rank-fusion

⚡️A Blazing-Fast Python Library for Ranking Evaluation, Comparison, and Fusion 🐍

Repository with basic information about Rank Aggregation Problem.

Learning Latent Semantic Representations of Paintings for Personalized Recommendation

Keyless, uv-native web search + read for AI agents: ddgs multi-engine search with de-correlated rank fusion, Trafilatura extraction to paginated Markdown, keyless arxiv and github search. Self-hosted SearXNG without Docker, an optional Tor layer for .onion search and fetch, NordVPN/SOCKS5 egress, one-call init, and a doctor.

Easy to use rank aggregation software for recommendation systems

RAG pipeline for medical question-answering. Fuses lexical and dense retrieval (MedCPT, Contriever, Specter + FAISS) with OpenAI, Gemini, and HuggingFace LLMs. Supports iterative multi-round reasoning, strict typing, structured observability, and a clean layered architecture

Question-Answering (QA) system powered by Retrieval-Augmented Generation (RAG). The system leverages advanced methods such as Rank Fusion and Cascading Retrieval for optimized document retrieval and contextual QA generation.

Training-free, CPU-only lexical–dense fusion for conversational-memory retrieval — 0.752 vs 0.640 Hit@1 over BM25 on LoCoMo, fully reproducible. Paper + reproduction code.

Weighted, adaptive, calibration-free Reciprocal Rank Fusion (RRF) of heterogeneous retrieval channels

Rank fusion and reranking

Ship a search index to a device as signed, content-addressed bundles: fail-closed Ed25519 verification, FastCDC chunking so the next release downloads only the chunks that changed, then FAISS + BM25 rank fusion locally. Python. 309 tests, 98% coverage.

Hybrid product search on Qdrant: semantic + BM25 with RRF fusion, exact/fuzzy SKU & EAN-13 matching with typo/OCR correction. FastAPI, FastEmbed, multilingual (uk/ru/en).

Merge 768-dim and 1536-dim retrieval indexes during a re-embedding migration by calibrated rank, not raw score

Leakage-aware multi-retriever fusion study on MovieLens-1M, with a reviewable FunRec service overlay.

Implementation of various vector rank fusion algorithms

Fusão Híbrida para Recuperação de Informação Jurídica Brasileira (STIL 2026)

Scientific citation retrieval: eight retrievers fused with weighted RRF, 0.6965 NDCG@10 held out. M1 AI Information Retrieval challenge, Paris-Saclay.

Collapse retrieved chunks to parents and near-duplicates to one candidate, before fusion, so ranking does not depend on chunk size

RAG service over your own documents using the Claude API: chunk, embed, retrieve, and answer with citations - or say 'I don't know' when the context doesn't cover it

Keyword, vector & hybrid search with MongoDB — interactive demo comparing $text, $vectorSearch and $rankFusion (MongoDB 8.3), with a minimal RAG pipeline using OpenAI or Ollama.