Morphological and Contextual Meaning Analysis in Colloquial Tamil Using Yarowsky Influenced Modified Apriori Algorithm

Abstract

Colloquial Tamil, often referred to as "spoken Tamil" or "conversational Tamil," is the informal variant of the Tamil language used in everyday conversations among native speakers. Colloquial Tamil is rich in idiomatic expressions and colloquialisms. The pronunciation in colloquial Tamil can vary significantly from region to region. The use of colloquial Tamil is influenced by social factors such as age, education level, and urban versus rural upbringing. It serves as a marker of identity and belonging within Tamil-speaking communities. Performing morphological analysis is the need of the hour, especially when regional dialect influenced Tamil content is used across social media. Morphological analysis over the informal Tamil content becomes complex as the language has varying meaning for the same word. To arrive at the actual meaning conveyed by the word, it is imperative to consider the coexisting words in the contextual sentence. These co-existing words roots to the prominent word. In this paper, to support and map the actual contextual meaning of the word under investigation, Yarowsky algorithm influenced modified Apriori algorithm is used. This algorithm along with the generated frequent item sets prove to be a suitable fit for the morphological analyzer in determining the actual contextual meaning. Often the frequent datasets are referred to as associative datasets.  As per the literature survey, experimental results demonstrate that Apriori based prediction and time complexity excels over the 2-level grammar based approach. This article concludes with the suggestion that prediction accuracy can be further improved, after performing Yarowsky algorithms initial pruning over the associative datasets.