EFFICIENT COMPUTATION OF DIVERSE QUERY RESULTS

Abstract

Query result diversification aims to enhance the quality of query results presented to users by ranking the results based on diversity so that more informative results are presented first. In this thesis, we study three problems related to the efficient computation of diverse query results. Firstly, we study the problem of evaluating diversity queries in the context of relational database systems where query results are diversified with respect to a sequence of attributes (known as the d-order) such that attributes that appear earlier in the d-order have higher priority for result diversification. Secondly, we study the optimization problem of evaluating multiple diversity queries in an online environment, and develop three new evaluation techniques. Finally, we study the novel problem of computing diverse query results in the context of spatial keyword search which is useful for applications such as trip-planning.