A study on machine learning techniques for the schema matching network problem

Schema matching is the problem of finding semantic correspondences between elements from different schemas. This is a challenging problem since disparate elements in the schemas often represent the same concept. Traditional instances of this problem involved a pair of schemas. However, recently, there has been an increasing interest in matching several related schemas at once, a problem known as schema matching networks. The goal is to identify elements from several schemas that correspond to a single concept. We propose a family of methods for schema matching networks based on machine learning, which proved to be a competitive alternative for the traditional matching problem in several domains. To overcome the issue of requiring a large amount of training data, we also propose a bootstrapping procedure to generate training data automatically. In addition, we leverage constraints that arise in network scenarios to improve the quality of this data. We also study a strategy for receiving user feedback to assert some of the matchings generated and, relying on this feedback, improve the final result’s quality. Our experiments show that our methods can outperform baselines, reaching F1-score up to 0.83.