The Impact of Metaphoric Text and Its Detection Process in Bangla Language for Enhanced NLP Performance

Metaphor detection plays a critical role in under-standing figurative language, particularly in morphologically rich languages like Bangla. Metaphorical expressions enhance the depth and cultural nuances of communication but pose significant challenges for natural language processing (NLP) systems. This paper addresses the lack of dedicated resources and methodologies for Bangla metaphor detection by developing a comprehensive framework. A benchmark dataset of annotated Bangla sentences is created, and both traditional machine learning and state-of-the-art deep learning models, including LSTM and BERT, are applied. The findings demonstrate that incorporating metaphor detection significantly improves the performance of downstream NLP tasks, such as sentiment analysis, machine translation, and text classification. This research not only highlights the importance of metaphor detection in Bangla NLP but also paves the way for future advancements in the domain by integrating linguistic, semantic, and contextual features.