A SYSTEMATIC REVIEW OF ASPECT-BASED SENTIMENT ANALYSIS: DEEP LEARNING TECHNIQUES, CHALLENGES, AND FUTURE SCOPES

Abstract

Aspect-based sentiment analysis (ABSA) is an important research area of the natural language processing (NLP) task concerned with detecting the sentiment polarity towards different aspects or features mentioned in text data. Deep learning based techniques have impacted the progress and success of ABSA methods from 2015 until 2025.The objective of this work is to provide a systematic overview of the state-of-the-art studies using deep learning techniques for ABSA and to synthesize the most significant challenges and opportunities for further research in this field. This study performed a systematic search in different databases and consider studies whose main focus is on deep learning techniques for aspect-based opinion mining. Deep learning techniques in ABSA have advanced from the RNN/CNN (2015-2018) to the transformer-based/hybrid models (2019-2025). However, implicit sentiment treatment, cross-domain adaption and computational efficiency are also far from solved. Future efforts would focus on hybrid architectures, their applications, multimodal integration, and real-world deployment.