Recall is a performance metric in machine learning and AI that measures the percentage of correct results retrieved by a system out of all the correct results that should have been retrieved. It is particularly useful in applications such as search, categorization, and entity recognition.
For example, if an application is tasked with identifying dog breeds in a document and the document mentions 10 dog breeds, but the application only returns 5 correct breeds, the system’s recall would be 50%. This means that the system identified half of the relevant results, but missed the other half. Recall is important when it’s crucial to capture as many relevant results as possible, even if some false positives are included.