Bengali Text Summarization with Attention-Based Deep Learning
Anupam Singha, N. R. Rajalakshmi · 2023
This research paper is about summarizing text using natural language processing with limited resources. Different techniques have been created, including abstractive and extractive methods, but the most recent is the use of recurrent neural network algorithms, which have shown better results. While other languages have automatic text summarizers, very few exist for Bengali. Furthermore, high-quality data is quite scarce. The research suggests a new abstractive text summarizer for Bengali that uses an encoder-decoder system with an attention mechanism. The dataset size evaluated for this work is small, but the quality is great; it is mostly composed of news articles. This work used a Long Short-Term Memory (LSTM) network model based on the seq2seq technique to make summarizing Bengali text more efficient. This abstractive approach generates new sentences containing information, indicating the model's ability to generate meaningful sentences.