Extension of Semantic Based Urdu Linguistic Resources Using Natural Language Processing

Urdu is the 8th largest spoken language in the world with more than 140 million speakers in subcontinent. Despite its abundant usage, the research work in computational linguistics in Urdu language is scarce. Identification of important terms is one of the pivotal process in knowledge representation in any domain. However, such work is not carried out yet for Urdu language. Furthermore, inclusion of multiword terms in semantic based language resources such as WordNet is not realized yet. This paper focuses on using automatic techniques to identify multi-word Urdu terms from several corpora using a variation of C-Value method and semantically connect them in Urdu WordNet. We have created an automatic process comprising of crawlers, scrappers that create semantic linking and retrieve example sentences from various online sources. The resulting resource, named as Extended Urdu WordNet, comprises of 735 synsets containing only multi-word terms as members. The resource is then manually validated by domain experts.