RESOLVING ANAPHORA IN MARATHI LANGUAGE
NLP is a sub-domain of artificial intelligence (AI) that focuses on how computers interact with human languages. It is a research field that focuses on the interaction between human language and computers. Anaphora resolution (AR) is one of the most exciting areas of research in natural language processing. Anaphora is a common occurrence in both written and spoken texts. It is evident that the text is not just the sentence strings; there could be numerous utterances with a variety of nouns and pronouns, each with its own set of references. Anaphora is resolved in a variety of foreign languages, including English, Spanish, and French, as well as in Indian languages like Hindi, Tamil, Telugu, Kannada, and others, but there has been very little work done in Marathi. Marathi is very difficult language due to its structure. One word has multiple meaning, for finding the correct tag of each word we have developed our own Marathi POS tagger to resolve anaphora because the existing POS tagger failed to resolve anaphora in Marathi text when tested with our corpus. We generated a corpus for the Marathi text because the standard corpus for the Marathi anaphora resolution was unavailable. This paper presents the rule based approach for the resolution of anaphora in Marathi language on the basis of Marathi transformational grammar. We also concentrate on the binding theory for the resolution of anaphora in Marathi. This system aids in the interpretation of Marathi text and attempts to resolve anaphora from the given text.
Anaphora Resolution, Antecedents, NLP, Binding Theory, Discourse Anaphora, Gender and Number Agreement, Marathi