Abstract: The tasks that encloses the Semantic Web often requires semantic comparisons between classes, properties, relations and instances, that mainly consist in semantic similarity. As a consequence, the number of similarity functions used in this field has greatly increased in the last years. However, the selection of the best similarity functions to use in each case is a difficult task usually guided by heuristics and intuition. To overcome this problem, in this paper we propose a taxonomy of similarity functions as a first step to a semantic catalog of similarity functions. This taxonomy is induced by another taxonomy of data types. Our taxonomy allows selecting the similarity functions applicable to a pair of entities based on the data type used to represent them. It also eases the composition of functions according the composition of data types, and supports the inheritance of applicable functions from the more abstract to the more specific data types. As a consequence, it facilitates the application of similarity functions in an automated way. Finally, we also populate this taxonomy with some of the most popular similarity functions used in Semantic Web.

Keywords: Semantic Web, semantic comparison, similarity measures

ACM Classification Keywords: 1.5.4 Pattern recognition applications. I.2.4 Knowledge Representation Formalisms and Methods.

Link:

TOWARDS A SEMANTIC CATALOG OF SIMILARITY MEASURES

Alfredo Sánchez-Alberca?, Rafael Lahoz-Beltra? and Juan Castellanos-Peñuela?

http://www.foibg.com/ijicp/vol01/ijicp01-02-p03.pdf