Taxonomy of Deep Learning Models for Aspect-based Sentiment Analysis: A Comprehensive Survey

In aspect-based sentiment analysis (ABSA), the aspects of an entity are extracted from a sentence before the sentiment polarity towards each aspect is determined. In a number of research domains, including ABSA, deep learning has recently taken the lead position as the dominant branch of machine learning. Given the tremendous advancements made by ABSA tasks in a variety of industries, including finance, customer feedback, movie reviews, etc., it is important to find deep learning algorithms that may be used to detect the elements and categorize the emotion. The surveys that are now available on ABSA tasks are insufficient to study the varied viewpoints, difficulties, and potential future research areas of deep learning for ABSA tasks. This research offers a comprehensive survey of various deep-learning models for ABSA problem resolution. This research presents a taxonomy that divides various deep learning models into 8 major categories in light of the findings. Additionally, this paper summarizes the works it has reviewed and discusses their key traits and difficulties. This work also provides future research directions by critically analyzing the outcomes of the existing deep learning models, which is helpful for the readers of this paper.