Abstract:
Identifying people from anonymous location histories is important for two purposes. i.e. to clarify privacy risks in using the location histories and to find evidence of ...Show MoreMetadata
Abstract:
Identifying people from anonymous location histories is important for two purposes. i.e. to clarify privacy risks in using the location histories and to find evidence of who went where and when. Although linking with social network accounts is an excellent approach for such identification, previous methods need information about social relationships and have a limitation on the number of target data sets. Moreover, they make limited use of time information. We present models that overcome these problems by estimating the sameness and difference of people by using combinations of time and distance. Our proposed method uses these models along with multi-resolution models for both sides of linking, i.e. location histories and social network accounts. Evaluation using real data demonstrated the effectiveness of our method even when linking only one pseudonymized and obfuscated location history to 1 of 10,000 social network accounts without any information about social relationships.
Published in: 2020 IEEE 20th International Conference on Software Quality, Reliability and Security Companion (QRS-C)
Date of Conference: 11-14 December 2020
Date Added to IEEE Xplore: 21 December 2020
ISBN Information: