Automatic and Graceful Repairing of Data Inconsistencies Resulting from Retroactive Updates in Temporal Xml Databases

Abstract

In multi-temporal XML databases supporting schema versioning, end users manipulate temporal XML elements, in a transparent manner, via suitable and corresponding schema versions. Furthermore, they could insert, modify or delete past, future or current temporal XML elements; manipulating past data are titled retroactive updates. However, updating a past element is not always a graceful task, since it could give rise to inconsistencies in the database. In fact, modifying a past element due to a detected error means that the database has included erroneous information during some period and, therefore, its consistency should be restored by correcting all errors and inconsistencies that have occurred in the past. Indeed, all processings that have been carried out during the inconsistency period and have used erroneous information have normally produced erroneous information. In this paper, we propose an approach which preserves data consistency in multi-temporal and multi-version XML databases. More precisely, after any retroactive update, the proposed approach allows (i) detecting and analyzing periods of database inconsistency, which result from that update, and (ii) repairing of all inconsistencies and recovery of all side effects.