Data privacy protection method and device, equipment, storage medium and computer program product

Abstract

The invention discloses a data privacy protection method and device, equipment, a storage medium and a computer program product, and the method comprises the steps: determining a confidence coefficient upper bound and a confidence coefficient lower bound of a purchase intention set according to the number of real intentions included in the purchase intention set, based on the purchase intention set, the confidence coefficient upper bound, the confidence coefficient lower bound, the data element background knowledge and the purchase record of the data element purchaser, establishing a release intention-based attack model, an efficiency maximization attack model and a purchase record inference attack model for the data element purchaser; a uniform attack defense model, an efficiency maximization attack defense model and a purchase record inference attack defense model based on release intention are constructed by taking a confidence coefficient threshold as a constraint condition, and privacy protection is performed on a data element purchaser according to the attack defense models, so that the privacy security of the data element purchaser is guaranteed, and the security of the data element purchaser is improved. And thus, the privacy leakage risk of the data element purchaser is reduced.

Description

Technical Field

[0001] The present application relates to the field of data security technology, and in particular to a data privacy protection method, apparatus, device, storage medium and computer program product. Background Art

[0002] Data has become the fifth largest production factor after land, labor, capital, and technology, a national basic strategic resource, and the core engine for the development of the digital economy. Especially with the advent of the era of big models, large-scale, high-quality, and diversified data sets are needed to improve model effects and generalization capabilities, and the strategic position of data elements has been further highlighted.

[0003] However, existing data privacy protection solutions mainly focus on protecting the privacy of data providers and data trading platforms, but do not involve the privacy protection of data buyers, who are at risk of privacy leakage. Summary of the invention

[0004] The main purpose of this application is to provide a data privacy protection method, device, equipment, storage medium and computer program product, aiming to solve the technical problem that the existing data privacy protection schemes mainly focus on protecting the privacy of data providers and data trading platforms, but do not involve the privacy protection of data buyers, and data buyers are at risk of privacy leakage.

[0005] To achieve the above objectives, the present application provides a data privacy protection method, which includes:

[0006] Acquire a purchase intention set published by a data element buyer, wherein the purchase intention set includes at least one real intention;

[0007] Determining an upper confidence limit and a lower confidence limit of the purchase intention set according to the number of real intentions included in the purchase intention set;

[0008] Establishing a release intention-based attack model, an efficiency maximization attack model, and a purchase record inference attack model for the data element buyer based on the purchase intention set, the confidence upper bound, the confidence lower bound, the data element background knowledge, and the purchase record of the data element buyer;

[0009] According to the attack model based on publishing intention, the attack model for maximizing efficiency and the attack model for inferring purchase records, a uniform attack defense model based on publishing intention, an attack defense model for maximizing efficiency and a purchase record inferring attack defense model are constructed with confidence threshold as a constraint condition;

[0010] The privacy of the data element purchaser is protected according to the uniform attack defense model based on publishing intention, the efficiency maximization attack defense model and the purchase record inference attack defense model.

[0011] Optionally, the privacy protection of the data element purchaser according to the uniform attack defense model based on publishing intention, the efficiency maximization attack defense model, and the purchase record inference attack defense model includes:

[0012] Expanding the purchase intention set to minimize the optimization problem in the uniform attack and defense model based on release intention or the efficiency maximization attack and defense model;

[0013] Sampling is performed based on the total number of data records that the data element purchaser is to purchase, and a purchase data set corresponding to the data element purchaser is obtained to ensure that the purchase data set meets the privacy protection constraints corresponding to the purchase record inference attack defense model.

[0014] Optionally, the step of expanding the purchase intention set to minimize an optimization problem in a uniform attack and defense model based on release intention or in the efficiency maximization attack and defense model includes:

[0015] The purchase intention set is set to be the same as the real intention, and in each iteration, each dimension in the data space is traversed;

[0016] Searching for optional neighborhood feature values ​​on the traversed dimension, and selecting the best neighborhood feature value from the optional neighborhood feature values;

[0017] After traversing each dimension, the best neighborhood feature value is added to the purchase intention set to minimize the optimization problem in the uniform attack and defense model based on release intention or the efficiency maximization attack and defense model.

[0018] Optionally, searching for optional neighborhood feature values ​​on the traversed dimension and selecting the best neighborhood feature value from the optional neighborhood feature values ​​includes:

[0019] Calculate the cumulative distance of eigenvalues ​​on the traversed dimension, and select the eigenvalue with the shortest cumulative distance as the optional neighborhood eigenvalue;

[0020] Calculating the number of additional records of the optional neighborhood feature value, the incremental impact of the privacy protection constraint, and the effectiveness score;

[0021] An optimal neighborhood feature value is selected from the optional neighborhood feature values ​​according to the number of additional records, the privacy protection constraint impact increment, and the effectiveness score.

[0022] Optionally, sampling according to the total number of data records to be purchased by the data element purchaser to obtain a purchase data set corresponding to the data element purchaser to ensure that the purchase data set satisfies the privacy protection constraint corresponding to the purchase record inference attack defense model includes:

[0023] Generating a data set by random sampling according to the total number of data records to be purchased by the data element purchaser, and eliminating the data set that does not meet the privacy protection constraint;

[0024] The utilization rates of the remaining data sets are calculated, and the remaining data set with the highest utilization rate is selected as the purchased data set.

[0025] Optionally, sampling according to the total number of data records to be purchased by the data element purchaser to obtain a purchase data set corresponding to the data element purchaser to ensure that the purchase data set satisfies the privacy protection constraint corresponding to the purchase record inference attack defense model includes:

[0026] Initialize the purchase data set by distributed sampling according to the total number of data records to be purchased by the data element purchaser;

[0027] Performing addition and deletion processing on the initialized data set, and checking whether the processed data set satisfies the privacy protection constraint;

[0028] If the privacy protection constraint is satisfied, then calculating the acceptance probability of the processed data set;

[0029] Whether to select the processed data set as the purchase data set is determined according to the acceptance probability.

[0030] In addition, to achieve the above purpose, the present application also proposes a data privacy protection device, which includes:

[0031] An intention acquisition module, used to acquire a purchase intention set published by a data element purchaser, wherein the purchase intention set includes at least one real intention;

[0032] A boundary determination module, used to determine an upper confidence limit and a lower confidence limit of the purchase intention set according to the number of real intentions included in the purchase intention set;

[0033] A model building module, for establishing a release intention-based attack model, an efficiency maximization attack model, and a purchase record inference attack model for the data element buyer based on the purchase intention set, the confidence upper bound, the confidence lower bound, the data element background knowledge, and the purchase record of the data element buyer;

[0034] The model building module is further used to build a uniform attack defense model based on publishing intention, an efficiency maximization attack defense model and a purchase record inference attack defense model based on publishing intention, the efficiency maximization attack model and the purchase record inference attack model with a confidence threshold as a constraint condition;

[0035] The privacy protection module is used to protect the privacy of the data element purchaser based on the uniform attack defense model based on the publishing intention, the efficiency maximization attack defense model and the purchase record inference attack defense model.

[0036] In addition, to achieve the above-mentioned purpose, the present application also proposes a data privacy protection device, which includes a memory, a processor, and a data privacy protection program stored in the memory and executable on the processor, and the data privacy protection program is configured to implement the data privacy protection method described above.

