hybrid_search
Search by fusing semantic, lexical, and metadata signals to improve recall. Optionally weigh graph connectivity, expand neighbors, or explain results with graph paths.
Instructions
Search using combined semantic, lexical, and metadata signals. Provides better recall than single-signal search by fusing multiple relevance signals. v3 additive options: graphWeight adds a graph-connectivity (PageRank) channel, expandNeighbors appends one-hop neighbors of top results, explain annotates results with evidence paths from query anchors, lookFor ranks expansion neighbors by a free-text connection description.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| limit | No | Maximum results to return (default: 10) | |
| query | Yes | Search query text | |
| explain | No | Annotate each result with evidencePaths — graph paths connecting query anchor matches to the result (default: false) | |
| filters | No | Symbolic/metadata filters | |
| lookFor | No | Free-text description of the desired connection; expansion neighbors are ranked by similarity to it (lookForScore) | |
| weights | No | Layer weights (automatically normalized to sum to 1.0) | |
| graphWeight | No | Weight for the graph-connectivity channel (normalized PageRank). 0/omitted = channel disabled (default) | |
| expandNeighbors | No | One-hop neighbor expansion: neighbors of the top-K results are appended with a damped score and re-sorted |