Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Survey on Aspect-Based Sentiment Analysis: Tasks, Methods, and Challenges

Wenxuan Zhang, Xin Li, Yang Deng, Lidong Bing, Wai Lam

TL;DR

ABSA seeks fine-grained sentiment understanding at the level of aspects by jointly identifying sentiment elements and their relations. The paper provides a comprehensive taxonomy and survey of ABSA tasks, from single-element to compound tasks, and reviews modeling paradigms, PLM-based approaches, and cross-domain/cross-lingual transfer. It highlights key datasets, evaluation practices, and practical directions such as multimodal ABSA and lifelong learning, offering guidance for researchers and practitioners. The findings underscore PLMs as a central driver of progress while identifying challenges in data, robustness, and cross-domain generalization with implications for real-world deployment.

Abstract

As an important fine-grained sentiment analysis problem, aspect-based sentiment analysis (ABSA), aiming to analyze and understand people's opinions at the aspect level, has been attracting considerable interest in the last decade. To handle ABSA in different scenarios, various tasks are introduced for analyzing different sentiment elements and their relations, including the aspect term, aspect category, opinion term, and sentiment polarity. Unlike early ABSA works focusing on a single sentiment element, many compound ABSA tasks involving multiple elements have been studied in recent years for capturing more complete aspect-level sentiment information. However, a systematic review of various ABSA tasks and their corresponding solutions is still lacking, which we aim to fill in this survey. More specifically, we provide a new taxonomy for ABSA which organizes existing studies from the axes of concerned sentiment elements, with an emphasis on recent advances of compound ABSA tasks. From the perspective of solutions, we summarize the utilization of pre-trained language models for ABSA, which improved the performance of ABSA to a new stage. Besides, techniques for building more practical ABSA systems in cross-domain/lingual scenarios are discussed. Finally, we review some emerging topics and discuss some open challenges to outlook potential future directions of ABSA.

A Survey on Aspect-Based Sentiment Analysis: Tasks, Methods, and Challenges

TL;DR

ABSA seeks fine-grained sentiment understanding at the level of aspects by jointly identifying sentiment elements and their relations. The paper provides a comprehensive taxonomy and survey of ABSA tasks, from single-element to compound tasks, and reviews modeling paradigms, PLM-based approaches, and cross-domain/cross-lingual transfer. It highlights key datasets, evaluation practices, and practical directions such as multimodal ABSA and lifelong learning, offering guidance for researchers and practitioners. The findings underscore PLMs as a central driver of progress while identifying challenges in data, robustness, and cross-domain generalization with implications for real-world deployment.

Abstract

As an important fine-grained sentiment analysis problem, aspect-based sentiment analysis (ABSA), aiming to analyze and understand people's opinions at the aspect level, has been attracting considerable interest in the last decade. To handle ABSA in different scenarios, various tasks are introduced for analyzing different sentiment elements and their relations, including the aspect term, aspect category, opinion term, and sentiment polarity. Unlike early ABSA works focusing on a single sentiment element, many compound ABSA tasks involving multiple elements have been studied in recent years for capturing more complete aspect-level sentiment information. However, a systematic review of various ABSA tasks and their corresponding solutions is still lacking, which we aim to fill in this survey. More specifically, we provide a new taxonomy for ABSA which organizes existing studies from the axes of concerned sentiment elements, with an emphasis on recent advances of compound ABSA tasks. From the perspective of solutions, we summarize the utilization of pre-trained language models for ABSA, which improved the performance of ABSA to a new stage. Besides, techniques for building more practical ABSA systems in cross-domain/lingual scenarios are discussed. Finally, we review some emerging topics and discuss some open challenges to outlook potential future directions of ABSA.
Paper Structure (34 sections, 4 equations, 5 figures, 4 tables)

This paper contains 34 sections, 4 equations, 5 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Introduction
  3. Background
  4. Four Sentiment Elements of ABSA
  5. ABSA Definition
  6. Modeling Paradigms
  7. Sequence-level Classification (SeqClass)
  8. Token-level Classification (TokenClass)
  9. Machine Reading Comprehension (MRC)
  10. Sequence-to-Sequence (Seq2Seq)
  11. Pipeline Method (Pipeline)
  12. Datasets & Evaluations
  13. Single ABSA Tasks
  14. Aspect Term Extraction (ATE)
  15. Aspect Category Detection (ACD)
  16. ...and 19 more sections

Figures (5)

  • Figure 1: An example of the four key sentiment elements of ABSA.
  • Figure 2: Taxonomy of ABSA tasks, with representative methods of each task.
  • Figure 3: The relations between the four sentiment elements, single ABSA tasks, and compound ABSA tasks.
  • Figure 4: Demonstrations of the four types of unified methods for the ACSA task.
  • Figure 5: Different modeling paradigms for tackling the ASTE task, where (a), (b), (c), and (d) are simplified illustrations of the methods proposed in TwoStage aaai20-aste, JET emnlp20-aste-position, BMRC aaai21-aste-bimrc, and GAS acl21-gabsa respectively.