In accordance with the May 2017 revision of the Act on the Protection of Personal Information, under certain rules, "anonymously processed" *1 personal information can now be used for purposes other than originally intended without the consent of each individuals. This will fast-track the utilization of personal data to create new businesses and services, and could lead to further improvement of public convenience.
NTT is conducting R&D of safe and highly useful anonymization technologies that process data in accordance with their attributes and intended use, such as our original Pk-anonymization technology*2. Also, leveraging the knowhow from usage of different anonymization technologies, our top-class performance and knowhow in national contests on anonymization technologies, and the knowhow from data assessment experiments conducted in various fields, we are pursuing various initiatives to provide support in creating more useful anonymously processed information, while ensuring proper compliance to anonymization criterions.
*1 Anonymously processed information refers to "information that has been processed to make personal information impossible to identify a specific individual, and from which it is impossible to restore this person's personal information."
*2 Pk-anonymization technology is a type of randomization technique for rewriting parts of the data, and involves stochastically changing pieces of data and inferring the original data through a machine learning method called Bayesian inference, to create highly useful anonymized data that satisfy the k-anonymity.
Image of anonymization processing
- Dai Ikarashi, Ryo Kikuchi, Koji Chida, Katsumi Takahashi: "k-Anonymous Microdata Release via Post Randomisation Method," International Workshop on Security (IWSEC), 2015
- Eizen Kimura, Satoshi Hasegawa, Koji Chida, Shoko Gamo, Satoshi Irino, Haku Ishida, Yukio Kurihara: "Evaluation of the anonymity and utility of de-identified clinical data based on Japanese anonymization criteria," MedInfo2017 - The 16th World Congress on Medical and Health Informatics(Poster), 2017
- Eizen Kimura, Koji Chida, Dai Ikarashi, Koki Hamada, Ken Ishihara: "Statistical disclosure limitation of health data based on Pk-anonymity," MIE 2012 - The 24th European Medical Informatics Conference (Poster), 2012
- Satoshi Hasegawa, Shogo Masaki, Rina Okada: "Design and Evaluation of A Practically Efficient Anonymization Library for Large Scale Data," Computer Security Symposium, 2017
- Satoshi Hasegawa, Ryo Kikuchi, Shogo Masaki, Koki Hamada: "Probabilistic K Anonymity High-Dimensional Data Publication via Matrix Factorization," Computer Security Symposium, 2016
- Satoshi Hasegawa, Shogo Masaki, Koki Hamada, Ryo Kikuchi: "A theoretical analysis on accuracy of reconstruction in probabilistic k-anonymization," Symposium on Cryptography and Information Security, 2016.
- Ryo Kikuchi, Dai Ikarashi, Koji Chida, Koki Hamada: "Data-Dependent Pk-Anonymization Method for Publishing Useful Anonymized Table," Symposium on Cryptography and Information Security, 2013
- Dai Ikarashi, Satoshi Hasegawa, Tatsuya Osame, Ryo Kikuchi, Koji Chida: "A Privacy Preserving Cross-tabulation which Guarantees k-Anonymity by Randomization for Numeric Attributes," Computer Security Symposium 2012
- Dai Ikarashi, Koji Chida, Katsumi Takahashi: "A Probabilistic Extension of k-Anonymity," Computer Security Symposium 2009