Recall measures the coverage of the relevant documents of an information retrieval (IR) system. It is the fraction of all relevant documents that are retrieved. Consider a test document collection and an information need Q. Let R be the set of documents in the collection that are relevant to Q. Assume an IR system processes the information need Q and retrieves a document set A. Let |R| and |A| be the numbers of documents in R and A, respectively. Let |R ∩ A| denote the number of documents that are in both R and A. The recall of the IR system for Q is defined as R = |R ∩ A|∕|R|.
Precision and recall are the most frequently used and basic retrieval performance measures. Many other standard performance metrics are based on the two concepts.