Hierarchical Clustering of Verbs

摘要: In this paper we present an unsupervised learning algorithm for incremental concept formation, based on augmented version of COBWEB. The is applied to the task acquiring a verb taxonomy through systematic observation usages in corpora. Using Machine Learning methodology Natural language problem required adjustments both sides. fact, formation algorithms assume input information as being stable, unambiguous and complete. At opposite, linguistic data are ambiguous, incomplete, possibly erroneous. A NL processor used extract semiautomatically from corpora thematic roles verbs derive feature-vector representation instances. order account multiple instances same verb, measure category utility, defined COBWEB, has been with notion memory inertia. Memory inertia models influence that previously classified given have classification subsequent verb. Finally, method identify basic-level classes acquired hierarchy, i.e. those bringing most predictive about their members.