An Instance-Based Schema Matching Method with Attributes Ranking and Classification

Schema matching is a critical problem in many applications of database system, such as information integration, data warehouses, e-commerce, etc. So far, many solutions based on schema and element have been proposed. In this paper we present a new approach of instance-based matching building on the hypothesis that the corresponding attributes have equal relative importance. The framework of our approach consists of three parts: attribute ranking, attribute classification and matching phase. Unlike traditional approaches considering all attributes with the same importance, we take machine learning methods to prioritize all schema attributes by ranking and classification. During the matching phase, we construct an optimal objective function to find all equivalent attributes. In the end, our approach is validated by real datasets and the results show good accuracy.