100

RésuméThis article presents a survey of the interactions between information retrieval and temporal aspects of documents. First the time evoked by the discourse and the historic situation time of the documents themselves are distinguished. The universe of discourse time should be taken into account in the indexation step of information retrieval. It can be processed by named entities extraction and more finely by natural language processing methods in order to determine temporal relations. The processing of metadata information if the latter does not follow very strict rules in fact is a neighbour problem. The publication time, which is in the traditionnal publishing world the main temporal metadata, becomes in the digital document world a fundamental piece of information. We introduce the notion of mutable and immutable collections. We quote the questions of time representation granularity.