[0037] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, on which a data privacy protection program is stored, and when the data privacy protection program is executed by a processor, the data privacy protection method as described above is implemented.

[0038] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a data privacy protection program, and when the data privacy protection program is executed by a processor, it implements the data privacy protection method described above.

[0039] One or more technical solutions proposed in this application have at least the following technical effects:

[0040] In the present application, a method of obtaining a purchase intention set published by a data element purchaser is disclosed, wherein the purchase intention set includes at least one real intention, an upper confidence bound and a lower confidence bound of the purchase intention set are determined according to the number of real intentions included in the purchase intention set, and an attack model based on publication intention, an efficiency maximization attack model, and a purchase record inference attack model for the data element purchaser are established based on the purchase intention set, the upper confidence bound, the lower confidence bound, the background knowledge of the data element, and the purchase record of the data element purchaser. According to the attack model based on publication intention, the efficiency maximization attack model, and the purchase record inference attack model, a uniform attack defense model based on publication intention, an efficiency maximization attack defense model, and a purchase record inference attack defense model are constructed with a confidence threshold as a constraint condition. According to the uniform attack defense model based on publication intention, the efficiency maximization attack defense model, and the purchase record inference attack defense model, the privacy of the data element purchaser is protected. Since the present application takes into account various background knowledge that may be used by attackers, problem modeling is performed on the privacy protection problem of the data element purchaser in the data circulation transaction to solve the above problem, thereby ensuring the privacy security of the data element purchaser and reducing the risk of privacy leakage of the data element purchaser. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0042] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0043] Figure 1 This is a flowchart of the first embodiment of the data privacy protection method of this application;

[0044] Figure 2 This is a flow chart of the second embodiment of the data privacy protection method of this application;

[0045] Figure 3 This is a flowchart of the third embodiment of the data privacy protection method of this application;

[0046] Figure 4 This is a flow chart of a uniform attack / efficiency maximization attack defense algorithm based on publishing intention in an embodiment of a data privacy protection method of this application;

[0047] Figure 5 A schematic diagram of extended characteristic values ​​on the horizontal axis of an embodiment of the data privacy protection method of the present application;

[0048] Figure 6 This is a flow chart of a Monte Carlo simulation algorithm for an embodiment of a data privacy protection method of the present application;

[0049] Figure 7 This is a flow chart of a Markov chain Monte Carlo algorithm of an embodiment of a data privacy protection method of the present application;

[0050] Figure 8 This is a schematic diagram of the module structure of the data privacy protection device according to an embodiment of the present application;

[0051] Fig. 9 Schematic diagram of the device structure of the hardware operating environment involved in the data privacy protection method in the embodiment of the present application.

[0052] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0053] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0054] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0055] At present, data has become the fifth largest production factor after land, labor, capital, and technology, becoming a national basic strategic resource and the core engine for the development of the digital economy. Especially with the advent of the era of big models, large-scale, high-quality, and diversified data sets are needed to improve model effects and generalization capabilities, and the strategic position of data elements has been further highlighted.

[0056] Nowadays, applications and services based on big data widely rely on data from multiple sources. Whether in hot fields such as big data mining, data science or artificial intelligence, there is a strong demand for diversified data sources, especially for the secondary use of data. As an online platform, the data circulation market connects data providers and users, and promotes the discovery, exchange, sharing and integration of data resources.

[0057] In the data circulation market, privacy and its protection are particularly important because the privacy of all parties involved may be leaked during data transactions. In a broad sense, privacy protection refers to the ability of an individual or group to prevent their information from being identified or accessed without authorization. Data providers face privacy risks. For example, hospitals have valuable medical data, which may be needed by medical device companies. If hospitals can protect personal identity security while collecting and anonymizing patient treatment data, when providing these data in the data market, buyers may still obtain information about the success rate of treatment for specific diseases through the data, which may expose the privacy of the hospital. In addition, data transactions may also involve third-party privacy leakage, that is, the privacy leakage of the provider of the data trading platform. For example, a company that provides machine learning model building services to data buyers may face the risk of model theft, which also infringes on the company's privacy. However, current research and technology on privacy protection in data transactions mostly focus on protecting the privacy of data providers and data trading platforms, while the privacy protection of buyers is often overlooked. In fact, the buyer's identity, purchase location and time, purchased products, price and quantity and other information may be accidentally leaked during the transaction. Frequent customer information leakage incidents on e-commerce platforms highlight the urgency of protecting buyers' privacy.

[0058] Privacy protection has become a key technology area, involving a variety of innovative methods and patents. These technologies include privacy computing technologies, such as homomorphic encryption, multi-party secure computing, and federated learning, which can perform analysis and collaborative computing without exposing the original details of the data. Data anonymization and de-identification technologies, such as K-anonymization and L-diversity algorithms, achieve privacy protection by blurring personal information. Blockchain technology is also used to build a transparent and secure data exchange platform, combining smart contracts and zero-knowledge proofs to enhance privacy protection. In addition, data watermarking and traceability technologies ensure the transparency of data copyright and propagation paths, and hardware and software solutions, such as dedicated encryption processors and secure enclaves, improve the efficiency of privacy computing.

[0059] For example, the related technology proposes a privacy protection method and system for big data transactions, which realizes the effective protection of data privacy between buyers and sellers during the transaction process through firewall isolation, homomorphic encryption, data desensitization and blockchain notarization. The buyer obtains data samples through trial order requests for fusion experiments, and purchases big data after confirming the demand, while the seller notarizes the big data and encrypts the transmission through blockchain, and the server finally delivers the calculation results based on big data using federated learning technology.

[0060] Related technologies also propose a data circulation and privacy protection method based on blockchain, which involves calculating the asymmetric key of the data user, obtaining file data and file keywords, calculating the file data ciphertext and keyword index ciphertext, and determining access rights, so as to achieve the safe circulation of data and privacy protection. However, the use of asymmetric key encryption may introduce a high computing load and affect the efficiency of data circulation, especially in the case of large data volumes, the encryption and decryption process may become a bottleneck.

[0061] The relevant technology also proposes a method to provide safe and compliant personal data privacy protection, using differential privacy algorithms, chaotic asymmetric encryption algorithms and multi-source heterogeneous data governance technologies to achieve multi-layer encryption protection of data.

[0062] However, relevant data privacy protection schemes mainly focus on protecting the privacy of data providers and data trading platforms, and do not involve the privacy protection of data buyers, who are at risk of privacy leakage.

[0063] Therefore, in order to overcome the above-mentioned defects, the present application provides a solution, which includes: obtaining a purchase intention set published by a data element buyer, wherein the purchase intention set includes at least one real intention, determining the upper confidence limit and the lower confidence limit of the purchase intention set according to the number of real intentions included in the purchase intention set, establishing an attack model based on the release intention, an efficiency maximization attack model and a purchase record inference attack model for the data element buyer based on the purchase intention set, the upper confidence limit, the lower confidence limit, the background knowledge of the data element and the purchase record of the data element buyer, according to the attack model based on the release intention, the efficiency maximization attack model and the purchase record inference attack model, constructing a uniform attack defense model based on the release intention, an efficiency maximization attack defense model and a purchase record inference attack defense model with the confidence threshold as a constraint condition, and protecting the privacy of the data element buyer according to the uniform attack defense model based on the release intention, the efficiency maximization attack defense model and the purchase record inference attack defense model; because the present application takes into account various background knowledge that may be used by attackers, it models the problem of privacy protection of data element buyers in data circulation transactions to solve the above-mentioned problem, thereby ensuring the privacy security of data element buyers and reducing the risk of privacy leakage of data element buyers.

