Search Index

A search index is a structure organized to find relevant records efficiently. A text index can map terms to documents containing them; a vector index supports finding nearby representations under a chosen similarity measure. An index may store content as well as lookup information, depending on the system.

A useful copy can become outdated

Imagine a policy assistant whose index contains searchable passages from a return-policy document. Editing the original policy does not necessarily update the index immediately. A successful source edit and a fresh search result are different events.

Elastic's index documentation illustrates how documents and searchable fields are organized in one search system. The general design must also record the source identity, version, and transformation that produced each indexed entry.

Searchability does not establish authority

An index can be rebuilt from retained sources if the necessary data and processing rules remain available. It should not be assumed to be disposable when it contains the only surviving information. Decide explicitly which store is authoritative.

For restricted documents, enforce current access rules on the retrieval path. Remove or invalidate obsolete entries and related caches as required. A high search score does not show that a passage is current, permitted, or sufficient to answer a question. Keyword and vector retrieval are possible methods; neither defines the complete quality of the search service.

Reference: Elastic: Index fundamentals.