[Math] What’s the meaning of the term RECALL in information retrieval

probabilityterminology

From the wiki of Precision and recall:

recall (also known as sensitivity) is the fraction of relevant
instances that are retrieved.

I can understand the literal meaning of "sensitivity", but quite confused with the term "recall".

Since I'm not a native English speaker, I can merely understand this word as "remember sth." or "a call on sb. to return", which has nothing to do with the meaning in such a context. Could someone explain why they use this terminology? Is there any origin of it?

Best Answer

From the dictionary:

verb [...]

  • cause one to remember or think of.
  • bring the memory or thought of someone or something to (a person or their mind).
  • call up (stored computer data) for processing or display.

noun

[...]

  • the action or faculty of remembering something learned or experienced. "he has amazing recall"

  • the proportion of the number of relevant documents retrieved from a database in response to an inquiry.

So "recall" here means the instances when something was recalled (retrieved) from the data.

The negation would be an instance when something relevant existed, but wasn't retrieved. This is like when you have heard someone's name before, but can't recall their name — can't retrieve it from your memory. Maximum recall means there are no false negatives: whenever there exists something relevant that can be recalled (retrieved), it is recalled (retrieved). It is no surprise that the phrase total recall is used in the context of perfect memory.

(I do feel that it may have been better to call the quantity being measured something like "level of recall" or "amount of recall" or "fraction of recall" or something, but I disagree that the usual meaning (the ones you stated) "has nothing to do with the meaning in such a context". It is the same meaning.)

Related Question