A Text Processing and Translation Framework for Advancement of Maithili Language

This paper introduces a system for Maithili language processing that combines spell checking, stemming, dictionary lookup, and sentence reconstruction to build an accurate and efficient Maithili to English text translator. Additionally, the proposed system provides a feature for word auto-completion, which suggests words based on a given prefix. The process begins with a Trie-based spell-checking mechanism, which takes advantage of the Levenshtein distance to identify and correct spelling errors. Next, stemming reduces words to root forms by analyzing POS tags and applying Maithili-specific s uffix rules. Dictionary lookup follows, where the system retrieves word meanings by either exact matching or finding the closest term using longest substring matching. Finally, to form coherent sentences from jumbled words, the system uses spaCy for syntax analysis and GPT-2 to generate meaningful sequences. This multistage pipeline enhances accessibility to the Maithili language by unifying its components into a functional text translator, advancing language processing and enabling effective multilingual communication.