[0064] It should be noted that the executor of the embodiment of the present application may be a data privacy protection device with data processing, network communication and program running functions, such as a server, etc., or other electronic devices that can achieve the same or similar functions, and this embodiment does not impose any restrictions on this.

[0065] Based on this, the present application embodiment provides a data privacy protection method, referring to Figure 1 , Figure 1 This is a flowchart of the first embodiment of the data privacy protection method of this application.

[0066] In a first embodiment, the data privacy protection method includes:

[0067] Step S10: Obtain a purchase intention set published by a data element purchaser, wherein the purchase intention set includes at least one real intention.

[0068] It should be understood that data is a collection of information in a broad sense, and data elements are units extracted from data with specific value and meaning. This extraction process usually involves a series of operations such as data cleaning and processing, and it is necessary to consider the privacy leakage issues that may be caused by transactions. The purpose is to make data more in line with market demand and improve its usability and security. Suppose the buyer of data circulation transaction wants to Get data from, where D1,…,D n Represents the dimensions (also called attributes) of the data. Representing data space Without loss of generality, let’s assume that each dimension D i Data buyers want to obtain data space The data subset of the buyer, that is, the buyer’s true intention, that is, the data that the buyer wants to buy, can be expressed using the conjunction normal form V1∧…∧V n Indicates that When V i =D i When V i =ALL, that is, the data buyer does not have any i Specify any constraints.

[0069] In an effective data circulation market, it is assumed that data buyers make their true intentions clear in an honest manner, that is, all data records belonging to the true intentions are indeed meaningful to the buyers. For example, consider a data set representing customer transactions on an e-commerce platform, which has three dimensions: D1 is the product category (such as electronic products, clothing, books, etc.), D2 is the payment method (such as bank card, credit card, e-wallet, etc.), and D3 is the shipping address (such as Beijing, Shanghai, Guangdong, etc.). Assume that there is an online electronic retailer as a buyer of data element circulation transactions who is interested in purchasing a specific subset of data to help it analyze targeted marketing activities. The buyer's true purchase intention is (V1 = {electronic products}) ∧ (V2 = ALL) ∧ (V3 = {Beijing}). If the buyer's true purchase intention is directly disclosed, the potential target of the data element buyer can be easily obtained by competitors, that is, its strategic focus on selling electronic products to customers in Beijing is exposed. This information may put the data element buyer under greater competitive pressure, because competitors can use this information to improve their marketing strategies, intensify competition in the market, and formulate strategies to attract customers in the same geographical area. Therefore, data element buyers need privacy protection so that attackers cannot determine the buyer's true intentions based on the information observed from the buyer.

[0070] Consider different assumptions about observable information and the corresponding attack models. In order to quantify the attacker’s true intention to invade the privacy of data element buyers, For a data record in the data element, it is necessary to calculate the attacker's confidence in whether the record belongs to the real intention of the data element buyer. This application summarizes three types of attacks: attacks based on publishing intentions, efficiency maximization attacks, and purchase record inference attacks, and models are built around these three types of attacks.

[0071] It is understandable that the first attack model is based on the purchase intention published by the data element buyer. In order to protect the real purchase intention of the data element buyer and maintain his privacy, the data element buyer can publish a U1∧…∧U n A superset of the true intent, i.e. (i.e., the set of purchase intentions published by data element buyers) so that potential data element providers can understand the interests of data buyers. And there is the following assumption: in an attack based on published purchase intentions, the attacker can only observe the purchase intentions published by data element buyers and does not have control over the data space Related information.

[0072] Since the published purchase intent is the only information the attacker has about the buyer’s intent, If the data record in the published intent is not within the scope of the published intent, the attacker can know that the buyer is not interested in the data record. For the data record included in the published intent, the attacker's confidence in whether the data record is included in the buyer's true intent can be analyzed as follows:

[0073] Since the attacker does not have any information about the data space, such as probability density, the best assumption is Each possible point in has the same probability of appearing, that is, The data distribution in is uniform. Therefore, in the proposal, the attack based on the intention of publishing is called the uniform attack based on the intention of publishing.

[0074] Step S20: determining an upper confidence limit and a lower confidence limit of the purchase intention set according to the number of real intentions included in the purchase intention set.

[0075] It should be understood that since the attacker does not have any information about the data space, such as probability density, the best assumption is Each possible point in has the same probability of appearing, that is, The data distribution in is uniform. Therefore, in the proposal, the attack based on the intention of publishing is called the uniform attack based on the intention of publishing.

[0076] Since the real purchase intention is a subset of the published purchase intention, and the minimum size of the real purchase intention is 1, that is, the real purchase intention contains at least one specific value in each published intention. Therefore, the lower bound of the confidence is: And this is known information to the data element buyer and any attacker. Similarly, the upper bound of the confidence is: This information is only known to the purchaser of the data element, not to the attacker.

[0077] Step S30: Based on the purchase intention set, the confidence upper limit, the confidence lower limit, the data element background knowledge and the purchase record of the data element purchaser, establish a publishing intention-based attack model, an efficiency maximization attack model and a purchase record inference attack model for the data element purchaser.

[0078] For ease of understanding, the following examples are given, but are not intended to limit the present application. The steps of establishing an attack model based on publishing intent include:

[0079] Since the published purchase intent is the only information the attacker has about the buyer’s intent, If the data record in the published intent is not within the scope of the published intent, the attacker can know that the buyer is not interested in the data record. For the data record included in the published intent, the attacker's confidence in whether the data record is included in the buyer's true intent can be analyzed as follows:

[0080] Since the attacker does not have any information about the data space, such as probability density, the best assumption is Each possible point in has the same probability of appearing, that is, The data distribution in is uniform. Therefore, in the proposal, the attack based on the intention of publishing is called the uniform attack based on the intention of publishing.

[0081] Since the real purchase intention is a subset of the published purchase intention, and the minimum size of the real purchase intention is 1, that is, the real purchase intention contains at least one specific value in each published intention. Therefore, the lower bound of the confidence is: And this is known information to the data element buyer and any attacker. Similarly, the upper bound of the confidence is: This information is only known to the data element buyer, not to the attacker. Therefore, we can get: in, It indicates that when there is a data record x in the intention published by the data element buyer, the attacker's confidence that the data record x is the true intention of the data element buyer through a uniform attack based on the publication intention.

[0082] For ease of understanding, the following examples are given, but are not intended to limit the present application. The steps of establishing an efficiency maximization attack model include:

[0083] In addition to the observable publishing intentions, the attacker may also have some background knowledge about the data. The attacker can use this background knowledge to increase the confidence in whether the data record belongs to the real intention of the data buyer. The core idea of ​​efficiency maximization attack is: in order to achieve economic efficiency, data element buyers tend to maximize the efficiency of published purchase intentions. That is, a large part of the published purchase intentions are within the real intentions of the data element buyers. Therefore, this type of attack is called efficiency maximization attack. And there is the following assumption: in the efficiency maximization attack, the attacker can observe the purchase intentions published by the data element buyers, and also master the data space Data distribution information in

[0084] If the attacker knows the data distribution in the subspace of the purchase intentions published by the data element buyers, then given a data record that matches the published intention, the attacker's upper bound of confidence that the record belongs to the buyer's intention is proportional to the probability of the record x. That is, in, Representation based on data space The probability of the data record x calculated from the data distribution in .

