What is the definition of recall?

Recall is a metric used in information retrieval and classification to assess the effectiveness of a model or system in identifying relevant instances. In more precise terms, recall refers to the percentage of truly positive documents (those that are relevant) that have been successfully retrieved by the search or classification system.

Choosing the option that states recall is defined as "the percentage of truly positive documents found" aligns perfectly with the standard definition used in evaluation metrics. This measure helps determine how well the system is performing in capturing all relevant information, highlighting its ability to minimize false negatives.

In a project context, focusing solely on the number of relevant documents identified, the total count of documents reviewed, or the accuracy of human coding decisions does not directly address the essence of recall, which centers specifically around the retrieval of true positives in relation to all actual relevant instances. Thus, the chosen definition encapsulates the key concept of recall in evaluating the effectiveness of classification and retrieval methodologies.