|Semantic Web: is an expression for an approach for clearly describing objects on the web for archiving, managing search queries, and controlling processes. Object classifications in this approach, in principle, require lists of attributions of properties based on predicate logic. Competitive approaches are e.g. conceptual spaces that represent objects as points in multi-dimensional spaces, which allow faster processing. See also knowledge representation, semantic maps, conceptual spaces._____________Annotation: The above characterizations of concepts are neither definitions nor exhausting presentations of problems related to them. Instead, they are intended to give a short introduction to the contributions below. – Lexicon of Arguments. |
Semantic Web/Gärdenfors: Berners-Lee, Hendler, and Lassila (2001) Thesis: the Semantic Web is an extension of the current web in which the meaning of the given information is well-defined which facilitates the cooperation between human and robot.
GärdenforsVsBerners-Lee: instead of developing RDF (for information representation) and OWL (for expressing the ontology), one should examine how people process words.
Semantic Web: the dream is to develop a unified language for the representation of everything that exists on the web.
Berners-Lee: (Berners-Lee, 1998) Thesis: the semantic web is what we get when we apply the same globalization process, that the web originally applied to the hypertext, to the representation of knowledge. We create the central concepts of absolute truth, total knowledge, and total demonstrability, and see what we can do with limited knowledge.
Semantic Web/Shirky/Gärdenfors: Shirky (2004) Thesis: The Semantic Web assumes that many important aspects of the world can be represented in an unambiguous way and in a universally accepted way. After that, a lot of time is spent on the question which is the ideal XML format for such descriptions. This places too much emphasis on a wrong part of the problem: if the world were easy to describe, one could describe it in Sanskrit.
Solution/Gärdenfors: Conceptual Spaces.
Semantic Web/GärdenforsVsSemantic Web: Classification/similarity/Gärdenfors: when we consider how people deal with concepts, then the class structures mostly capture similarities between the objects. (Goldstone, 1994, Gärdenfors, 2000).
Problem: precisely a term like similarity cannot be expressed in the ontology of the Semantic Web.
Semantic Web/Shirky/Gärdenfors: (Shirky, 2003): Thesis: the Semantic Web is a machine for the creation of syllogisms. So if the Semantic Web wants to improve all domains where we use syllogisms, then one has to say that is practically nowhere.
Semantic Web/Conceptual Space/Gärdenfors: when we use conceptual spaces (with the representation of objects with properties as points in dimensions and regions) for the formation of the Semantic Web, taxonomies (concept hierarchies) occur by themselves. These hierarchies are used for symbolic structures. E.g. a robin is a bird: automatically originates from the fact that the region robin is a subregion of the region that represents birds.
Ontology/Computation: a Voronoi tessellation of the conceptual space can be created automatically by the computer and that is all that is necessary for an ontology. (See Noy & McGuinnes, 2001).
Identity: also results automatically from the coincidence of points.
Properties: Characteristics of properties such as transitivity and symmetry result from the format of conceptual spaces. E.g. the transitivity of "earlier than" follows from the linear structure of the time dimension.
((s) Conclusion: instead of word lists and lists of word lists with an explanation of concept hierarchies and distinction of word classes (such as e.g. the particular role of links and operators or quantifiers), one only needs the geometric properties of points.
The explanations of the differences in the word classes would in turn be given in words, which becomes superfluous by conceptual spaces.
Thus a) the handling by computer becomes easier and b) a possible circularity in the explanation of the role of words by words becomes more improbable.)
Conclusions/Gärdenfors: then you do not need any more inference engines on the symbolic level.
E.g. product search: you can find a similar wine with many similar and some different properties through the geometrical localization in the conceptual space._____________Explanation of symbols: Roman numerals indicate the source, arabic numerals indicate the page number. The corresponding books are indicated on the right hand side. ((s)…): Comment by the sender of the contribution.
The Geometry of Meaning Cambridge 2014