[0085] For ease of understanding, the following examples are given, but are not intended to limit the present application. The steps of establishing a purchase record inference attack model include:

[0086] In some cases, attackers will obtain the purchase records of data element buyers and try to infer the intentions of data element buyers. In order to protect their privacy, data element buyers may purchase more data records than their true intentions. Suppose a data element buyer purchases a set of data records It contains data records of real purchase intentions and some redundant data records used for disguise to protect privacy. The attacker may try to infer whether the data element buyer is interested in a given data record based on such purchase records. This type of attack is a purchased record inference attack. And there is the following assumption: In a purchase record inference attack, the attacker not only has the data record set purchased by the data element buyer, but also has the relevant data space Background knowledge of the data distribution in

[0087] Given a data record, an attacker can infer whether the record is of interest to the data buyer through two steps. In the first step, the attacker can use the data record set observed by the data buyer to infer a pseudo-release intention of the data element buyer. That is, the attacker can try to find a minimum intention So that each purchased data record is in this intent, it can be expressed as The solution to this problem can be written as: Among them, x[i] is the i-th dimension feature of x.

[0088] Given data space If a data record in the pseudo-publication intent is not in the pseudo-publication intent, the attacker knows that the record is not of interest to the data element buyer. For the records in the pseudo-publication intent, the attacker’s confidence in whether the record is in the buyer’s true intent can be analyzed based on the distribution of records purchased in the pseudo-publication intent and the attacker’s background knowledge.

[0089] The attacker can infer the observed data distribution in the purchased data record set For a data record x in the pseudo-publishing intention, the probability of x appearing in the purchase set can be estimated as Among them, h(x) represents The data in the table records the frequency of occurrence of x, while Representing a collection The number of elements in .

[0090] In the simplest case, the attacker can only observe the data records purchased by the data element buyer and has no control over the data space. Then, given a record that matches the fake publication intention, the attacker's confidence that the record belongs to the real intention of the data element buyer is bounded by the record in The probability of occurrence is proportional to

[0091] If the attacker gains control of the data space Data distribution in By comparing and the observed distribution The attacker can infer the probability that the purchased data record x belongs to the true intention based on the efficiency maximization hypothesis. In the observed distribution The more frequently x appears, the more The smaller the probability of x appearing in , the more likely it is that x is the true intention.

[0092] Quantitatively speaking, for a purchase data record x, the p-value, i.e. the proportion in the sample set, can be used to measure the probability that a purchase data record x is in a given distribution. In the case of the observed distribution The probability of frequency x appearing in . First, according to Calculating pseudo-publication intentions The probability of x appearing in is as follows

[0093] if And the p-value is small, then the data record x is likely to belong to the true intention. In order to calculate its p-value, Monte Carlo simulation and permutation test can be used to perform model-free hypothesis testing. Specifically, according to Extract from pseudo-publishing intent Group records, the size of each group record is the same as the set purchased by the buyer Same (denoted as ).for Satisfied For each data record x, calculate The p-value (i.e., the proportion of the sample set in which the probability of record x occurring is at least Based on this, the attacker's upper confidence bound on the data element buyer's intention containing the record can be written as:

[0094] According to the above description of the three types of attacks:

[0095] 1) The attack model based on publishing intention is used to represent the scenario where the attacker attacks only based on the purchase intention published by the data buyer. Based on the fact that the data buyer publishes a superset of the real intention and the attacker has no information about the data space, assuming that the data distribution in the data space is uniform, the attacker obtains the upper and lower bounds of the confidence of whether each data record is included in the real intention of the buyer by calculating the ratio of the maximum and minimum values ​​of the number of intentions published by the data buyer to the number of intentions published.

[0096] (2) The efficiency maximization attack model represents a scenario where the attacker not only attacks based on the purchase intention published by the data buyer, but also combines background knowledge and data distribution information. The basis is that the data element buyer tends to maximize the efficiency of publishing purchase intentions, and the attacker has data distribution information. The attacker's confidence that the record is the intention of the data buyer is proportional to the probability of the record appearing.

[0097] (3) The purchase record inference attack model is used to represent a scenario where the attacker not only has the probability distribution of the data space, but also has the actual purchase records of the data purchaser, and attempts to infer the buyer's true purchase intention through these records. The basis is that the upper bound of the attacker's confidence that the record belongs to the true intention of the data element purchaser is proportional to the probability of the record appearing in the purchase record. If the attacker observes that a certain data item appears more frequently in the distribution of published data and the probability of appearing in the data space is smaller, then it is more likely to belong to the true intention.

[0098] Step S40: Based on the attack model based on publishing intention, the efficiency maximization attack model and the purchase record inference attack model, a uniform attack defense model based on publishing intention, an efficiency maximization attack defense model and a purchase record inference attack defense model are constructed with a confidence threshold as a constraint condition.

[0099] It should be understood that buyer privacy attack modeling is a scenario simulation to understand the attacker's behavior and capabilities, while buyer privacy protection task modeling is based on these attack scenarios to design effective defense solutions to ensure the privacy security of data buyers. Through attack modeling, we can understand potential threats, and then formulate protection strategies, and ultimately achieve privacy security.

[0100] Given a buyer's true intention and a buyer-specified maximum confidence threshold λ>0, the data element buyer either publishes a purchase intention in the publication intention uniform attack and efficiency maximization attack models, or purchases a larger range of data records in the purchase record inference attack, so that any attacker's confidence in the records in the published intention or the records purchased by the data element buyer that belong to the buyer's true intention is no higher than λ. If satisfied, the published intention is called a λ-privacy-preserving publication intention, and the set of data records to be purchased is called a λ-privacy-preserving purchase record set.

[0101] Since data buyers need to pay for the set of data records they intend to publish or purchase, the scope of the data records they intend to publish and purchase should be minimized for economic reasons. The data buyer privacy protection problem is to calculate the minimum set of λ-privacy-preserving records of the intention to publish or purchase.

