An abstractive text summarization using deep learning in Assamese

Abstractive text summarization with long short-term memory (LSTM) is a prominent strategy in natural language processing that tries to construct a compact and coherent summary of a given text by learning the semantic representation of the input text. In this study, we provide a seq2seq-based LSTM network model with attention to the encoder–decoder to construct a short sequence of words containing coherent and human-like created summaries, including crucial information from the original text. We obtained a dataset from Asomiya Pratidin, an Assamese online news website, including around 10,000 Assamese text articles and related human-written summaries. Our primary objective is to create an abstractive summarizer in Assamese while minimizing train loss. We reduced train loss to 0.3032 and produced fluent summaries during our study. © 2023, The Author(s), under exclusive licence to Bharati Vidyapeeth's Institute of Computer Applications and Management.