# What is Reranking? Definition & Technical Mechanics

> **Source:** [SPIDITS AI AI Glossary](https://spidits.com/ai-glossary)  
> **Category:** Information Retrieval  
> **Canonical Citation:** [https://spidits.com/ai-glossary/reranking](https://spidits.com/ai-glossary/reranking)

## Definition
Reranking is a secondary step in RAG (Retrieval-Augmented Generation) pipelines where a highly accurate model evaluates and re-orders the candidate documents fetched during initial vector search, ensuring the most relevant context is placed at the top.

## Architectural Mechanics & Deep Dive
Reranking is a critical optimization step in RAG pipelines. While initial retrieval using bi-encoder vector similarity is fast, it can miss contextual nuances. A reranking step uses a high-capacity cross-encoder model to re-evaluate and score the similarity between the query and retrieved documents, filtering out low-relevance chunks.

## Industry Applications & Real-World Use Cases
RAG accuracy optimization, search engine optimization, and QA pipeline enhancement.

## Frequently Asked Questions

### Why is reranking needed if we already have vector search?
Vector search is fast but can miss complex context nuances. Reranking filters out semantic noise, keeping only the highly relevant chunks.

### What type of model is commonly used for reranking?
A Cross-Encoder model, which compares the query and document together to compute a deep semantic score.

## Related Terminology
- [rag](https://spidits.com/ai-glossary/rag)
- [semantic-search](https://spidits.com/ai-glossary/semantic-search)
- [cross-encoder](https://spidits.com/ai-glossary/cross-encoder)

---
*From the SPIDITS AI Knowledge Base (336+ terms indexed). Explore full technical definitions: [https://spidits.com/ai-glossary](https://spidits.com/ai-glossary)*