Analysis and Comparison of Antecedent Type of Demonstrative Pronoun in Context of Co-reference Resolution: A Corpus Based Study of Hindi for Monologue and Dialogue

In machine translation the co reference resolution has important significance. Without correct co-reference any translation has wrong interpretation. In this paper we investigated the type of antecedent for Hindi. We discussed the five values of this feature of antecedent. We analyzed 165 news items of Ranchi Express from EMILEE corpus of plain text. It consist 2382 sentences. Eight file of dialogue based from the same corpus has been analyzed which consist 871 sentences and from plain text 1511 sentences have been taken. In plain text 966 sentences have nominal antecedent, 387 have clausal antecedent and 153 do not have antecedent (abstract anaphora). On the other hand dialogue based files have 428 nominal antecedent, 334 have clausal antecedent and 108 do not have antecedent (abstract anaphora). Rest two features, Adjectival and propositional / Factual are either zero value or too low to be considered.