Named entity recognition in bangla

Abstract

Named Entity Recognition (NER) is a crucial task in natural language processing (NLP) that involves identifying and categorizing named entities within a body of text into predefined categories such as persons, organizations, locations, dates, and more. While substantial progress has been made in NER for high-resource languages like English, the development of NER systems for Bangla, a widely spoken language with complex morphology and syntax, remains challenging. In the context of the Bengali (Bangla) language, NER aims to extract and classify named entities specific to this linguistic and cultural context. The application of NER in Bangla is essential for numerous NLP tasks, including information extraction, question answering, and text summarization, which are particularly valuable for Bengali-speaking communities. Several challenges associated with Bangla language, which make NER a complex task. The language's morphological complexity and syntactic structure pose significant hurdles. Additionally, there is a scarcity of well-annotated datasets for Bangla NER. Variations in how named entities are represented in Bangla, such as different spellings and forms, further complicate the task. The development and effectiveness of Bangla NER systems largely depend on the availability of resources like labeled datasets, language models, and linguistic tools specifically designed for the Bangla. In this thesis, we are focusing on leveraging advanced deep learning models such as Convolutional Neural Network (CNN), Bidirectional Long Short-Term Memory (BiLSTM) network, and BERT-based model. The aim is to improve the accuracy and reliability of NER systems in Bangla.