TELUGU TEXT SUMMARIZATION USING EXTRACTIVE METHOD

Abstract

Text summarization plays a crucial role in managing and understanding large volumes of textual data, especially in regional languages like Telugu, where limited resources and tools are available. This project proposes an extractive text summarization system for Telugu language using deep learning techniques such as Long Short-Term Memory (LSTM) and Bidirectional LSTM (Bi-LSTM). The system is designed to process Telugu text documents and generate concise summaries while preserving the original context and meaning. The dataset used for training consists of Telugu text and corresponding summaries, which are preprocessed using techniques such as tokenization, normalization, and TF-IDF vectorization to convert textual data into numerical form suitable for model training. The LSTM and Bi-LSTM models are trained using an encoder-decoder architecture, where the encoder learns the representation of input text and the decoder generates the corresponding summary. LSTM is particularly effective in capturing sequential dependencies and contextual information, making it suitable for summarization tasks. Experimental results show that the LSTM model achieves better performance with an accuracy of 71%, while the Bi-LSTM model achieves comparatively lower accuracy of 31%. The system also provides a user interface where users can upload Telugu text documents and obtain summarized outputs in real time. The proposed system demonstrates the effectiveness of deep learning approaches for regional language processing and provides a scalable solution for automatic text summarization. It can be applied in various domains such as news summarization, educational content analysis, and document management, contributing to advancements in natural language processing for Indian languages.