Affiliations: [a] Resbee Info Technologies Private Limited, Thuckalay, Tamil Nadu, India | [b] Cerebrum Technology Ltd, Pune, Maharashtra, India | [c] AP/CSE, Kalasalingam Institute of Technology, Krishnankoil, Tamil Nadu, India | [d] Computer Science and Engineering, Sathyabama Institute of Science and Technology, Chennai, Tamil Nadu, India
Correspondence:
[*]
Corresponding author: N.S. Ninu Preetha, Research Associate, Resbee Info Technologies Private Limited, Thuckalay, Tamil Nadu 629175, India. E-mail: ninu.n.s@gmail.com.
Abstract: Recently, Aspect-based Sentiment Analysis (ABSA) is considered a more demanding research topic that tries to discover the sentiment of particular aspects of the text. The key issue of this model is to discover the significant contexts for diverse aspects in an accurate manner. There will be variation among the sentiment of a few contexts based on their aspect, which stands as another challenging point that puts off the high performance. The major intent of this paper is to plan an analysis of ABSA using twitter data. The review is concentrated on a detailed analysis of diverse models performing the ABSA. Here, the main challenges and drawbacks based on ABSA baseline approaches are analyzed from the past 10 years’ references. Moreover, this review will also focus on analyzing different tools, and different data utilized by each contribution. Additionally, diverse machine learning is categorized according to their existence. This survey also points out the performance metrics and best performance values to validate the effectiveness of entire contributions. Finally, it highlights the challenges and research gaps to be addressed in modeling and learning about effectual, competent, and vigorous deep-learning algorithms for ABSA and pays attention to new directions for effective future research.
Keywords: Sentiment analysis, aspect-based sentiment analysis, machine learning algorithms, deep learning algorithms, performance analysis, research gaps and challenges
