A fuzzy document representation supporting user adaptation in information retrieval

One of the main limitations of information retrieval systems is that user subjectivity can be handled at the level of query formulation, while the document representation is rigidly interpreted by the retrieval mechanism. Starting from the ideal that information retrieval can be seen as a multicriterion decision making activity, the authors introduce a new representation of documents which can be adapted to the user information needs. Documents are represented as compound entities, consisting of logical subparts, each one giving a distinct contribution to the overall significance of terms. The user intervention is allowed for both the specification of the documents´ subparts which must more heavily affect the retrieval and the indication of a linguistic quantifier to aggregate the single subpart degrees of significance. This model is defined within fuzzy set theory by the Yager OWA aggregation operators of criteria with unequal importance. It is shown how a document, with respect to the same query, has a different relevance according to the user interpretation of documents