Privacy Preserving OLAP over Distributed XML Data: A Theoretically-Sound Secure-Multiparty-Computation Approach

Privacy Preserving Distributed OLAP is becoming a critical challenge for next-generation Business Intelligence (BI) scenarios, due to the “natural suitability” of OLAP in analyzing distributed massive BI repositories in a multidimensional and multi-granularity manner. In particular, in these scenarios XML-formatted BI repositories play a dominant role, due to the well-know amenities of XML in modeling and representing distributed business data. However, while Privacy Preserving Distributed Data Mining has been widely investigated, the problem of effectively and efficiently supporting privacy preserving OLAP over distributed collections of XML documents, which is relevant in practice, has been neglected so far. In order to fulfill this gap, we propose a novel Secure Multiparty Computation (SMC)-based privacy preserving OLAP framework for distributed collections of XML documents. The framework has many novel features ranging from nice theoretical properties to an effective and efficient protocol, called Secure Distributed OLAP aggregation protocol (SDO). The efficiency of our approach has been validated by an experimental evaluation over distributed collections of synthetic, benchmark and real-life XML documents.