boudinfl/pke

Health v2 · maintenance only

Adoption not part of the score

1590 stars · 297 forks observed · 2026-08-28

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

pke is an open-source Python toolkit for keyphrase extraction that provides an end-to-end pipeline where each component (candidate selection, weighting, ranking) can be modified or extended. It implements many state-of-the-art unsupervised and supervised models and supports easy benchmarking on standard datasets.

Use cases

• extract keyphrases from documents in python
• extract keywords from text automatically
• compare keyphrase extraction models on benchmarks
• build a tagging system for articles
• implement a custom keyphrase extraction model
• get top n keywords for SEO or indexing

When to choose

• you need a standardized API to experiment with multiple keyphrase extraction algorithms
• you want to benchmark state-of-the-art keyphrase models like TopicRank or YAKE
• you need supervised keyphrase extraction models trained on SemEval-2010
• you want a pipeline where candidate selection and weighting are easily customizable

When to avoid

• you need keyphrase generation (abstracting new phrases) rather than extraction from the text
• you need production-grade speed or heavy-duty NLP beyond keyphrases
• you cannot use GPL-3.0 licensed code in your project
• you need a maintained library with frequent updates and recent dependency support

Facets

library · maturity maintenance

nlp search-engine machine-learning python cross-platform keyphrase-extraction keyword-extraction spacy unsupervised-learning supervised-models benchmarking natural-language-processing information-retrieval computational-linguistics

Member repositories

For agents

markdown · JSON · MCP: product_card(name="boudinfl/pke")