Fault diagnosis in databases for business processes

Abstract

Business processes involve data that can be modified or updated by various activities. These data must satisfy the business rules associated to the process. As the information treated in a business process tends to be extensive, data are normally stored in a relational database, and hence the database has to be analyzed to determine whether the business rules are satisfied and what values are incorrect. This paper proposes the use of model-based diagnosis in the business processes scenario. This scenario combines business processes, business rules, relational databases and where the faults are the instances of the variables introduced by the users. These considerations make it necessary to introduce a new way for representing the model, and the design of new algorithms to solve it. This model provides a means for the detection of incorrect tuples of different tables of the database by avoiding the analysis of the full database. Furthermore, in order to manage the current business rules, the use of a constraint paradigm is proposed and by using Max- CSPs to isolate incorrect values.