A clerk of courts wants a resident bot that finds answers in docket guides by meaning, not only exact keyword matches. Which grounding search style should they plan?

A clerk-of-courts resident bot that must find answers in docket guides by meaning needs semantic or vector-capable search for grounding. Residents paraphrase legal concepts; they rarely type exact headwords from the guide. Semantic retrieval maps natural-language questions to conceptually related passages so grounding can cite the right docket-guide content.

Lexical or keyword-based search that retrieves exact-term matched docket-guide passages for grounding fails because it requires exact-term matches, so paraphrased resident questions miss conceptually related passages and grounding cannot cite the right docket-guide content. Exact-phrase lookup search that retrieves verbatim matched phrase docket-guide passages for grounding fails because it requires verbatim phrase matches, so meaning-based resident questions that reword guide language miss the needed passages for grounding. Term-frequency ranked search that retrieves keyword-scored lexical docket-guide passages for grounding fails because it ranks by keyword frequency rather than meaning, so conceptually related but differently worded passages score poorly and grounding misses the right docket-guide content.

Exam caveat: semantic search still depends on quality chunking and domain vocabulary; hybrid keyword+vector can help when residents do use exact docket numbers. Operational check: test paraphrased resident questions against known guide sections, verify hits come from conceptually related passages rather than only exact-term matched, verbatim matched phrase, or keyword-scored lexical passages, and ensure grounded answers include enough passage context for clerks to verify.