implementing-hybrid-search

Name: implementing-hybrid-search
Availability: InStock
Author: ionmidori

Combine vector and keyword search results using Reciprocal Rank Fusion.

Updated Dec 17, 2025

Search Tags:#search #RAG #retrieval #vector search #hybrid search #RRF #keyword search

Ready to run

npx skills add https://github.com/ionmidori/SYDBioedilizia --skill implementing-hybrid-search

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: implementing-hybrid-search
Source: https://github.com/ionmidori/SYDBioedilizia/tree/main/.gemini/skills/implementing-hybrid-search
Command: npx skills add https://github.com/ionmidori/SYDBioedilizia --skill implementing-hybrid-search

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill addresses the limitations of relying solely on vector or keyword search by providing methods to combine both, leading to more comprehensive and accurate search results.

Core Features & Use Cases

Hybrid Search Architecture: Implements patterns for combining vector and keyword search results.
Fusion Methods: Supports Reciprocal Rank Fusion (RRF), Linear Combination, and Cross-encoder reranking.
Use Case: When building a RAG system, you can use this skill to ensure that queries with specific keywords (like product codes or names) are not missed, while still leveraging semantic understanding for broader queries.

Quick Start

Use the implementing-hybrid-search skill to combine vector and keyword search results using Reciprocal Rank Fusion.

Frequently Asked Questions about implementing-hybrid-search

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I combine vector search and keyword search to improve RAG retrieval accuracy?▼

Hybrid search combines vector similarity and keyword-based search techniques to enhance retrieval accuracy in RAG systems by implementing fusion methods like Reciprocal Rank Fusion (RRF) and linear combination.

What is Reciprocal Rank Fusion and how does it merge search result lists?▼

Reciprocal Rank Fusion (RRF) is a fusion method that merges the ranked lists from vector and keyword search to boost search recall without requiring score calibration.

When should I use hybrid search instead of vector similarity alone in my search engine?▼

Use hybrid search when queries contain specific keywords like product codes or names that pure vector search might miss, ensuring comprehensive results while retaining semantic understanding for broader queries.

Does this hybrid search implementation support sparse-dense vector querying for Pinecone?▼

Yes, the implementation provides Python implementations for fusion methods and potentially supports sparse-dense vector querying for platforms like Pinecone to combine search results.

What is the best way to implement cross-encoder reranking for hybrid search results?▼

Cross-encoder reranking serves as a supported fusion method within the hybrid search architecture to refine and rerank the combined vector and keyword search results for greater accuracy.

How does linear combination compare to RRF for merging keyword and vector search ranked lists?▼

Linear combination and Reciprocal Rank Fusion (RRF) are both supported fusion methods for merging ranked lists, offering different mathematical approaches to consolidating vector and keyword search scores.