Concepts and semantic relations in information science

Abstract

Concept. A concept is determined via its intension and extension as well as by definition. We will meet the problem of vagueness by introducing prototypes. Some important definitions are concept explanations (after Aristotle) and the definition of family resemblances (in the sense of Wittgenstein). We will model concepts as frames (according to Barsalou). The most important paradigmatic relation in KOS is hierarchy, which must be arranged into different classes: Hyponymy consists of taxonomy and simple hyponymy, meronymy consists of many different part-whole-relations. For practical application purposes, the transitivity of the given relation is very important. Unspecific associative relations are of little help to our focused applications and should be replaced by generalizable and domain-specific relations. We will discuss the reflexivity, symmetry, and transitivity of paradigmatic relations as well as the appearance of specific semantic relations in the different kinds of KOS (folksonomies, nomenclatures, classification systems, thesauri, and ontologies). Finally, we will pick out KOS as a central theme of the Semantic Web.

Concept-based document classification using Wikipedia and value function

“…However, this is not the only approach for building semantic-rich applications. Instead, increasingly many researchers have argued in favor of simplified concept systems (Stock, 2010;Weller, 2010). In particular, when considering the vision of building the Semantic Web (Berners-Lee, Hendler, & Lassila, 2001), ontologies appear to be far too complicated in terms of structure and use (Weller, 2010).…”

“…Just considering Wikipedia's articles as concepts, and by regarding its overall link structure-including redirects, hyperlinks, and category links-as relations, it is warranted to argue that Wikipedia is the largest semantic network available today. Furthermore, as discussed by Weller (2010) and Stock (2010), a semantic-rich knowledge system can be well realized without implementing all the aspects of fully fledged ontologies; for example, by using a less expressive model such as a thesaurus. Now, considering the success of Wikipedia, there are many reasons to believe the argument.…”

“…The recent article by Stock (2010) gives a nice coverage of this discussion. It is broadly accepted that if a system is expected to understand semantics in text, closer to the way human brains do, then it needs an access to extensive background knowledge describing the domain concepts and their dependencies (Hendler & Feigenbaum, 2001).…”

“…Nowadays, the necessity of this kind of concept mapping is recognized as a precondition for enabling high-level meaning-based search. As Feldman (2000) and Stock (2010) put it, the guideline for building modern retrieval systems is to make them find what the users mean and not what they say. This is what separates concept-based information retrieval (CBIR) from the classical approaches relying on bag of words and literal term matching.…”

A Pluralistic Approach to the Philosophy of Classification

“…Yet, in practice, philosophers tend to argue for the superiority of one particular philosophical approach to an issue. As both Stock (2010) and Szostak (2011) have recently noted, it would be a mistake for information scientists to ground their work in any one philosophical approach. This preference for plurality is shared by most, but not all, of the authors in Ibekwe-SanJuan and Dousa (2014).…”