Logistic regression for automatic lexical level morphological paradigm selection for Konkani nouns

Abstract

Automatic selection of morphological paradigm for a noun lemma is necessary to automate the task of building morphological analyzer for nouns with minimal human interventions. Morphological paradigms can be of two types namely surface level morphological paradigms and lexical level morphological paradigms. In this paper we present a method to automatically select lexical level morphological paradigms for Konkani nouns. Using the proposed concept of paradigm differentiating measure to generate a training data set we found that logistic regression can be used to automatically select lexical level morphological paradigms with an F-Score of 0.957.