information retrieval: relevance, pertinence, precision and recall

The relevance of information in relation to some question was defined in the late 1950s when the Cranfield test was developed at the Cranfield College of Aeronautics . The two measures that were developed are precision and recall.


The extent to which information retrieved in a search of a library collection or other resource, such as an online catalog or bibliographic database, is judged by the user to be applicable to (“about”) the subject of the query. Relevance depends on the searcher’s subjective perception of the degree to which the document fulfills the information need, which may or may not have been
expressed fully or with precision in the search statement. Measures of the effectiveness of information retrieval, such as precision and recall, depend on the relevance of search results.

Compare with pertinence.

In information retrieval, the extent to which a document retrieved in response to a query actually satisfies the information need, depending on the user’s current state of knowledge–a narrower concept than relevance. Although a document may be relevant to the subject of the inquiry, it may already be known to the searcher, written in a language the user does not read, available in a format the reseacher is unable or unwilling to use, or unacceptable for some other reason.

In information retrieval, a measure of search effectiveness, expressed as the ratio of relevant records or documents retrieved from a database to the total number retrieved in response to the query;

Compare with recall.

recall In information retrieval, a measure of the effectiveness of a search, expressed as the ratio of the number of relevant records or documents retrieved in response to the query to the total number of relevant records or documents in the database;One of the main difficulties in using recall as a measure of search effectiveness is that it can be nearly impossible to determine the total number of relevant records in all but very small databases.

source: ODLIS: Online Dictionary for Library and Information Science


Fairthorne, Robert A. in “The Symmetries of Ignorance” distinguishes between two kinds of aboutness, extensional and intentional:

Robert Fairthorne writes: “The problem of helping those who are ignorant, in detail, of what people have said about things, is therefore solved by defining ‘aboutness’ in extension. That is by listing the things that are mentioned in a document. . . .” […]
(1) extensional “aboutness” takes into account the environment of the use and the production of a document (thus it is a relation, not an attribute);
and (2) intentional “aboutness,” which clearly cannot be determined from the study of the text alone: “It entails knowledge of how it is going to be used by what class of readers.”The Role of Classification in Subject Retrieval in the Future by Rolland-Thomas, Paule

[Photomedia Forum post by T.Neugebauer from Jan 13, 2007 ]