On the Feasibility of Informative Answers

In this chapter, we investigate the situation where a user’s request for data cannot be answered in the desired way because the data base does not contain the data requested. We argue that in such a situation it is not sufficient to report to the user the fact that the query failed, without providing the deeper reasons for the failure; an informative answer to a failing query, however, will adjust the user’s wrong assumptions about the contents of the data base and often will free him/her from the need of asking additional queries. We describe a formal method by means of which the reasons for the failure of a user’s query expressed in a relational calculus based language may be inferred and reported to the user as answer; special attention is paid to the amount of computation required for this answer generation procedure. We then show that a wide class of integrity constraints of a data base may be used to further improve the inference process and that any abuse of informative answers for the deduction of confidential information may be prevented easily.