Automation and Validation of Annotation for Hindi Anaphora Resolution

The process of labelling any language genre by which one can extract useful information is called annotation. This provides syntactic information about a word or a word phrase. In this paper, an effort has been made to provide the algorithm for semiautomatic annotation for Hindi text to cater anaphora resolution only. The study was conducted on twelve files of Ranchi Express available in EMILLE corpus. The corpus is originally tagged for demonstrative pronouns. The detection of the pronouns is supported by the incorporation of seven tags. However the semantic interpretation of the demonstrative pronoun is not supported in the original corpus. In this paper an effort has been made to automate the process of tagging as well as the handling of semantic information through addition tags. It was conducted on 1485 demonstrative pronouns. The average accuracy of precision, recall and F measure is 74, 71 and 72 respectively. Keywords—Annotation; natural language processing; demonstrative pronoun; semantic category; indirect anaphora; semiautomatic annotation