, Topic Maps is an ISO standard whose purpose is to represent information about the structure of information resources (ISO13250). Both ideas began in different communities, at different times and for different purposes.
However, both RDF (the Semantic Web syntax) and XTM (the Topic Maps syntax) became official as respectively a W3C recommendation and an ISO standard in the same year, and their similarities didn’t pass unnoticed by the two communities. Pepper (2008c) tells the story of this link and how at some point choosing one over the other was a desired purpose for the future of the Web.
However, Topic Maps and the Semantic Web (this relation is referred often as ‘Topic Maps and RDF’) have a different scope and are made for different purposes, among one of them, to serve for organizing information in the Web. RDF as well can be used for non Web related purposes, and that’s the reason why both Topic Maps and RDF are usually encompassed with the term “Semantic web technologies”.
What is behind this encompassing term is an idea about the need to apply to the Web some Information Organization principles that would allow more structured searches and results, as opposed to the searches done through word-based and ranking algorithms. Tim Berners-Lee, the creator of the W3C and the person considered to be the inventor of the Web, published in 2001 an article in the journal Scientific American where he said that “The Semantic Web will bring structure to the meaningful content of Web pages, creating an environment where software agents roaming from page to page can readily carry out sophisticated tasks for users.”
Here is where Topic Maps fit in: the idea of the Web that can use ontology-based systems to give structure to the Web. Since structuring information is one of the main purposes of Information and Knowledge Organization, the LIS community has started to adopt those ‘semantic Web technologies’ for these purposes, and also the RDF and Topic Maps community have looked at its historical principles for applying them in their visions.
Both the approaches to solving this vision and the vision of a semantic Web itself are
problematic and have been criticized: Shirky, 2003 and Veltman, 2004, for instance. Research on the semantic Web abounds in the different communities as well as within the LIS
community, for example, on its role and the implication of these technologies for its theories and practices.
Even though the differences between RDF and Topic Maps were not within the scope of this thesis, it is important to remark that they are highly interoperable due to task forces that have worked for this purpose.
What the literature on Topic Maps in LIS seems to agree upon (and this is explained by the fact that the literature comes either from the Topic Maps community or by people in the LIS community who have adopted Topic Maps) is that Topic Maps has advantages over RDF (Garshol (2002; Yi, 2008; Tramullas & Garrido, 2006; Oh, 2009 and Tuhoi, 2005). The main reason for this consideration is basically the approach to subjects instead of documents, and the inherently richer semantics of Topic Maps as a model over RDF due to its simplification of relations and identity mechanisms that are considered problematic in representing
knowledge. Direct vs. indirect addressing seems, though, to be the main differential characteristic.
Tramullas & Garrido (2006) for instance, decided to adopt Topic Maps because of its “structure and syntax [are] more modern” and because it is a more “flexible and abstract paradigm”(p.2). These authors found in the development of their application (Potnia) that even though Topic Maps and RDF are interoperable in the sense that it is possible to represent RDF structures through Topic Maps, the other way (representing Topic Maps into RDF) represents a loss of the semantics. The authors don’t specify the details of this conversion, but the literature on the interoperability of Topic Maps and RDF gives account of the reasons. Tuhoi (2005) considered also that “Topic Maps are “higher-level” than RDF, including a few extra features”. Oh (2009) concluded in his prototype development and study that “TM implementation is relatively easy compared with RDF/OWL so one can expect a better return on investment.”
To conclude, one remarkable conclusion that has implications for Information and Knowledge Organization was made by Yi (2008) while comparing Topic Maps and RDF/OWL in relation
“While RDF/OWL is optimal for making inferences about information, Topic Maps is better for finding information. RDF/OWL is suitable for the physical sciences or biomedical domains, where terms are less ambiguous; however, as terms from the humanities for social sciences have multiple meanings, making inferences using
RDF/OWL is not an easy task. Topic Maps is therefore more appropriate for nonscientific fields than is RDF/OWL because Topic Maps can represent multiple meanings for each term and can build complex relationships among terms. (Yi, 2008)
However, one of the limitations of the application of Topic Maps in MLA, instead of RDF, is that precisely because the focus of Topic Maps is not specifically the Web, the solutions and implementations are often reduced to specific domains or not widely distributed and
disseminated through the network. That explains the reasons why initiatives such as Linked Data appeal more to the LIS community. However, the Topic Maps community is starting to develop similar strategies to make its solutions and perspectives widely accessible. For instance, at the time of the writing of this thesis, The Ontopia Knowledge Suite (OKS) for creating topic maps was under discussion to become open source.
The Simple Knowledge Organization System (SKOS) is an application of the Semantic Web technology RDF to knowledge organization systems (KOS). However, in the literature on Topic Maps in LIS there no existing research was found on the comparison of Topic Maps and SKOS from an Information and Knowledge Organization perspective.
Some authors refer to the inherent advantages of SKOS. For instance, Sigel (2006) says that “SKOS provides a model for expressing the basic structure and content of concept schemes such as thesauri, classification schemes, subject heading lists, taxonomies, ‘folksonomies’, other types of controlled vocabulary, and also concept schemes embedded in glossaries and terminologies” (Sigel, 2006). Sigel defines all these “concept schemes” as semiformal ontologies that can be represented through the use of SKOS, in which purpose would be “to bring the worlds of library classification and Web technology together” (p.?). However, these statements are also valid if applied to Topic Maps and more research on their similarities and differences is required.
SKOS is being used already in libraries, for instance at the Deutsche Nationalbibliothek, where it was used to represent the Dewey Classification System.