Measures of semantic similarity in folksonomies

Student thesis: Master thesis (including HD thesis)

  • Jan Kadlec
4. term, Computer Science, Master (Master Programme)
Folksonomies are new, user driven classification structures and an important part of Web 2.0. Folksonomies are the only one approach that can keep up with todays web expansion rate, by utilizing users as classificators of web\'s content. Folksonomies, when containing sufficient amount of data, can be exploited in several ways. This particular work concentrates on measures of semantic similarity in folksonomies. The aim of this work is to evaluate several semantic similarity measures on a sample of three datasets - delicious.com, Last.fm and medworm.com. Evaluation was done using grounding data from WordNet, Open Directory Project and medical oriented ontology. Results presented by this work indicate, that measures of semantic similarity can be used to successfully measure the similarities in folksonomies in several domains, among other, in domain of music and the one of web pages.
LanguageEnglish
Publication dateJun 2010
ID: 61077938