A Systematic Development of Stopwords for Manipuri Language Processing
摘要
Manipuri Language Processings are still in their nascent stages due to the lack of foundational resources, including annotated corpora, part-of-speech taggers, stemmers, and stopwords lists. This paper focuses on the development of a stopwords list for Manipuri language, a crucial step for text preprocessing in various NLP tasks. A corpus-based stopwords list was developed using term frequency (TF) and term frequency-inverse document frequency (TF-IDF) methods, and subsequently refined through the integration of English stopwords translated to Manipuri. This process resulted in the creation of a comprehensive and linguistically accurate 300 approx Manipuri stopwords. The efficacy of stopword removal was evaluated in the context of Manipuri Text Classification. The performance was assessed using precision, recall, and F1-score metrics, which revealed a 7% improvement in both accuracy and F1-score following stopwords removal.