Abstract: A data search method generates a search request at the client's data node and sends the request and provision of request for the search request to the client's data node. The client's data set is specified during the search request generation that comprises objects and labels matching the objects. The search request is sent to the data node of at least one data owner. Upon receipt of the search request, the objects are characterized by a set of features that comprises at least one feature from the owner's data set, at least one function of dependence between the set of features and labels is identified, the value of at least one loss function for the function of dependence between the set of features and labels is calculated; and information comprising the calculated loss function value is used as a response to the search request.

Abstract: The invention relates to data protection means and can be used for cooperative computing by two or more parties without revealing to each other the private data used. The technical result is to ensure the privacy of large amounts of data belonging to different parties, and to reduce the amount of memory secured from unauthorized reading during computations. Such an effect is achieved due to forming the first and the second private datasets, where the first and the second private datasets use user identifiers, wherein the first private dataset contains user ratings matched to the user identifiers, and the second private dataset, for each of the user identifiers, contains an array of user identifiers from the social environment of the respective users; by the user identifiers from the social environment of the selected user, computing the social rating of the selected user as a function of the number of users from the social environment and the rating values of the users from the social environment.

Abstract: The invention relates to data protection means and can be used for cooperative computing by two or more parties without revealing to each other the private data used. The technical result is to ensure the privacy of large amounts of data belonging to different parties, and to reduce the amount of memory secured from unauthorized reading during computations. Such an effect is achieved due to forming the first and the second private datasets, where the first and the second private datasets use user identifiers, wherein the first private dataset contains user ratings matched to the user identifiers, and the second private dataset, for each of the user identifiers, contains an array of user identifiers from the social environment of the respective users; by the user identifiers from the social environment of the selected user, computing the social rating of the selected user as a function of the number of users from the social environment and the rating values of the users from the social environment.