Learning user Preferences for Top-k Querying - improving Learning of the Higher Ranked Objects

Alan Eckhardta and Peter Vojtášb

Department of Software Engineering, Charles University in Prague, Czech Republic.


In this paper we deal with user preference learning to enable top-k querying. Input for learning is user overall rating of a sample set of objects. We introduce a new method which favours the higher rated objects. We present a method for evaluating the top-k query results according to this preferences favouring top of the list. We compare our method to several methods and evaluate experiments on a real data set.

Keywords: Preference learning, User preferences, Top-k querying.

     Back to TOC