Privacy Sensitive Distributed Data Mining from Multi-party Data

Privacy is becoming an increasingly important issue in datamining, particularly in security and counter-terrorism-related applicationswhere the data is often sensitive. This paper considers the problemof mining privacy sensitive distributed multi-party data. It specificallyconsiders the problem of computing statistical aggregates like the correlationmatrix from privacy sensitive data where the program for computingthe aggregates is not trusted by the owner(s) of the data. It presents abrief overview of a random projection-based technique to compute thecorrelation matrix from a single third-party data site and also multiplehomogeneous sites.