Personalizing Queries over Large Data Tables

摘要

We present a formal framework for the processing of preference queries over large data tables, in which user preferences are expressed as comparisons between attribute values (e.g. "I prefer Red to Black").The main contributions of the paper are as follows: (a) a formal framework for the statement of the problem, under no restrictions whatsoever on the preferences expressed by the user, (b) a rewriting algorithm that takes as input a preference query and returns a sequence of ordinary sub-queries whose evaluations construct the answer to the preference query, (c) a general definition of "skyline" and (d) a user-friendly interface supporting preference query formulation and incremental query evaluation with on-the-fly modification.