Research On Sensitive Information Collection Methods Based On The Negative Survey And Their Applications

The negative survey is an emerging method of sensitive information collection.It just asks the participants to return the categories to which they do not belong to the collector.The collector can obtain the distribution of sensitive information by some dedicated statistical methods(called reconstruction methods).Because the participants only provide the categories to which they do not belong,the negative survey can accom-plish the task of collecting sensitive information while preserving individual privacy.Due to the simplicity and efficiency,the negative survey has attracted more and more attention.This dissertation studies the reconstruction methods of the negative survey,and proposes two reconstruction methods.Furthermore,this dissertation improves the existing location information collection method based on negative survey.Moreover,this dissertation also expands application areas of the negative survey,and applies the negative survey to collecting rating information and time-series sensitive information.Specifically,the contributions are listed as follows.(1)A novel reconstruction method(called NStoPS-LP)is proposed.It transforms the problem of reconstructing positive survey results from negative survey results into a linear programming problem,and the interior point method is used to solve this prob-lem.The experimental results on synthetic data indicate that comparing with existing reconstruction methods,NStoPS-LP returns reconstructed results without negative val-ues with higher efficiency.Moreover,the performance of the NStoPS-LP is also veri-fied by a real educational evaluation dataset.The experimental results demonstrate that the performance of NStoPS-LP is still excellent,even in practical application scenarios.(2)A new method based on the negative survey that aggregates location infor-mation is proposed to find gathering locations.Comparing with the existing negative survey based method,the proposed method can protect users' location privacy better,especially for the moving users' location privacy.The proposed method's control gran-ularity of balancing the accuracy and the privacy is more flexible and finer than the existing negative survey based method.The experimental results on synthetic data and real data demonstrate that the proposed method can find gathering locations easily.(3)The concept of consistency for the reconstructed results of multi-question nega-tive surveys is defined.The consistency of the existing two typical reconstruction meth-ods' reconstructed results is analyzed,i.e.NStoPS and NStoPS-I.Besides,a new recon-struction method(called NStoPS-C)is proposed which returns the consistent and non-negative multi-question reconstructed results.The experimental results on the synthetic dataset demonstrate that NStoPS-C can obtain more reasonable and accurate multi-question reconstructed results than NStoPS and NStoPS-I.Furthermore,we also use the negative survey to collect healthcare data,and then employ NStoPS-C to reconstruct the positive survey results.The simulated experimental results indicate that even on the real dataset the NStoPS-C can still obtain more accurate multi-question reconstructed results than existing typical reconstruction methods.(4)The negative survey is applied to rating the credits of online merchants,and the negative rating model is proposed.In comparison with the traditional rating model,the negative rating model can better preserve the privacy of the purchaser.To test the effectiveness of the negative rating model,we collect some real data from the Amazon website.Experiments are performed on these data,and the experimental results demon-strate that the negative rating model can preserve the privacy of the purchaser and the score obtained by the negative rating model is similar to the real score of the merchants.(5)We find that the traditional negative survey might disclose the privacy of the user when it is used to collect time-series data.We propose an improved strategy on the negative survey that can enable the negative survey to preserve the privacy of individual sensitive information when collecting the time-series data,and then apply the improved negative survey to collect the power consumption data.The security of the proposed method is analyzed.The experiments are conducted on synthetic data and real data to verify the performance of the proposed method.On the whole,this dissertation studies the theory and applications of the negative survey,which has important reference to the researches of sensitive information collec-tion methods.