Aspect-Based Sentiment Analysis (ABSA)
Also known as: ABSA, aspect-level sentiment analysis, feature-based sentiment analysis, Konu Bazlı Duygu Analizi (ABSA)
Aspect-based sentiment analysis (ABSA) is a fine-grained natural-language-processing task that detects sentiment separately for each aspect or feature mentioned in a text — such as a product's quality, price, or service — rather than scoring the document as a whole. It was consolidated as a shared task by Pontiki et al. in SemEval-2014 Task 4.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
ABSA fits when you have text data and want sentiment resolved at the level of specific aspects or features rather than a single overall label, and when aspect categories are either predefined or can be extracted automatically. A corpus of at least roughly 50 documents is recommended. If you only need one polarity per document, plain sentiment analysis is simpler; if there is no text data, ABSA cannot run.
Strengths & limitations
- Resolves sentiment at the aspect or feature level, capturing mixed opinions that a single document label would hide.
- Produces an interpretable, aspect-level profile of what is praised and what is criticised across a corpus.
- Supports both classification and explanatory goals on text data, including multilingual settings.
- Requires aspect categories to be predefined or reliably extracted; errors in aspect extraction propagate into the sentiment step.
- Needs a reasonably sized corpus (about 50 documents or more) to give stable aspect-level results.
- More complex than document-level sentiment analysis, since it adds an aspect-identification stage.
Frequently asked
How does ABSA differ from ordinary sentiment analysis?
Ordinary sentiment analysis assigns one polarity to a whole document. ABSA detects sentiment separately for each aspect or feature mentioned in the text, so a review that praises food but criticises service yields two distinct, opposing polarities instead of one blended label.
Do I need to define the aspects in advance?
Aspect categories can be predefined (for example quality, price, service) or extracted automatically from the corpus. Either way, the quality of aspect identification directly affects the per-aspect sentiment results.
How much text do I need?
A corpus of at least about 50 documents is recommended for stable aspect-level estimates. With far fewer documents the per-aspect results become unreliable.
Does ABSA work in languages other than English?
Yes, multilingual aspect-polarity classifiers can be used, but the model must match the language of your corpus for both aspect extraction and polarity classification to work well.
Sources
- Pontiki, M. et al. (2014). SemEval-2014 Task 4: Aspect Based Sentiment Analysis. Proceedings of SemEval 2014, 27-35. DOI: 10.3115/v1/S14-2004 ↗
- Schouten, K. & Frasincar, F. (2016). Survey on Aspect-Level Sentiment Analysis. IEEE Transactions on Knowledge and Data Engineering, 28(3), 813-830. DOI: 10.1109/TKDE.2015.2485209 ↗
How to cite this page
ScholarGate. (2026, June 1). Aspect-Based Sentiment Analysis (ABSA). ScholarGate. https://scholargate.app/en/text-mining/aspect-based-sentiment
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Named Entity RecognitionText mining↔ compare
- Sentiment AnalysisText mining↔ compare
- Text ClassificationText mining↔ compare
- Topic ModelingDeep learning↔ compare