311 wants classic keyword search over request summaries in SQL. What should developers implement?
Select an answer to reveal the explanation.
Short Explanation
For classic keyword search over 311 request summaries, implement full-text catalogs so you can search the words residents typed. You’ll meet the SQL keyword-search requirement without relying on approximate matches or downloading every request.
Full Explanation
Classic keyword search over 311 request summaries in SQL calls for full-text catalogs on those summaries. Implementing full-text catalogs lets you search request text using keyword semantics, matching the requirement to find the words residents typed rather than relying on approximate embedding similarity. Only approximate nearest-neighbor vector search with no full-text catalog fails this requirement: it finds semantic neighbors, not necessarily exact keywords. A client-side filter that downloads every request row before matching keywords does not implement SQL full-text search, scales poorly, and moves the search out of the database. Storing summaries as encrypted blobs that cannot be searched makes the text opaque to catalogs and queries, so keyword search cannot work. Exam caveat: if the city later adds semantic search, a hybrid pattern may combine full-text and vectors, but this request is specifically for classic keyword search. Practical check: create a full-text catalog and index on request summaries, search for a distinctive keyword in a known request, confirm that it returns a hit, and verify that encrypted blobs or client-side downloads are not used as substitutes.