Privacy Preservation Of Semi-structured Data Based On X-K~m-Anonymity

In the information age,people's various behavioral data are collected in large quantities,for example,personal income bills,medical information and other information collected by enterprises or organizations themselves,as well as take-out platforms,e-commerce shopping platforms,social platforms,etc.by third-party applications.These collected information can be published and shared by some organizations.The sharing of information provides convenience for some scientific investigations,commercial predictions,etc.,but it also poses a great hidden danger to personal privacy.Personal privacy includes personal identity information,relationship information between individuals and groups,etc.The privacy preservation of personal identity information is the most basic focus in privacy protection work.Other privacy preservation work must be carried out on this basis..Among them,the privacy preservation of relational tables and social network diagrams has a wide range of applications.XML(Extensible Markup Language)is widely accepted and used as an important Internet data expression and data exchange format It has very flexible expressive capabilities and is not restricted by Internet platforms,operating systems and programming languages.At the time of data release,there are also a large number of document data in XML format containing personal information.At present,research hotspots on privacy protection mainly focus on relational tables or social network diagrams.Research on relationship tables or social network diagrams is very rich,and privacy models and protection methods are also becoming mature.While on semi-structured data,the research on privacy preservation is rare,the research is not deep enough,and it is difficult to adapt to XML documents in different DTD formats.This paper proposes a semi-structured data privacy preservation model with strong versatility:X-km-anonymity model,and designs a reasonable data searching method for attack scenarios with knowledge background to obtain the sensitive data which has risk of leakage.Sensitive data is then anonymized with an anonymous method based on restricted publishing.The anonymous algorithm uses a bottom-up heuristic algorithm to provide privacy guarantee by satisfying X-km-anonymity.This paper proposes two kinds of reliable data utility analysis indicators.They are comprehensively measured from the two aspects of document similarity and query accuracy,and the feasibility of the anonymous method proposed by the paper and the advantages and disadvantages under different utility indicators are verified and compared by experiments.