Searchplex ID Retrieval terminology registry

Task

Query Routing

Canonical ID https://id.searchplex.net/query-routing/

Choosing where or how a query should be handled, such as selecting a source, index, retriever, tool, model, strategy, or no-retrieval path.

Scope

Query routing uses information about the query to make a control-flow decision. In distributed IR and federated search, it often means selecting sources or servers. In RAG and agentic systems, it may mean selecting among retrievers, tools, indexes, models, or retrieval strategies.

Observed terms

  • Source Selection academic
  • Resource Selection academic
  • Router industry

Part of

  • Query Understanding Often uses query-understanding signals to decide which source, retriever, model, or strategy should handle the query.

Not equivalent to

  • Query Routing is not Agentic Retrieval. Query routing chooses a path for a query; agentic retrieval describes an agent-controlled retrieval loop that may include routing decisions.
  • Query Routing is not Federated Search. Query routing may be used inside federated search for source selection, but federated search is the broader multi-source retrieval pattern.
  • Query Routing is not Query Classification. Query classification assigns a query to labels or categories; query routing uses query signals to choose a handling path, source, retriever, model, or strategy.
  • Query Routing is not Query Decomposition. Query decomposition creates subqueries or subproblems; query routing chooses the handling path or retrieval resource.
  • Query Routing is not Retrieval Planning. Retrieval planning chooses an overall retrieval strategy or sequence; query routing chooses a handling path, source, retriever, tool, or model for a query.
  • Query Routing is not Routing. Routing in classic IR uses standing queries or profiles over incoming documents; query routing chooses where or how a live query should be handled.

Learn more

ID