Retrieve (Search)
AI-generatedSummary
Search the selected RAGFlow dataset using a question and return matching chunks with a combined context string for downstream use.
Inputs
- Dataset ID (required) — Selects the dataset to search.
- Question / Query (required) — The question or query sent to RAGFlow for retrieval.
- Similarity Threshold — Minimum similarity score, default 0.5; sent as the retrieval similarity threshold.
- Top K — Maximum number of chunks requested, default 5.
Output shape
One item per input item, containing a single record with chunks, context_text, total_chunks, and query.
Each chunk includes content, similarity, doc_name, and img_id. Missing chunk fields are replaced with empty strings or 0. context_text joins chunk content with separators; total_chunks is the number of returned chunks. Retrieval errors stop execution unless Continue On Fail is enabled.