[0102] For ease of understanding, the following examples are given, but are not intended to limit the present application. The steps of constructing a uniform attack defense model based on release intent include:

[0103] For uniform attacks based on publishing intent, ensure that: The confidence upper bound expressed in does not exceed λ, that is, Right now Therefore, the privacy-preserving defense against uniform attacks based on publishing intention can be modeled as an optimization problem:

[0104] For ease of understanding, the following examples are given, but are not intended to limit the present application. The steps of constructing an efficiency maximization attack defense model include:

[0105] In the efficiency maximization attack scenario, it is necessary to ensure that the confidence upper bound based on the attacker's background knowledge of data distribution is not higher than the confidence threshold λ of all data records that belong to the true intention. According to the efficiency maximization attack model, the attacker's confidence upper bound that the record belongs to the buyer's intention is proportional to the probability of the record x. For any x∈V1∧…∧V n , both have It can be seen that That is to say

[0106] Therefore, the defense against the efficiency-maximizing attack can be modeled as an optimization problem:

[0107]

[0108] For ease of understanding, the following examples are given, but are not intended to limit the present application. The steps of constructing a purchase record inference attack defense model include:

[0109] In the purchase record inference attack scenario, it is also necessary to ensure that the confidence upper bound based on the attacker's background knowledge of data distribution is not higher than the confidence threshold λ of all data records belonging to the true intention. According to the purchase record inference attack model, the attacker's confidence upper bound on the data element buyer's intention to include the record can be obtained for any x∈V1∧…∧V n , both have Right now

[0110] Therefore, it is necessary to ensure The p-value is greater than or equal to 1-λ. Taking utility as the main consideration, where buyers seek to maximize the ratio of the number of records in the true intent to the number of records in the posted intent, the defense against purchase record inference attacks can be modeled as an optimization problem:

[0111] Step S50: The privacy of the data element purchaser is protected according to the uniform attack defense model based on publishing intention, the efficiency maximization attack defense model and the purchase record inference attack defense model.

[0112] It can be understood that by solving the uniform attack defense model based on publishing intention, the efficiency maximization attack defense model, and the purchase record inference attack defense model constructed in the above steps, the privacy of data element buyers can be ultimately protected.

[0113] This embodiment takes into account various background knowledge that may be exploited by attackers, and models the privacy protection issue of data element buyers in data circulation transactions to solve the above problems, thereby ensuring the privacy security of data element buyers and further reducing the risk of privacy leakage of data element buyers.

[0114] Reference Figure 2 , Figure 2 This is a flow chart of the second embodiment of the data privacy protection method of this application, based on the above Figure 1 The first embodiment shown proposes a second embodiment of the data privacy protection method of the present application.

[0115] In the second embodiment, the step S50 includes:

[0116] Step S501: Expand the purchase intention set to minimize the optimization problem in the uniform attack and defense model based on release intention or the efficiency maximization attack and defense model.

[0117] It should be understood that, for the above three defense task models, this embodiment proposes corresponding algorithms to solve the privacy protection problem of data buyers, namely: a uniform attack / efficiency maximization attack defense algorithm based on publishing intention, and a purchase record inference attack defense algorithm.

[0118] It is understandable that for the uniform attack and efficiency maximization attack algorithms based on publishing intention, an "expansion" strategy is adopted to find the best neighborhood feature value and add it to the publishing intention by expanding the set of true purchase intention feature values ​​to minimize the optimization problem obtained by the privacy protection constraint conversion. This allows the published intention to minimize the scope of the published intention and purchased data records as much as possible while satisfying the privacy protection constraint, thereby reducing the possibility of attackers inferring the buyer's true intention through the published intention, and protecting the buyer's privacy.

[0119] Step S502: Sampling is performed based on the total number of data records that the data element purchaser is to purchase, to obtain a purchase data set corresponding to the data element purchaser, so as to ensure that the purchase data set satisfies the privacy protection constraints corresponding to the purchase record inference attack defense model.

[0120] It should be understood that, for the purchase record inference attack algorithm, a "sampling" strategy is adopted. The Monte Carlo simulation algorithm uses random sampling to generate data sets and merge and screen data sets that meet privacy protection constraints. The Markov chain Monte Carlo algorithm calculates the state transition probability of the data set by iteratively selecting data record addition or deletion operations. According to the Monte Carlo simulation algorithm sampling or the Markov chain Monte Carlo algorithm iteration, the privacy security of the data element purchaser can be guaranteed because the final purchased data set is ensured within the privacy constraints.

[0121] In response to the above three defense task models, this embodiment proposes corresponding algorithms for solving the privacy protection problem of data buyers, namely, a uniform attack / efficiency maximization attack defense algorithm based on publishing intention, and a purchase record inference attack defense algorithm, thereby reducing the possibility of attackers inferring the buyer's true intentions through publishing intentions and ensuring the privacy security of data element buyers.

[0122] Reference Figure 3 , Figure 3 This is a flow chart of the third embodiment of the data privacy protection method of this application. Figure 1 The first embodiment shown proposes the third embodiment of the data privacy protection method of the present application.

[0123] In the third embodiment, the step S501 includes:

[0124] Step S5011: setting the purchase intention set to be the same as the real intention, and traversing each dimension in the data space in each iteration.

[0125] It should be understood that given the data space and data element buyer’s real purchase intention V1∧…∧V n , we can continuously expand the feature value set V in the real purchase intention i , until a release intention PI (i.e., purchase intention set) is obtained that can fully disguise and protect the true intention, that is, it satisfies the constraints of the optimization problem in the uniform attack and defense model based on release intention or the efficiency maximization attack and defense model.

[0126] Step S5012: Search for optional neighborhood feature values ​​on the traversed dimension, and select the best neighborhood feature value from the optional neighborhood feature values.

[0127] It should be understood that searching for optional neighborhood feature values ​​on the traversal dimension and selecting the best neighborhood feature value from the optional neighborhood feature values ​​may include calculating the cumulative distance of the feature values ​​on the traversal dimension and selecting the feature value with the shortest cumulative distance as the optional neighborhood feature value; calculating the number of additional records, the incremental impact of the privacy protection constraint, and the effectiveness score of the optional neighborhood feature value; and selecting the best neighborhood feature value from the optional neighborhood feature values ​​based on the number of additional records, the incremental impact of the privacy protection constraint, and the effectiveness score.

[0128] For ease of understanding, refer to Figure 4 This invention is provided for illustration, but is not intended to limit the present application. Figure 4 This is a flow chart of a uniform attack / efficiency maximization attack defense algorithm based on publishing intention in an embodiment of the data privacy protection method of this application. Figure 4 In the above, the uniform attack / efficiency maximization attack defense algorithm based on publishing intention can be divided into the following three steps:

[0129] Step 1: Search for optional neighborhood feature values. Figure 5 As shown, Figure 5 This is a schematic diagram of extended characteristic values ​​on the horizontal axis of an embodiment of the data privacy protection method of this application. Figure 5 The published intent is represented by the blue area, in this example, That is, the data space has two dimensions. The horizontal D1 represents the education level, and the vertical D2 represents the job category. The characteristic value of the release intention in dimension D1 is U1 = {"Master", "PhD"}, and the possible characteristic values ​​of U1 can be expanded to add "College" and "Bachelor" and so on.

[0130] Step 1.1: Calculate the cumulative distance of eigenvalues. First, for each expanded eigenvalue, calculate the cumulative distance between the eigenvalue and all eigenvalues ​​in U1. Eigenvalue u∈D i \U i The cumulative distance is defined as Assuming that index i=1, 2, 3, 4 are assigned to "college", "bachelor", "master" and "doctor" respectively, the cumulative distance between "college" and all features in U1 is 5, and the cumulative distance of "bachelor" relative to U1 is 3.

[0131] Step 1.2: Select the feature value with the shortest cumulative distance. Then, select the feature value with the shortest cumulative distance as the nearest neighbor feature value on D1, that is, "bachelor". On dimension D2, there is no semantic distance between feature values, that is, every pair of feature values ​​on D2 has the same distance. Therefore, for feature U2 = {"civil servant"} in the publishing intent on dimension D2, all remaining features "unemployed", "private enterprise" and "individual" are considered as potential extended feature values.

