PPM: Privacy Protection Method for Outsourcing Data Entry

Abstract

In China, manual approaches are still widely used in outsourcing data entry because of the poor recognition of OCR for Chinese. However, manual entries from images may cause leakage of users’ privacy. In this paper, we propose a privacy protection method to reduce privacy disclosure. First, we use the EAST algorithm to segment whole images into sub-images. Second, we propose a privacy association separation algorithm to protect users’ complete information which could be derived from privacy associations between attributes. Finally, we propose a sub-image allocation algorithm that varies for different relations between attributes, sub-sets, and data-entry clerks. We use AHP to model and optimize sub-contractor selection as well. The experimental results on OutsourcingData and U.S. Census Adult dataset show the advantages of our proposed method in terms of both the attribute privacy protection rate and the member privacy protection rate.