Retrieval

Retrieval is the process of obtaining information from a storage system in response to a query or cue. In information retrieval, the aim is to identify documents or records that are relevant to an information need, often within a large corpus such as a library or the web. Retrieval systems typically build an index to organize content, then apply a ranking procedure to present the most relevant results first.

Common retrieval models include boolean retrieval, which uses exact matches; vector space models that represent documents and queries as vectors with term weights; and probabilistic or language-model approaches that estimate the likelihood of relevance. Evaluation relies on metrics such as precision and recall, and ranking-focused measures like mean average precision and normalized discounted cumulative gain. Benchmarks and annotated query sets are used to compare systems.

Retrieval also refers to memory recall in cognitive psychology, where information stored in memory is retrieved using cues, context, or by reducing interference. In databases and file systems, retrieval involves answering queries to fetch data or documents, often aided by indexes and query optimization to reduce latency and improve throughput. Retrieval tasks span text, multimedia, and structured data, with increasingly sophisticated methods for similarity, semantics, and context.

Challenges include linguistic variation, ambiguity, scale, dynamic collections, and user satisfaction. Recent advances in retrieval combine traditional IR with machine learning and neural models to improve relevance estimation, personalisation, and interactive search, enabling systems to adapt to user intent and feedback.