[0132] After iterating each dimension, four possible expanded neighborhood eigenvalues ​​are obtained: “Bachelor” and “Unemployed” on the D1 dimension, and “Private enterprise” and “Individual” on the D2 dimension.

[0133] Step 2: Determine the optimal neighborhood feature value. In order to determine the optimal feature value, this application designs a scoring function to comprehensively measure the effectiveness of adding the feature value to the release intention. The scoring considers two factors: one is the number of additional data records added to the data element buyer's release intention after adding the feature value, and the other is the incremental impact of adding the feature value on satisfying the privacy protection constraint.

[0134] Step 2.1: Calculate the number of additional records to add neighborhood feature values. Given a dimension D to be added i The potential eigenvalue u on ij , first consider the number of additional data records added to the data element buyer's intention after adding this feature value add(u ij ). Given the eigenvalue u ij and the current release intention U1∧…∧U i …∧U n , the number of eigenvalues ​​added is

[0135]

[0136] Step 2.2: Calculate the incremental impact of the privacy protection constraint of the added neighborhood eigenvalue. Secondly, in order to measure the impact of adding this eigenvalue on satisfying the optimization problem constraints in the uniform attack defense model based on publishing intention or the efficiency maximization attack defense model, the increment Δ(u ij 0. For a uniform attack based on publishing intention, Δ(u ij )=|U1∧…∧U i-1 ∧{u ij}∧U i+1 …∧Un |. For the efficiency maximization attack, the increment of the constraint after adding the eigenvalue is

[0137] Step 2.3: Calculate the effectiveness score of adding neighborhood feature values. For each neighborhood feature value, add the number of data records added (u ij ) and constraint influence increment Δ(u ij ) is normalized using the minimum-maximum method. Define the formula to calculate the added neighborhood eigenvalue u ij The score is Among them, α is the weight of the evaluation factor, and u ij The normalized added quantity and constraint impact increment under the constraints. Represents the maximum number of added values ​​for all possible neighborhood feature values.

[0138] Step 3: Select neighborhood feature values ​​based on effectiveness scores. The buyer of the final data element selects the feature values ​​included in the published intent for purchase based on the scores of all neighborhood feature values ​​to achieve privacy protection.

[0139] Step S5013: After traversing each dimension, the optimal neighborhood feature value is added to the purchase intention set to minimize the optimization problem in the uniform attack and defense model based on release intention or the efficiency maximization attack and defense model.

[0140] For ease of understanding, the following examples are given, but are not intended to limit the present application. The core idea of ​​the uniform attack / efficiency maximization attack defense algorithm based on release intention is to convert the initial release intention PI = U1∧…∧U n is set to be the same as the true intention, that is, for 1≤i≤n, U i =V i In each iteration, the data space is traversed For each dimension in and its corresponding feature value set U in the release intention i , select D i \U i After traversing each dimension, the nearest neighbor eigenvalue is added to the release intention PI to minimize the optimization problem in the uniform attack defense model based on release intention or the efficiency maximization attack defense model.

[0141] This embodiment adopts an "expansion" strategy for the uniform attack and efficiency maximization attack algorithms based on publishing intentions. By expanding the set of feature values ​​of real purchase intentions, the optimal neighborhood feature values ​​are found and added to the publishing intentions to minimize the optimization problem obtained by the privacy protection constraint conversion. This allows the published intentions to minimize the scope of the published intentions and purchased data records as much as possible while satisfying the privacy protection constraints, thereby reducing the possibility of attackers inferring the buyer's true intentions through the published intentions, thereby protecting the buyer's privacy.

[0142] In the third embodiment, the step S502 includes:

[0143] Step S5021: Generate a data set by random sampling based on the total number of data records that the data element purchaser is going to purchase, and eliminate the data set that does not meet the privacy protection constraints.

[0144] It should be understood that the optimization problem derived from the attack defense model inferred from the purchase records is to make the function The value of p for any x∈V1∧…∧V n are greater than or equal to 1-λ. First, we need to determine the buyer's release intention (which is a false release intention in the attacker's view). ). Therefore, given the buyer’s true intention, the proposed extended method is firstly used to calculate the published intention.

[0145] Then, given the total number of data records q that the data element buyer wants to purchase and the published purchase intention PI = U1∧…∧U n , q records need to be assigned to U1∧…∧U n In this regard, the present application proposes two algorithms: (1) Monte Carlo simulation algorithm; (2) Markov chain Monte Carlo algorithm.

[0146] Step S5022: Calculate the utilization rates of the remaining data sets, and select the remaining data set with the highest utilization rate as the purchase data set.

[0147] For ease of understanding, refer to Figure 6 This invention is provided for illustration, but is not intended to limit the present application. Figure 6 This is a flow chart of the Monte Carlo simulation algorithm of an embodiment of the data privacy protection method of this application. Figure 6 In the Monte Carlo simulation algorithm, there are four specific steps:

[0148] Step 1: Generate a data set by random sampling. Generate Z data sets Each dataset contains q records sampled from published intents.

[0149] Step 2: Eliminate the data sets that do not meet the privacy protection constraints. The p-value is less than 1-λ.

[0150] Step 3: Calculate the utilization of the remaining data sets. Calculate the utilization rate for each remaining data set after deletion, that is, the ratio of the number of data records in the real purchase intention to the number of data records in the release intention. Given a remaining data set The utilization rate is

[0151] Step 4: Select the dataset with the highest utilization rate as the purchased dataset. Taking utilization rate as the primary consideration, the data element buyer strives to maximize the ratio of the number of records of real intention and release intention, so the dataset with the highest utilization rate is finally selected as the purchased dataset.

[0152] This embodiment adopts a "sampling" strategy for the purchase record inference attack algorithm. The Monte Carlo simulation algorithm uses random sampling to generate data sets and merge and filter data sets that meet the privacy protection constraints. The sampling iteration of the Monte Carlo simulation algorithm ensures that the final purchased data set is within the privacy constraints, thereby protecting the privacy security of the data element purchaser.

[0153] In the third embodiment, the step S502 further includes:

[0154] Step S5021': Initialize the purchase data set through distribution sampling according to the total number of data records to be purchased by the data element purchaser.

[0155] Step S5022': perform addition and deletion processing on the initialized data set, and check whether the processed data set satisfies the privacy protection constraint.

[0156] Step S5023': If the privacy protection constraint is satisfied, then the acceptance probability of the processed data set is calculated.

[0157] Step S5024': determine whether to select the processed data set as the purchase data set according to the acceptance probability.

[0158] For ease of understanding, refer to Figure 7 This invention is provided for illustration, but is not intended to limit the present application. Figure 7 This is a flow chart of the Markov chain Monte Carlo algorithm of an embodiment of the data privacy protection method of this application. Figure 7 In the Markov Chain Monte Carlo algorithm, there are four specific steps:

[0159] Step 1: Initialize the purchase data set based on distribution sampling. First, infer the pseudo-publication intention in the attack defense model based on the purchase records. The probability of x appearing in the data is calculated based on the data distribution. Extract q data records to initialize the purchased data set

[0160] Step 2: Select basic processing operations for the data set. Given an initialized record set It contains the records that the data element buyer really wants (the buyer's true intention) and some disguised data records used to hide privacy. Four basic processing operations are defined for this data set:

[0161] a) Delete a disguised data record and add a data record that the buyer actually intended;

[0162] b) deleting a disguised data record and adding another disguised data record;

[0163] c) Delete a data record that is the real intention of the purchaser and add a fake data record;

[0164] d) Delete a data record that is actually intended by the buyer, and add another data record that is actually intended by the buyer.

