Comparative Analysis of Multilingual and Cross-Lingual Models for Aspect-Based Sentiment Analysis

• Published in: International Conference on System Modeling & Advancement in Research Trends (Online) pp. 206 - 210
• Main Authors: Hussain, Sajithunisa, Khan, Rubina Liyakat, Quraishi, Suhail Javed, Singh, Anupam, George, Remya P., Ahmad, Nazia
• Format: Conference Proceeding
• Language: English
• Published: IEEE 06.12.2024
• Subjects: Analytical models; Aspect-Based Sentiment Analysis; Context modeling; Cross-Lingual; Data models; Deep learning; Multilingual; Natural Language Processing; Reviews; Sentiment analysis; Technological innovation; Training
• ISBN: 9798350380569
• ISSN: 2767-7362
• DOI: 10.1109/SMART63812.2024.10882527

Aspect-Based Sentiment Analysis (ABSA) is a finegrained sub-task of Natural Language Processing concerned with opinion bearing on certain aspects contained in text. The shift towards multicultural content on the digital platforms implies the need to have models which will perform the sentiment analysis of the multilingual content. In this study, a detailed comparison of ten developed methodologies of ABSA models is presented, including the theoretical background, data sets, and the assessment of performances of each model. We explore various techniques that are based on the lexical resources, statistical and machine learning, and state of the art deep learning networks. The study also sheds light on the potential strength and weakness of the proposed method and points out the possible further research directions that can be undertaken for improving the multilingual or cross-lingual sentiment analysis.