An Automatic Pun Word Identification Framework for Code Mixed Text

The computer systems must be able to process and comprehend inherent human language phenomena like humor in order to be more human-like, with a higher emotional quotient. This study will look at a subtype of humour-puns, which are a common category of wordplay-based joke with a focus on code- mixed puns, which have become increasingly popular on social media, in casual conversations, and in commercials, with the aim of developing a system that can automatically locate the pun position and recover the pun target. First investigate and categorize code-mixed sentence into two categories: Hindi words and English words, and then suggest a HLST based algorithm for recovering the pun targets. To achieve the desired results, the algorithm employs language models and similarity-based functionality known as Levenshtein distance. The method is test on a small collection of code-mixed advertisements containing pun words and it recovers 89% of the puns correctly.