AutoParSe: An automatic paradigm selector for nouns in Konkani

Abstract

In this paper, we discuss a rule based method which automatically assigns paradigms to Konkani nouns using morphophonemic rules, stem formation rules and relevance score of the paradigms. The first contribution is computation of relevance score of a paradigm, which is computed using a corpus and paradigm differentiating measure assigned to inflectional suffixes in the paradigm. Relevance score helps assign multiple paradigms to the input word wherever appropriate. The other contribution is a method for computing paradigm differentiating measure for inflectional suffixes. We have proposed a pruning technique based on derivational suffixes to further improve the precision. The experimental study has been carried out using the Konkani WordNet and the Asmitai Corpus. The proposed method successfully assigned relevant paradigms to 10,068 nouns with F-Score of 0.93.