Enriching an ontology with WordNet based on similarity measures

Abstract

In this work we have used five semantic similarity measures and WordNet to add information to an ontology, the Common Procurement Vocabulary. The added information is used to automatically classify product descriptions according to the Common Procurement Vocabulary.
It is shown that the similarity measure proposed by Leacock and Chodorow is the most suitable for this task, out of the five measures compared. Leacock-Chodorow shows average precision between 0.684 and 0.711 and recall between 0.845 and 0.977, depending on whether threshold is used or not. Baseline average precision peaks at 0.592.