# Document definition in information retrieval

Is there common formal definition of Document in information retrieval field? In many researches authors don't define the term Document(maybe because it is evident for them). Wikipedia says "text file, document (computer science) a computer file that contains text" but it doesn't seem as common formal definition.

Do you know the common formal definition of Document, or, if not, do you know any researches where that term was defined?

• Why do you want a formal definition of document? What do you plan to use this definition for? What good would knowing this definition do you? Personally, I don't see why the information retrieval field needs a formal definition of document. – Peter Shor Aug 19 '13 at 20:12
• I'm going to make some research in text mining. It's difficult for me to use the term without its definition – Ilya Boltnev Aug 19 '13 at 20:47
• @PeterShor: perhaps because many IR measures (recall precision, ...) are based on the number of relevant/non-relevant/retrieved/... documents. Originally a document was synonymous of "unstructured text". Nowadays it is a more general term (it can be a multimedia file, a web page, ...) and you can see it like "unit of information on a particular subject" that can satisfy (or help to satisfy) the information need of the user that queries the IR-system. – Vor Aug 19 '13 at 21:14
• some/many IR methods eg LSA/LSI regard a document as a "set of words [symbol strings separated by space, punctuation etc]" – vzn Aug 23 '13 at 17:14