Research On Privacy Preservation For Information Analysis In Location Big Data

In the era of big data,data mining technology for information analysis has developed rapidly,and the location information of people and things has been digitized by using position sensing technology such as mobile communication and sensing equipment.It can provide users with more comprehensive and intelligent location-based information of big data services.However,due to the diversified use of location services and cross-redundancy of content in mobile social networks,how to maintain the optimal balance between privacy protection and service availability is an urgent problem to be solved.According to the degree of protection of location privacy,the existing privacy protection technology for location services can be divided into three categories,based on heuristic privacy metrics,probabilistic guessing and location-based privacy protection based on information retrieval.Classical privacy control methods based on access control and anonymous location can not fully protect user privacy,and the existing research results still have the following shortcomings:1)Most of the existing researches ignore the attack pattern which matches the position data with the non-position data,so that the attacker can obtain or analyze the user's historical position data from other angles to get the background knowledge about the user.2)The data attributes of big data services for information analysis are complex and diverse,and only a single type of data distribution model can not meet the real needs.3)Security mining algorithm requires a reasonable allocation of privacy budget,improve classification accuracy while reducing errors.However,it neglects the problem of equality of authority among multiple participants in distributed environment.4)The traditional spatial heuristic method based on heuristic privacy measure can not generate an accurate and reasonable anonymous region ASR,which is easy to generate space debris,resulting in the redundancy of the algorithm,so that the associated mass data processing method is not efficient.In view of the above shortcomings,this paper presents a big data services for location-oriented information analysis of privacy protection methods.The work and innovation are as follows:Firstly,a new location-based data distribution model based on differential privacy protection is proposed,which can protect the sensitive information of location data and non-location data in big data query service.It is completely independent of the background knowledge of the excavator.Even if a malicious attack has grasped most of the relevant background information,it will not reveal the unknown information if the data set is repeatedly guessed on the whole data set.Simulation results show that the proposed algorithm can distribute location data with high accuracy for the query service under the specified privacy conditions.Secondly,an improved algorithm of decision tree based on interface privacy protection is proposed.The protection mechanism is built entirely on the access interface,and the interface provider is responsible for privacy protection.The data miner will not need to have the knowledge about privacy protection,and need not modify the mining task because of the security access restriction.At the same time,a data mining algorithm based on homomorphic encryption technology is proposed for the vertical partitioning of two-dimensional position information.It can guarantee the security of each participant without revealing their own private data,through the opaque calculation of the joint center box,So that semi-trusted third-party decision-makers global decryption and analysis of the final joint indicators of the completion of the information model of the building.Finally,a trusted third-party location anonymous server between mobile users and servers is proposed,where the location-sensitive hash algorithm uses an heuristic measure of privacy to partition the anonymous region and keep the proximity of the search target in Euclidean distance.Location data,can effectively blur the target node against malicious privacy attacks.Experiments show that the algorithm can effectively optimize the anonymity space to improve the degree of privacy protection,and has a better time complexity in the process of massive data set construction.