Privacy Protection Technology escort program advertisement

Many Internet users have similar perceptions: just browsed a product from a shopping platform, and then opened another information app, and also brushed advertisements for similar products. This situation can not help causing doubts about the unknown truth. In fact, this is related to the recommendation technology of the platform's personalized advertising. The shopping platform is intelligently put into intelligent launch to other media based on the indirect portrait of the user in its APP. In this process, only technical docking between the platform will not obtain each other's personal information data. Among them, there is a basic guarantee for the realization of privacy protection technology for the realization of the commercial value.

It is reported that this kind of personalized advertisement recommendation is a model of programmatic advertising. Advertisers can procurely procure media resources and use technical means to match accurate audiences. Regarding the public's doubts, in the opinion of Huo Yan, deputy secretary -general of the China Advertising Association, this is a normal phenomenon. In the past, every major technical progress will be accompanied by similar concerns. "This is asymmetry with information, especially consumption. It is related to the reason why the platform's personalized advertising operation model, mechanism and technical principles behind them are unfamiliar. "

At the Seminar of "Construction of Technical Practice and Law Application of Mutual Data Environment" held in Beijing a few days ago, Douyin Group's data and privacy legal teams jointly released the "Report on the Application Rules of Privacy Computing Law" (hereinafter referred to as The "Report"), which proposes the adaptation of "Calculation of Privacy Protection" and data protection, provides solutions for personal information protection in programmatic advertising industries.

Among them, "Federal Learning", as one of the core technologies of privacy computing, can ensure that under the premise that the original data is not out of the library, it takes into account the protection of personal information and the realization of commercial value. "When advertisers and advertising platforms deploy federal learning systems in their respective environments, the data entered by all parties are only calculated in the input party's local federal learning system. ID ciphertexts, encrypted floating -point vectors (Embedding) or gradients (Gradients) and other data, and original data such as user features, labels, labels are only on local platforms, thereby avoiding the flow of original data. " Gao Zhen, deputy director of the Douyin Group Data and Privacy Law, talked about the operation of "Federal Learning" in digital advertising scenarios.

At the seminar, Liu Jingyan, the relevant person in charge of the huge engine privacy compliance, also introduced the "PRL-ECC" (Private Record LINKAGE-ERLIPTIC CRYPTOGRAPHY) used by the platform, which enabled the platform When the transaction request, first process the personal information data through the key, and then rely on the hash algorithm, so that neither the two parties cannot launch the explicit device number to further achieve the protection of user data privacy in the advertising link.

For the "PRL-ECC" solution, the director of the Digital Economy and Legal Innovation Research Center of the University of Foreign Economic and Trade expressed its recognition: "This plan is based on the basis Follow. "However, licenses also said that although PRL is one of the effective solutions for personal information sharing in the field of digital advertising, in the context of continuous improvement of public privacy, how to comprehensively implement the legitimate, legal, necessary, and integrity of personal information processing. In principle, there is still a long way to go.

"Privacy computing technology can sustainly escort the development of programmatic advertising, protection of personal information and stimulating economic vitality, but relevant practitioners need to increase popular science and help eliminate public concerns." Huo Yan particularly emphasized in an interview with the media. It is hoped that all parties represented by huge engines such as huge engines will continue to strengthen the user's privacy protection solutions in the programmatic advertising scenario, provide solid shields for ensuring the security of programmatic advertising business data, and lead the digital marketing industry to become healthier to be healthier. Development.

- END -