Building a WordNet for Persian verbs

Abstract. This article is a report of an ongoing project to develop a WordNet for Persian Verbs. To build this WordNet we apply the expand approach used in EuroWordNet and BalkaNet. We are now building the core WordNet of Persian verbs by translating the verbs of BalkaNet Concept Sets 1, 2 and 3. The translation process includes automatic suggested equivalences of English synsets in our WordNet editor, and then their manually refinement by a linguist using different dictionaries and corpora. We are also adding the frequent Persian verbs that are not included in the sets using an electronic Persian corpus. This core WordNet will be extended (semi)automatically. The most important fact about Persian verbs is that most of them are compound rather than simple. Compound verbs in Persian are formed in two major patterns: combination and incorporation. In many cases the compound verbs are semantically transparent, that is the meaning of compound verb is the function of the meaning of its verbal and non-verbal constituents. This suggests that many verbs in Persian WordNet can be directly connected to their non-verbal constituent in Persian WordNet and so inherit the existing relations among those words too.