Methods for measuring cross media content exposure across multiple publishers in a privacy preserving way are disclosed. The solutions described can accommodate publishing platforms that vary in size and computational resources without hindering performance or accuracy. Proposed is a solution to implement a privacy-safe HyperLogLog sketch implementation that can be applied to the cross media content exposure problem, as well as other problems that require privacy safe multi-set cardinality estimation. The processes described herein include publishers constructing encrypted HyperLogLog sketches and transmitting the encrypted sketches to worker computing devices, the worker computing devices joining the encrypted sketches, the worker computing devices finding the minimum value of each register in the joined sketch, and summing the registers to compute the multi-set cardinality estimate.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.