State-of-the-art in Nepali NLP
We will first do a comprehensive review of the current state-of-the art of NLP tools and techniques developed for the Nepali language. The study will cover the areas such as developing language corpora, language models with word embeddings, dependency parsing and morphological analysis.
We will then explore more efficient methods suited to resource-constrained language like Nepali with code-switching to handle both the commonly used encodings: Latin and Devanagiri.
The developed methods will be evaluated in applications such as such as named entity recognition, sentiment analysis and intent classifications for chatbots.