Combining quantifications for flexible query result ranking

Abstract

Databases contain data and database systems governing such databases are often intended to allow a user to query these data. On one hand, these data may be subject to imperfections, on the other hand, users may employ imperfect query preference specifications to query such databases. All of these imperfections lead to each query answer being accompanied by a collection of quantifications indicating how well (part of) a group of data complies with (part of) the user's query. A fundamental question is how to present the user with the query answers complying best to his or her query preferences. The work presented in this paper first determines the difficulties to overcome in reaching such presentation. Mainly, a useful presentation needs the ranking of the query answers based on the aforementioned quantifications, but it seems advisable to not combine quantifications with different interpretations. Thus, the work presented in this paper continues to introduce and examine a novel technique to determine a query answer ranking. Finally, a few aspects of this technique, among which its computational efficiency, are discussed.