Sarah Mallet
IRISA/INSA, Campus Universitaire de Beaulieu, France
Mireille Ducassé
IRISA/INSA, Campus Universitaire de Beaulieu, France
Download article
Published in: Proceedings of the 3rd International Workshop on Automatic Debugging; 1997 (AADEBUG-97)
Linköping Electronic Conference Proceedings 1:9, p. 87-102
Linköping Electronic Articles in Computer and Information Science vol. 2 1:9, p. 87-102
Published: 1997-09-10
ISBN:
ISSN: 1650-3686 (print), 1650-3740 (online)
The power of deductive systems in general is that programs express what should be done and not how it should be done. Nevertheless; deductive systems need debugging and explanation facilities. Indeed; their operational semantics is less abstract than the declarative semantics of the programs. If users have to understand all the low level details of the operational semantics much of the benefits of using a deductive system is lost.
Existing explanation systems for deductive databases produce proof trees to be shown to users. Although useful; proof trees give a fragmented view of query evaluations; and users face a; most of the time large; forest of proof trees.
We propose a new data structure; called the DDB tree; which merges the information of a proof tree forest into one concise tree. A DDB tree gives a global picture of a query evaluation in a dramatically reduced structure with no loss of information. DDB trees can be shown to users or can be analyzed further by an explanation system.
