Development of Punjabi WordNet

Abstract

Natural language processing (NLP) tasks such as word sense disambiguation, machine translation (MT) and part-of-speech tagging etc. require large scale lexical resources. These lexical resources have already been developed for digitally advanced languages, such as English, but yet to be developed for widely spoken but digitally young languages, such as Punjabi. WordNet is one such lexical resource that can be used for variety of NLP tasks ranging from digital dictionary to automated MT. The basic building block of WordNet is synset, a word sense with which one or more synonymous words are associated. Each synset in WordNet is linked to other synsets using lexical and semantic relations. Lexical relations are between word forms, whereas semantic relations exist between two whole synsets. This paper presents lexical and semantic relations of Punjabi WordNet, including synonymy, hypernymy/hyponymy, meronymy/holonomy, entailment and troponymy etc. It also illustrates the process of creation of Punjabi synsets from Hindi synsets, design of synsets and semantic databases for implementation of semantic relations for Punjabi WordNet.