A BELIEF-DRIVEN DISCOVERY FRAMEWORK BASED ON DATA MONITORING AND TRIGGERING

Abstract

A new knowledge-discovery framework, called Data Monitoring and Discovery Triggering (DMDT), is defined, where the user specifies monitors that “watch" for significant changes to the data and changes to the user-defined system of beliefs. Once these changes are detected, knowledge discovery processes, in the form of data mining queries, are triggered. The proposed framework is the result of an observation, made in the previous work of the authors, that when changes to the user-defined beliefs occur, this means that, there are interesting patterns in the data. In this paper, we present an approach for finding these interesting patterns using data monitoring and belief-driven discovery techniques. Our approach is especially useful in those applications where data changes rapidly with time, as in some of the On-Line Transaction Processing (OLTP) systems. The proposed approach integrates active databases, data mining queries and subjective measures of interestingness based on user-defined systems of beliefs in a novel and synergetic way to yield a new type of data mining systems.