Securing private data sharing in multi-party analytics

Gowtham Bellala, Bernardo Huberman


A general class of problems arises when datasets containing private information belong to multiple parties or owners and they collectively want to perform analytic studies on the entire set while respecting the privacy and security concerns of each individual party. We describe a solution to this problem in the form of a secure procedure for data mapping and/or linkage, which allows to identify the correspondence between entities in a distributed dataset. In contrast to existing methods this solution does not require either a trusted or semi-trusted third party, while being simple, efficient and scalable for both large datasets and number of parties.


privacy, data sharing, multiparty analytics

Full Text:



A Great Cities Initiative of the University of Illinois at Chicago University Library.

© First Monday, 1995-2019. ISSN 1396-0466.