Privacy-Preserving Mining of Web Service Conversations

Service Computing;Privacy;Service Interactions;K-anonymity;Data mining;Computer and Information Science;College of Engineering and Computer Science

Binzagr, FaisalMalik, Hafiz ;

University of Michigan

关键词: Service Computing;    Privacy;    Service Interactions;    K-anonymity;    Data mining;    Computer and Information Science;    College of Engineering and Computer Science;   

Others  :  https://deepblue.lib.umich.edu/bitstream/handle/2027.42/138104/Privacy-Preserving%20Mining%20of%20Web%20Service%20Conversations.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Organizations and businesses are exporting their applications as Web services seeking morecollaboration opportunities. These services are generally not used in silos. Indeed, the invocation of a service is often conditioned by the invocation of other services. We refer to the precedence relationships between service invocations as conversations or choreographies. As clients interact with Web services, they exchange an important quantity of sensitive data, hence raising the challenge to keep the privacy of various interactions. In addition to the data exchanged with Web services, users may consider the information about service usage as sensitive and would like to hide that information from third parties. However, conversation relationships may complicate the task of keeping such information secret.In this Thesis, we extend the traditional concept of k-anonymity introduced for databases to Web service conversations. The goal is to determine the extent to which the invocation of a service can be inferred from downstream invocations. We first use the FP-Growth algorithm for mining service invocation logs. The mining process returns the probabilities of service conversations.We then define a probabilistic k-anonymity technique for Web service conversations based on the results of the mining process. The proposed approach assists users in selecting Web services that best satisfy their anonymity requirements. We conducted extensive experiments using realworld Web services to prove the efficiency of the proposed approach.