Anserini

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

1191 stars · 673 forks observed · 2026-09-03

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

Pyserini is a Python toolkit for reproducible information retrieval research supporting both sparse (via Anserini/Lucene) and dense (via Faiss) representations. It bundles queries, relevance judgments, prebuilt indexes, and evaluation scripts for standard IR test collections, with a REST API and MCP server.

Use cases

• reproduce BM25 baseline runs on standard IR test collections
• run dense retrieval with prebuilt indexes for first-stage ranking
• build a multi-stage ranking pipeline with reproducible retrieval
• evaluate retrieval runs with trec_eval-style scripts
• search MS MARCO and other benchmark corpora
• expose retrieval via a REST API or MCP server

When to choose

• you need reproducible, benchmarked sparse and dense retrieval in Python
• you want prebuilt indexes and judgments for common IR test collections
• you're doing academic IR research or building RAG first-stage retrieval

When to avoid

• you need a production search engine with indexing, sharding, and serving out of the box
• you want a simple plug-and-play vector store without Java 21 dependencies
• your use case is general web search rather than benchmark-oriented retrieval

Facets

library · maturity active

search-engine rag machine-learning sdk python jvm cli cross-platform information-retrieval sparse-retrieval dense-retrieval bm25 faiss lucene test-collections first-stage-retrieval mcp-server rest-api search natural-language-processing research

Member repositories

For agents

markdown · JSON · MCP: product_card(name="castorini/anserini")