Published Paper Details:

Text summarization plays a crucial role in natural language processing, with applications spanning from document summarization to chatbots. This thesis examines various techniques for text summarization, focusing on both extractive and abstractive methods. It starts with a thorough literature review that traces the historical development of text summarization, discussing various methodologies and evaluation metrics. Following this, the study details data collection and preprocessing methods, then moves on to the implementation of both extractive and abstractive summarization models. Extractive summarization uses techniques like word frequency scoring and sentence selection, while abstractive summarization relies on deep learning models trained on preprocessed text. A novel approach is presented, utilizing a sequence-to-sequence (seq2seq) model with an attention mechanism. This method applies deep learning, specifically Long Short-Term Memory (LSTM) networks, to produce concise summaries from input texts. The model features an encoder-decoder structure, where the encoder processes the text and the decoder creates the summary. An attention mechanism is used to enable the decoder to concentrate on relevant sections of the input. The proposed method's effectiveness is tested on a diverse dataset of texts, including reviews and articles, and evaluated using ROUGE, BLEU, and METEOR metrics. The results show that the model performs competitively in generating abstractive summaries compared to extractive methods, effectively capturing key information while producing coherent and grammatically accurate summaries.