realitycheck-search

Natural language search finds related claims by converting your query into embeddings and retrieving relevant records from a vector index. It returns ranked results with IDs, text, type, domain, and similarity scores.

Semantic search translates natural language queries into vector embeddings and scans a vector index to match related content. It retrieves claims and supporting sources based on similarity scores rather than exact keyword matches.

Yes, semantic search supports domain and type filters to tailor results for specific research, policy analysis, or journalism contexts. This narrows the retrieved claims and sources to your target area.

Yes, semantic search requires an existing embeddings pipeline and a vector index to process natural language queries. These components convert text into vectors and store them for retrieval.

The best way to retrieve evidence trails is using natural language semantic search to scan your vector index for related claims and sources. This approach surfaces supporting records with similarity scores across domains.

Semantic search depends on the quality of your embeddings pipeline and vector index, requiring structured data to function. It retrieves related claims based on similarity scores rather than verifying their factual accuracy.