Hybrid Search¶
Hybrid Search combines the best of AI-Semantic search and Lexical (keyword) search to provide optimal search results for most use cases.
Overview¶
Recommended Default Choice
We recommend Hybrid Search as the default search type for most use cases. It provides an optimal balance between natural language understanding and keyword precision, delivering the most comprehensive search results.
Hybrid search intelligently blends two powerful search approaches:
- AI-Semantic Understanding: Captures meaning and context using AI, enabling natural language queries
- Lexical Precision: Ensures exact keyword matches aren't missed, especially for technical terms
- Smart Ranking: Combines results from both methods to provide the most relevant matches
Key Features¶
- Flexible Query Types: Works well with both natural language questions and specific keywords
- Comprehensive Coverage: Finds relevant content through both semantic understanding and exact matches
- Configurable Balance: Adjust the AI-to-keyword ratio to optimize for your use case
- Best-of-Both Reliability: Reduces the risk of missing relevant results
Best Use Cases¶
Hybrid search is ideal for most use cases, including:
- SaaS Applications: Search by natural language queries and keywords
- Content Sites: Search by conversational queries, keywords, and metadata
- Knowledge Bases: Support both technical searches and natural language questions
- Product Catalogs: Match both specific product codes and feature descriptions
- Documentation: Find content by both specific terms and conceptual queries
Why Choose Hybrid Search¶
Hybrid search is the recommended choice when you need:
- Maximum search coverage and reliability
- Support for diverse search patterns
- Balance between precision and understanding
- Optimal results without manual tuning
Fine-Tuning
Adjust the hybrid_search_ai_ratio
parameter in the Search Documents API to control the balance between AI-semantic and lexical matching for your specific use case.
Search Type Selection
While Hybrid Search works best for most cases, you can easily switch to pure AI-Semantic or Lexical search if your use case requires it by changing a single parameter in your API request: search_type
.