Privacy preserving schema and data matching

In many business scenarios, record matching is performed across different data sources with the aim of identifying common information shared among these sources. However such need is often in contrast with privacy requirements concerning the data stored by the sources. In this paper, we propose a protocol for record matching that preserves privacy both at the data level and at the schema level. Specifically, if two sources need to identify their common data, by running the protocol they can compute the matching of their datasets without sharing their data in clear and only sharing the result of the matching. The protocol uses a third party, and maps records into a vector space in order to preserve their privacy. Experimental results show the efficiency of the matching protocol in terms of precision and recall as well as the good computational performance.