Title: Bidirectional LSTM and self-attention mechanisms based multi-Label sentiment analysis

Authors: A.R. Arunarani

Addresses: Department of Computational Intelligence, School of Computing, College of Engineering and Technology, SRM Institute of Science and Technology, Kattankulathur, Chennai, 603203, India

Abstract: This study proposes the implementation of a novel optimisation-depend on deep learning algorithm for multi-label sentiment analysis (MLSA). The goal of the algorithm is to improve the accuracy of sentiment analysis, particularly in the context of e-commerce-related applications. This technique effectively categorise the text data into multiple sentiment classes, such as positive, negative, neutral, or other emotions, and to determine the overall sentiment expressed in a given text document. The challenge of MLSA on e-commerce data lies in the informal and often cryptic nature of the text, which can make sentiment analysis difficult. To address this, a novel optimisation-empowered bidirectional long-short term memory (Bi-LSTM) system with self-attention mechanisms is proposed in this research work. It uses the Bi-LSTM network to capture the sequential relationships between words in the manuscript and the self-attention mechanism to dynamically weigh the importance of different words in determining the overall sentiment expressed in the text.

Keywords: self-attention; bi-directional long short-term memory; multi-label sentimental analysis; deep learning; sentimental analysis; NLP; natural language processing.

DOI: 10.1504/IJSSE.2025.144565

International Journal of System of Systems Engineering, 2025 Vol.15 No.1, pp.67 - 78

Received: 31 May 2023
Accepted: 07 Jun 2023
Published online: 21 Feb 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article