Representation and Acquisition of Verbal Polysemy
Fumiyo Fukumoto · 1995
In this paper, we focused on an appropriate definition of polysemy in temas of distributional behaviour of words in monolingual texts and proposed a dustering method which recosnkes verbal polysemies from a textual corpus. The chara~eristic of ou~ clust~ 8 algoritkm is that it explicitly introduces new entities, i.e. hFpothetical t~’bs when an entity is judged poly~mous and mmociates them with contexts which are subcente~t~ of the context of the original entity. We report the results of two experiments. The first is concerned with verifying the ~ect of our algorithm which explidtly introduces a hypothetical ,3erb. The second one is based on the assumption of possible causes of failures mentioned in the firs t experiment sad is conducted to see how various parameters alfect the clustering results. The results of experiments demonstrate the ei~ectiveness of our proposed method.