[0165] By selecting different disguised data records or data records in the buyer's true intention, there are multiple operation choices, thus forming a Markov chain state space. For each iteration, an operation is randomly selected from a, b, c, d and the current data set is converted according to the operation. Transformed to

[0166] Step 3: Check whether the privacy protection constraints are met and calculate the acceptance probability. Whether the privacy protection constraint is satisfied, if so, calculate The acceptance probability Where η(·) is the utilization function in step 3 of the Monte Carlo simulation algorithm.

[0167] Step 4: Repeat steps 2 and 3 until the privacy constraint is satisfied. In each iteration, continue to select the next operation and update the purchased data set until the privacy constraint in the purchase record inference attack defense model is no longer satisfied. The p value is less than 1-λ. And in the iterative process, when the privacy threshold λ is equal to the attacker's record x∈V1∧…∧V n The Markov Chain Monte Carlo algorithm terminates when the minimum difference between the current confidences of is less than a certain threshold ∈.

[0168] Step 5: Decide whether to select as the purchase data set based on the acceptance probability. Transfer to The probability of As a purchased dataset.

[0169] Based on Monte Carlo simulation algorithm sampling or Markov Chain Monte Carlo algorithm iteration, the purchased data set is finally selected within the privacy constraints, thereby ensuring the privacy security of the buyer of the data elements.

[0170] This embodiment adopts a "sampling" strategy for the purchase record inference attack algorithm. The Markov Chain Monte Carlo algorithm calculates the state transition probability of the data set by iteratively selecting data record addition or deletion operations. According to the iteration of the Markov Chain Monte Carlo algorithm, the final purchased data set is ensured to be within the privacy constraint range, thereby protecting the privacy security of the data element purchaser.

[0171] Compared with the related art, this application has the following advantages:

[0172] 1. Modeling the privacy attack on data element circulation transactions to obtain three attack models, namely, the attack based on publishing intention, the attack based on efficiency maximization, and the attack based on purchase record inference, and modeling the privacy protection task of data element circulation transactions to obtain three defense task models. Two privacy protection defense algorithms are constructed based on the three attack models and the corresponding three defense task models, namely, the uniform attack / efficiency maximization attack defense algorithm based on publishing intention and the purchase record inference attack defense algorithm, to achieve privacy protection for data buyers;

[0173] 2. For modeling of three of the attack models, the specific method is to define the data space, data dimension, data probability distribution, etc., and calculate the upper and lower bounds of the attacker's confidence in whether the data record is contained in the buyer's true intention.

[0174] 3. For modeling of three defense task models, the specific method is to give a buyer's true intention and a maximum confidence threshold λ specified by the buyer. The data buyer privacy protection problem is to calculate the minimum λ-privacy protection release intention or purchase record set, and according to the attacker's confidence upper bound of whether the data record is included in the buyer's true intention does not exceed λ, the problem is converted into an optimization problem that satisfies the privacy protection constraints.

[0175] 4. For two of the privacy protection defense algorithms, the specific methods are the uniform attack / efficiency maximization attack defense algorithm based on the release intention. By defining the number of data records added and the incremental impact on the privacy constraint, the score of the neighborhood feature value of the added real purchase intention feature value is calculated, and the data element buyer chooses to purchase the feature value included in the released intention. The purchase record inference attack defense algorithm includes two optional algorithms, the Monte Carlo simulation algorithm and the Markov chain Monte Carlo algorithm. They respectively calculate the utilization rate under the privacy protection constraint conditions through sampling, and iteratively select the data record addition or deletion operation to calculate the state transition probability of the data set, and finally select the purchased data set within the privacy constraint range, thereby ensuring the privacy security of the data element buyer.

[0176] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the data privacy protection method of the present application. More simple transformations based on this technical concept are all within the protection scope of the present application.

[0177] This application also provides a data privacy protection device, please refer to Figure 8 , the data privacy protection device comprises:

[0178] The intention acquisition module 10 is used to acquire a purchase intention set published by a data element purchaser, wherein the purchase intention set includes at least one real intention;

[0179] A boundary determination module 20, configured to determine an upper confidence limit and a lower confidence limit of the purchase intention set according to the number of real intentions included in the purchase intention set;

[0180] A model building module 30, for establishing a release intention-based attack model, an efficiency maximization attack model, and a purchase record inference attack model for the data element buyer based on the purchase intention set, the confidence upper bound, the confidence lower bound, the data element background knowledge, and the purchase record of the data element buyer;

[0181] The model building module 30 is further used to build a uniform attack defense model based on publishing intention, an efficiency maximization attack defense model and a purchase record inference attack defense model based on publishing intention, the efficiency maximization attack model and the purchase record inference attack model with a confidence threshold as a constraint condition;

[0182] The privacy protection module 40 is used to protect the privacy of the data element purchaser according to the uniform attack defense model based on publishing intention, the efficiency maximization attack defense model and the purchase record inference attack defense model.

[0183] The data privacy protection device provided by the present application adopts the data privacy protection method in the above-mentioned embodiment, which can solve the technical problem that the existing data privacy protection scheme mainly focuses on protecting the privacy of data providers and data trading platforms, but does not involve the privacy protection of data buyers, and data buyers are at risk of privacy leakage. Compared with the prior art, the beneficial effects of the data privacy protection device provided by the present application are the same as the beneficial effects of the data privacy protection method provided by the above-mentioned embodiment, and the other technical features in the data privacy protection device are the same as the features disclosed in the above-mentioned embodiment method, which will not be repeated here.

[0184] The present application provides a data privacy protection device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the data privacy protection method in the above-mentioned embodiment one.

[0185] Reference below Fig. 9 , which shows a schematic diagram of the structure of a data privacy protection device suitable for implementing the embodiment of the present application. The data privacy protection device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Fig. 9 The data privacy protection device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0186] like Fig. 9As shown, the data privacy protection device may include a processing device 1001 (e.g., a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to the program stored in the ROM (Read Only Memory) 1002 or the program loaded from the storage device 1003 to the RAM (Random Access Memory) 1004. In the RAM 1004, various programs and data required for the operation of the data privacy protection device are also stored. The processing device 1001, the ROM 1002, and the RAM 1004 are connected to each other through a bus 1005. The input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, an LCD (Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the data privacy protection device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a data privacy protection device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or have alternatively.

[0187] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0188] The data privacy protection device provided by the present application adopts the data privacy protection method in the above embodiment, which can solve the technical problem that the existing data privacy protection scheme mainly focuses on protecting the privacy of data providers and data trading platforms, but does not involve the privacy protection of data buyers, and data buyers are at risk of privacy leakage. Compared with the prior art, the beneficial effects of the data privacy protection device provided by the present application are the same as the beneficial effects of the data privacy protection method provided by the above embodiment, and the other technical features in the data privacy protection device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0189] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0190] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0191] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the data privacy protection method in the above-mentioned embodiment.

