Hypothetical document embeddings (HyDE)
Draft a possible answer to guide a search for real passages; check the real sources afterward.
The fuller definition
A retrieval method that embeds a generated hypothetical document and uses that representation to search for relevant real documents.
- Generate
Draft a hypothetical passage
- Embed
Represent the passage
- Retrieve
Find actual sources
A simplified relationship, not a screenshot or a complete design.
A concrete example
A system generates a plausible answer-shaped passage before searching a document collection.
The distinction that matters
The generated passage guides retrieval; it is not evidence and must not be treated as a source.
Related terms
Embedding
A list of numbers used to compare items such as texts. Texts with similar meanings may have similar lists.
Documents & retrievalRetrieval
Searching a larger set of records for useful parts. Finding a passage does not prove it answers the question.
Documents & retrievalSemantic search
Searching by meaning rather than by exact words, so 'delay' can find 'postponed'.
Where it is taught
How document-grounded AI answers workSource and further reading
Gao et al. · Precise Zero-Shot Dense Retrieval without Relevance Labels (HyDE) (arXiv, December 2022)The definition and example are written for this library. The link gives background, official documentation or legal authority as labelled. Legal scope depends on the jurisdiction and current text; product features depend on release and settings.