Requeries

Requeries are follow-up queries issued during information retrieval tasks with the goal of refining, expanding, or redirecting the original search. The concept is common in interactive search environments, where users iteratively adjust terms, constraints, or topics after inspecting initial results. It also applies to automated systems that generate revised queries in response to user signals such as clicks, dwell time, or explicit relevance judgments, in order to improve subsequent results.

In practice, requeries occur in search engines, database query interfaces, and data exploration tools. They enable users to resolve ambiguity, locate subtopics, or meet specific constraints (for example, a date range or a geographic region). They also appear in natural language interfaces and conversational search, where subsequent turns may reformulate the question or add qualifiers.

Mechanisms for generating requeries include relevance feedback, query expansion, session-based minimization or diversification, and learning-to-rank loops. Requeries may be explicit, originating from user input, or implicit, driven by interaction data. Some systems treat requeries as part of a broader iterative search process or as a dialog in which the system prompts for clarification.

Benefits of requeries include higher precision, improved recall for targeted intents, and better user satisfaction in long-running searches. Challenges include added latency, risk of query drift or topic wandering, potential user fatigue, and the need to balance exploration with efficiency.

References to interactive information retrieval and query reformulation literature discuss similar ideas under various labels, such as iterative search, relevance feedback, and conversational search.