Using Domain Knowledge to Rank SPARQL Query Results According to User Preferences

Abstract

A growing number of applications for the Semantic Web are expected to be able to address user preferences in addition to well defined constraints when searching through large volumes of linked data. SPARQL, the most popular language for querying linked data repositories, does not have built-in tools for representing and processing preferences. The need for more adequate expression of user queries is well recognized by the Semantic Web community and two main approaches were offered to address this need. The quantitative approach associates a scoring function with each query result allowing results to be ordered according to their relevance to user preferences, while the qualitative approach relies on extending SPARQL with modifiers to express and rank preferences. The major concern with the former is that numerical scoring functions are hard to define, while the later cannot handle multiple preferences at the same time. In this paper, we suggest utilizing domain knowledge to process and rank SPARQL query results according to user preferences without the need to modify SPARQL functionality. An example application is used throughout the paper to illustrate the proposed approach.