Explaining Unexpected Answers of SPARQL Queries

"Why am I not getting the right answer?" is a question many Knowledge Base users may ask themselves. In particular, novice users can easily make mistakes and find differences between the answer they expected and the answer they got. This problem is known as the unsatisfactory answer problem. A subproblem, where no answers are returned, has been widely studied and identifying failure causes can help users modify their queries to fit their requirements. But users may be unhappy with their results for multiple other reasons: they may be overwhelmed by too many answers, expect a particular answer that is not included, or even encounter a combination of these problems. In this paper, we classify the various types of unsatisfactory answers, and propose algorithms to compute generalized failure causes. We evaluate the performance of our algorithms and show that they perform comparably to existing problem-specific methods, while being more extensive.