[0192] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, RAM (Random Access Memory), ROM (Read Only Memory), EPROM (Erasable Programmable Read Only Memory) or flash memory, optical fiber, CD-ROM (CD-Read Only Memory, portable compact disk read-only memory), optical storage device, magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination of the above.

[0193] The above-mentioned computer-readable storage medium may be included in the data privacy protection device; or it may exist independently without being assembled into the data privacy protection device.

[0194] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the data privacy protection device, the data privacy protection device executes the data privacy protection method in the above embodiment.

[0195] The computer program code for performing the operation of the present application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on the remote computer, or completely on the remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a LAN (Local Area Network) or a WAN (Wide Area Network), or it can be connected to an external computer (e.g., using an Internet service provider to connect through the Internet).

[0196] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0197] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0198] The readable storage medium provided in this application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned data privacy protection method, and can solve the technical problem that the existing data privacy protection scheme mainly focuses on protecting the privacy of data providers and data trading platforms, but does not involve the privacy protection of data buyers, and data buyers are at risk of privacy leakage. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the data privacy protection method provided in the above-mentioned embodiment, which will not be repeated here.

[0199] The present application also provides a computer program product, including a computer program, which implements the data privacy protection method as described above when executed by a processor.

[0200] The computer program product provided by this application can solve the technical problem that the existing data privacy protection scheme mainly focuses on protecting the privacy of data providers and data trading platforms, but does not involve the privacy protection of data buyers, and data buyers are at risk of privacy leakage. Compared with the prior art, the beneficial effects of the computer program product provided by this application are the same as the beneficial effects of the data privacy protection method provided by the above embodiment, which will not be repeated here.

[0201] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A data privacy protection method, characterized in that: The data privacy protection method comprises: Acquire a purchase intention set published by a data element buyer, wherein the purchase intention set includes at least one real intention; Determining an upper confidence limit and a lower confidence limit of the purchase intention set according to the number of real intentions included in the purchase intention set; Establishing a release intention-based attack model, an efficiency maximization attack model, and a purchase record inference attack model for the data element buyer based on the purchase intention set, the confidence upper bound, the confidence lower bound, the data element background knowledge, and the purchase record of the data element buyer; According to the attack model based on publishing intention, the attack model for maximizing efficiency and the attack model for inferring purchase records, a uniform attack defense model based on publishing intention, an attack defense model for maximizing efficiency and a purchase record inferring attack defense model are constructed with confidence threshold as a constraint condition; The privacy of the data element purchaser is protected according to the uniform attack defense model based on publishing intention, the efficiency maximization attack defense model and the purchase record inference attack defense model.

2. The data privacy protection method according to claim 1, characterized in that: The privacy protection of the data element purchaser according to the uniform attack defense model based on publishing intention, the efficiency maximization attack defense model and the purchase record inference attack defense model includes: Expanding the purchase intention set to minimize the optimization problem in the uniform attack and defense model based on release intention or the efficiency maximization attack and defense model; Sampling is performed based on the total number of data records that the data element purchaser is to purchase, and a purchase data set corresponding to the data element purchaser is obtained to ensure that the purchase data set meets the privacy protection constraints corresponding to the purchase record inference attack defense model.

3. The data privacy protection method according to claim 2, characterized in that: The step of expanding the purchase intention set to minimize the optimization problem in the uniform attack and defense model based on release intention or the efficiency maximization attack and defense model includes: The purchase intention set is set to be the same as the real intention, and in each iteration, each dimension in the data space is traversed; Searching for optional neighborhood feature values ​​on the traversed dimension, and selecting the best neighborhood feature value from the optional neighborhood feature values; After traversing each dimension, the best neighborhood feature value is added to the purchase intention set to minimize the optimization problem in the uniform attack and defense model based on release intention or the efficiency maximization attack and defense model.

4. The data privacy protection method according to claim 3, characterized in that: The step of searching for optional neighborhood feature values ​​on the traversed dimension and selecting the best neighborhood feature value from the optional neighborhood feature values ​​includes: Calculate the cumulative distance of eigenvalues ​​on the traversed dimension, and select the eigenvalue with the shortest cumulative distance as the optional neighborhood eigenvalue; Calculating the number of additional records of the optional neighborhood feature value, the incremental impact of the privacy protection constraint, and the effectiveness score; An optimal neighborhood feature value is selected from the optional neighborhood feature values ​​according to the number of additional records, the privacy protection constraint impact increment, and the effectiveness score.

5. The data privacy protection method according to claim 2, characterized in that: The sampling is performed according to the total number of data records to be purchased by the data element purchaser to obtain a purchase data set corresponding to the data element purchaser, so as to ensure that the purchase data set satisfies the privacy protection constraint corresponding to the purchase record inference attack defense model, including: Generating a data set by random sampling according to the total number of data records to be purchased by the data element purchaser, and eliminating the data set that does not meet the privacy protection constraint; The utilization rates of the remaining data sets are calculated, and the remaining data set with the highest utilization rate is selected as the purchased data set.

6. The data privacy protection method according to claim 2, characterized in that: The sampling is performed according to the total number of data records to be purchased by the data element purchaser to obtain a purchase data set corresponding to the data element purchaser, so as to ensure that the purchase data set satisfies the privacy protection constraint corresponding to the purchase record inference attack defense model, including: Initialize the purchase data set by distributed sampling according to the total number of data records to be purchased by the data element purchaser; Performing addition and deletion processing on the initialized data set, and checking whether the processed data set satisfies the privacy protection constraint; If the privacy protection constraint is satisfied, then calculating the acceptance probability of the processed data set; Whether to select the processed data set as the purchase data set is determined according to the acceptance probability.

7. A data privacy protection device, characterized in that: The data privacy protection device comprises: An intention acquisition module, used to acquire a purchase intention set published by a data element purchaser, wherein the purchase intention set includes at least one real intention; A boundary determination module, used to determine an upper confidence limit and a lower confidence limit of the purchase intention set according to the number of real intentions included in the purchase intention set; A model building module, for establishing a release intention-based attack model, an efficiency maximization attack model, and a purchase record inference attack model for the data element buyer based on the purchase intention set, the confidence upper bound, the confidence lower bound, the data element background knowledge, and the purchase record of the data element buyer; The model building module is further used to build a uniform attack defense model based on publishing intention, an efficiency maximization attack defense model and a purchase record inference attack defense model based on publishing intention, the efficiency maximization attack model and the purchase record inference attack model with a confidence threshold as a constraint condition; The privacy protection module is used to protect the privacy of the data element purchaser based on the uniform attack defense model based on the publishing intention, the efficiency maximization attack defense model and the purchase record inference attack defense model.

8. A data privacy protection device, characterized in that: The data privacy protection device includes: a memory, a processor, and a data privacy protection program stored in the memory and executable on the processor. When the data privacy protection program is executed by the processor, the data privacy protection method according to any one of claims 1 to 6 is implemented.

9. A storage medium, characterized in that: The storage medium stores a data privacy protection program, which, when executed by a processor, implements the data privacy protection method according to any one of claims 1 to 6.

10. A computer program product, characterized in that The computer program product includes a data privacy protection program, and when the data privacy protection program is executed by a processor, the data privacy protection method according to any one of claims 1 to 6 is implemented.

Images