Part-of-Speech Tagging of Hindi Corpus Using Rule-Based Method

The main goal of analysis of NLP (natural language processing) is to understand natural languages by parsing them. In the practice of analyzing natural languages there exist various sub-tasks. These sub-tasks depend on inbuilt structure of language and do not require complete knowledge and understanding of language. Part-of-speech tagging is one of them. Part-of-speech tagging is basically a practice of assigning language-specific grammatical tags to each word of language-specific input text, according to word’s appearance in the text. These tags can be like noun, adverb, number, negative, etc. There exist a variety of taggers for most popular language in the world, i.e., English. But such taggers cannot be used for morphologically rich Hindi language as difference exists between structures of both languages. A “Rule-based system” is presented in this paper. 29 standard part-of-speech tags are used, including two special tags for date and time also in multiple formats. The special tags like punctuation, time, and date are based on regular expressions. Main aim of the proposed system is to increase automaticity and maintain high precision, while limiting the size of human made corpus. Proposed system uses human made corpus of around 9,000 words to increase tagging and rule-based (lexical features based) approach to decrease the size of already trained corpus. The system yields 91.84 % of average precision and 85.45 % of average accuracy.