Query log streams based incremental schema matching

Abstract

This paper proposes an idea for a query log streams based incremental schema matching system. Existing techniques include utilizing the usage information of the attributes in matching schemas. This paper proposes a further advancement of the technique by taking into consideration the constant and bursty nature of the attributes usage patterns and predicting schema attributes usage frequency - frequent of infrequent and new schema attributes usage information to perform incremental schema matching tasks so that the system is able to provide a better schema matching result by performing integration which is more relevant to the user. To the best of our knowledge, this paper represents the first to propose the technique of leveraging constant and bursty patterns of usage data and predicting schema attributes usage frequency and new attributes usage in incremental data integration tasks. The details of solution presented in this paper represents a first-cut work-in-progress approach, developing more comprehensive solution to the problem in ongoing future work.