When should I use Reciprocal Rank Fusion (RRF)?

Use RRF when you want to combine multiple search rankings into a single list without needing to manually normalize scores from different scoring systems like BM25 and Cosine Similarity.

Does this skill support PostgreSQL?

Yes, it includes specific templates for implementing hybrid search using the pgvector extension for embeddings and built-in full-text search (FTS) for keywords.

How does cross-encoder reranking work?

A cross-encoder is a neural model that processes the query and a document together to predict relevance. It is typically used as a second-stage reranker on the top candidates returned by a faster hybrid search.

What is hybrid search in RAG systems?

Hybrid search combines semantic vector search (which understands meaning) with traditional keyword-based search (which finds exact matches) to provide more accurate and relevant results.

Hybrid Search Implementation

Name: Hybrid Search Implementation
Author: duanbiao2000

byduanbiao2000

0•

데이터 과학 및 ML

Implements robust hybrid search systems by combining vector similarity and keyword-based retrieval for enhanced RAG performance.

This skill provides standardized patterns and implementation templates for hybrid search, allowing developers to combine the semantic understanding of vector embeddings with the precision of keyword matching. It covers essential fusion techniques like Reciprocal Rank Fusion (RRF) and linear combination, while providing ready-to-use code for popular databases like PostgreSQL (pgvector) and Elasticsearch. It is particularly useful for building Retrieval-Augmented Generation (RAG) systems that need to handle domain-specific terminology, product codes, or exact identifiers that pure semantic search often misses.

주요 기능

01Cross-encoder reranking patterns for high-precision document retrieval

02Advanced result fusion using Reciprocal Rank Fusion (RRF) and linear weighting

03PostgreSQL integration templates using pgvector and full-text search (FTS)

040 GitHub stars

05Standardized scoring normalization for disparate search metrics

06Elasticsearch hybrid search configurations for BM25 and dense vectors

사용 사례

01Improving recall in RAG systems where semantic search misses technical terms or codes

02Optimizing document retrieval for large-scale enterprise knowledge bases

03Building domain-specific search engines requiring concept matching and keyword accuracy

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add duanbiao2000/obsidiandoc26 hybrid-search-implementation

For use in Claude.ai and ChatGPT

Download Skill