Google company cryptographic protocol for confidential multiparty computation , which allows analysis and calculations on encrypted data sets from several participants, maintaining the confidentiality of each participant’s data (each participant is not able to obtain information about the data of other participants, but can perform generalized calculations on them without decryption). Protocol implementation code licensed under Apache 2.0.
Private Join and Compute allows you to transfer a private set of records to a third party, who will be able to analyze it and generally evaluate the differences with their set, but will not be able to find out the values of specific records. For example, it is possible to obtain information from an encrypted data set, such as the number of identifiers that match its set and the sum of the values of records with matching identifiers. In this case, it is impossible to find out exactly what values and identifiers are present in the set.
Private Join and Compute protocol, also referred to as Private Intersection-Sum, on protocol combination (Random Oblivious Transfer), encrypted and double disguise .
The proposed system may be useful, for example, when one medical institution has information about the health status of patients, and another about the prescription of a new preventive medicine. The “Private Join and Compute” protocol allows you, without disclosing information, to combine encrypted data sets and display general statistics that will allow you to understand whether the prescribed drug reduces the incidence of disease or not. Another example is that based on the database of accidents from the state traffic inspectorate and the base of the use of improved safety equipment in cars, it is possible to assess whether the appearance of these equipment affects the number of accidents.
Another example is when, based on the employee base of one company and purchase data from another, you can calculate how many employees from the first company made purchases from the second and for what amount. In the context of advertising networks, similar calculations can be made to evaluate the effectiveness of advertising campaigns, using lists of users who were shown an advertisement (or who clicked on a link) and who made purchases in an online store.
Source: opennet.ru