# What is ColBERT? Definition & Technical Mechanics

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

## Definition
ColBERT (Contextualized Late Interaction over BERT) is a retrieval model that employs late interaction to compare queries and documents. By storing token-level embeddings and comparing them at search time, it maintains high retrieval quality while enabling fast queries.

## Architectural Mechanics & Deep Dive
ColBERT (Contextualized Late Interaction over BERT) is a retrieval model that employs token-level late interaction to compare queries and passages. Instead of compressing an entire document into a single vector, ColBERT encodes query and document tokens separately and computes maximum dot-product alignments at search time. This achieves high semantic precision while remaining fast.

## Industry Applications & Real-World Use Cases
Fine-grained document retrieval, passage search, and high-precision RAG systems

## Frequently Asked Questions

### What is late interaction in ColBERT?
A mechanism that encodes the query and document separately at token level and delays their interaction until query time using simple vector dot-products.

### How does ColBERT compare to standard bi-encoders?
It is significantly more precise because it aligns query tokens to document tokens directly rather than compressing the whole document into a single vector.

## Related Terminology
- [bert](https://spidits.com/ai-glossary/bert)
- [embedding](https://spidits.com/ai-glossary/embedding)
- [semantic-search](https://spidits.com/ai-glossary/semantic-search)

---
*From the SPIDITS AI Knowledge Base (300+ terms indexed). Explore full technical definitions: [https://spidits.com/ai-glossary](https://spidits.com/ai-glossary)*