A systematic review and research contributions on aspect-based sentiment analysis using twitter data

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.