Automated Abstractive Text Summarization Using Multidimensional Long Short-Term Memory

Abstract

Text summarization is a technique for condensing and presenting the most crucial information from a larger text in a succinct manner. It is a challenging task for human beings to manually summarize extensive documents due to the time required and cognitive effort involved. The increasing amount of textual data generated in recent years has led to a greater demand for efficient and automated text summarization methods. This study explores the utilization of deep learning and neural network concepts to create an automatic abstractive text summarization system for product reviews using multi-dimensional long short-term memory. Since, reading product reviews can be a time-consuming task, especially given the abundance and diversity of reviews written by numerous individuals. The proposed model was evaluated by comparing the generated summaries to reference summaries using metrics such as Rouge. Further, the performance of the model is compared with the prevalent state-of-the-art techniques. The proposed Multi-Dimensional LSTM model was evaluated using the Amazon food review dataset and compared with four other text summarization methods. The proposed model outperformed the prevalent models in terms of ROUGE-1 and ROUGE-L scores but was slightly outperformed in terms of the ROUGE-2 score.