Reconciling Schema Matching Networks Through Crowdsourcing

Schema matching is the process of establishing correspondences between the attributes of database schemas for data integration purposes. Although several automatic schema matching tools have been developed, their results are often incomplete or erroneous. To obtain a correct set of correspondences, usually human effort is required to validate the generated correspondences. This validation process is often costly, as it is performed by highly skilled experts. Our paper analyzes how to leverage crowdsourcing techniques to validate the generated correspondences by a large group of non-experts. In our work we assume that one needs to establish attribute correspondences not only between two schemas but in a network. We also assume that the matching is realized in a pairwise fashion, in the presence of consistency expectations about the network of attribute correspondences. We demonstrate that formulating these expectations in the form of integrity constraints can improve the process of reconciliation. As in the case of crowdsourcing the user's input is unreliable, we need specific aggregation techniques to obtain good quality. We demonstrate that consistency constraints can not only improve the quality of aggregated answers, but they also enable us to more reliably estimate the quality answers of individual workers and detect spammers. Moreover, these constraints also enable to minimize the necessary human effort needed, for the same expected quality of results.