yangheng95/PyABSA

Health v2 · maintenance only

Flags: no_readme

Adoption not part of the score

1102 stars · 174 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

PyABSA is a PyTorch-based library providing state-of-the-art models for aspect-based sentiment analysis, including aspect term extraction, aspect polarity classification, and text classification. It also supports text augmentation and adversarial defense for NLP models.

Use cases

• extract aspect terms from product reviews
• classify sentiment polarity toward specific aspects in text
• train and fine-tune aspect-based sentiment analysis models
• perform general text classification
• augment training data for NLP tasks
• defend NLP models against adversarial attacks

When to choose

• you need aspect-level sentiment analysis rather than document-level sentiment
• you want pretrained state-of-the-art ATE/APC models with a simple Python API
• you are working in PyTorch and need reproducible ABSA baselines

When to avoid

• you only need simple overall document sentiment scores without aspect extraction
• you need a production serving system rather than a modeling library
• your project is not Python/PyTorch based

Facets

library · maturity active

nlp machine-learning deep-learning machine-learning deep-learning python cross-platform aspect-based-sentiment-analysis aspect-term-extraction pytorch text-classification adversarial-defense text-augmentation natural-language-processing

Member repositories

For agents

markdown · JSON · MCP: product_card(name="yangheng95/PyABSA")