Mining Changes and Connections using Rough Set Theory

Mining data changes and connections from information systems (or databases) is made difficult by the different data behaviors and relationships across multiple data sets. When making a decision, such a dynamic and integrated knowledge base can be used to set useful rules (e.g., causality) that differ from the statistical associations in a single resource. In this paper, using techniques based on the rough set theory, we propose a change and connection mining algorithm for discovering a time delay between the quantitative changes in the data of two temporal information systems and for generating the association rules of changes from their connected decision table. We establish evaluation criteria for the connectedness of two temporal information systems with varying time delays by calculating weight-based accuracy and coverage of the association rules of changes, adjusted by a fuzzy membership function. keywords: rough set theory, fuzzy theory, change and connection mining, causality