Hybrid search
A retrieval approach that combines lexical and semantic search so exact matches and meaning reinforce each other.
TL;DR
Combines keyword and vector retrieval for better recall.
Quick facts
- Category
- AI & knowledge management
- Product tie-in
- KobiGPT RAG platform
- Related
- See compare and tools pages
- Locale
- TR and EN site
Why teams choose KobiGPT
- Understand terms before evaluating vendors.
- Link concepts to KobiGPT features (RAG, Kobi Kredi).
- Share glossary links with procurement and legal.
- Explore assistant use cases next.
Product facts
- Ücretsiz plan
- 100 doküman · 2 departman · 120 Kobi/ay(PLAN_CONFIG)
- Starter
- 1000 doküman · 5 departman · 1000 Kobi/ay(PLAN_CONFIG)
- Pro
- 12500 doküman · 25 departman · 12500 Kobi/ay(PLAN_CONFIG)
- Fusion query
- Dense and sparse prefetch with RRF fusion(apps/api/src/rag/qdrant.service.ts → searchHybrid)
Why combine two signals
Lexical retrieval is strong for expressions that must be written exactly, such as product codes and proper names. Semantic retrieval helps when the user phrases the question differently.
Hybrid retrieval combines those signals into one ranking. It keeps the strengths of both instead of trusting only exact matching or only vector proximity.
How KobiGPT applies it
KobiGPT’s Qdrant collections hold dense and sparse vector fields together. A query runs against both with a department filter, and results are combined with a Reciprocal Rank Fusion approach.
The practical implication is that scope matters as much as the model. A file not attached to the assistant should not enter retrieval merely because its text looks similar.
How to review quality
A hybrid score is not an absolute correctness score. The top passage can miss an exception or come from an outdated version, so the displayed source should be read.
The strongest archive improvements are selecting the right files, using clear headings, and maintaining versions. Adding files without organising scope makes hybrid retrieval noisy.
Applying Why combine two signals in a controlled workflow
A useful way to evaluate hybrid-search is to follow one real question from the source document to the final answer. Record which file was selected, what context reached the model, and what a reviewer would need to verify. This turns a definition into an operational check and makes the result comparable across teams.
The same check should include ownership and change management. Decide who updates the relevant documents, how an outdated result is reported, and which access boundary applies. KobiGPT can provide the assistant and the cited document context, but the organisation still owns the source material, permissions, and the decision made from the answer.
FAQ
What is Hybrid in practice?
Combines keyword and vector retrieval for better recall.
Does KobiGPT use this?
See product docs and feature pages for implementation details.
More reading?
Visit our blog and FAQ.
Accuracy disclaimer?
Educational content; verify for compliance decisions.
Does hybrid search show two result lists?
The user normally receives one combined ranking; dense and sparse signals are calculated underneath.
Does hybrid search guarantee the right document?
No. It improves ranking, but the source text and document freshness still need checking.
Comparison
| Feature | KobiGPT | Alternative |
|---|---|---|
| SME focus | Yes | N/A |
| Citations | When using RAG | N/A |
| Glossary depth | Growing | N/A |
| Tools | Interactive | N/A |