A Novel Clustering-Based Approach to Schema Matching

Schema matching is a critical step in data integration from multiple heterogeneous data sources. This paper presents a new approach to schema matching, based on two observations. First, it is easier to find attribute correspondences between those schemas that are contextually similar. Second, the attribute correspondences found between these schemas can be used to help find new attribute correspondences between other schemas. Motivated by these observations, we propose a novel clustering-based approach to schema matching. First, we cluster schemas on the basis of their contextual similarity. Second, we cluster attributes of the schemas that are in the same schema cluster to find attribute correspondences between these schemas. Third, we cluster attributes across different schema clusters using statistical information gleaned from the existing attribute clusters to find attribute correspondences between more schemas. We leverage a fast clustering algorithm, the K-Means algorithm, to the above three clustering tasks. We have evaluated our approach in the context of integrating information from multiple web interfaces and the results show the effectiveness of our approach.