Search approaches
To deliver relevant search results, you need to choose the right search approach for your data and use case.
The following table provides an overview of the fundamental search techniques available in Elasticsearch:
| Name | Description | Notes |
|---|---|---|
| Full-text search | Traditional lexical search with analyzers and relevance tuning | Essential foundation for keyword matching, works out of the box |
| Vector search | Machine learning-based search using natural language understanding | Includes semantic search, dense and sparse vector approaches |
| Hybrid search | Combines full-text and vector search approaches | Best balance for both keyword precision and semantic relevance |
| Ranking and reranking | Post-processing results to improve relevance | Optional ML-based enhancement for fine-tuned relevance |
| Geospatial search | Location-based search and spatial relationships | For maps, distance calculations, and shape queries |
Tip
Full-text search is a very powerful tool in itself. One of the key strengths of Elasticsearch is its flexibility, allowing you to start with full-text search and gradually incorporate more complex or resource-intensive approaches over time.
Once you've chosen your search approach(es), you'll need to select a query language to implement them. Refer to query languages for search use cases to learn about the available options.