Forward feedback data element cross-platform risk quantitative evaluation method and system

Abstract

The invention discloses a data element cross-platform risk quantitative evaluation method and system based on forward feedback, the method is applied to a first platform sharing data, and the method comprises the following steps: determining a risk value and a weight of to-be-shared data according to attribute distribution and sensitive information content of the to-be-shared data; according to a sensitivity level comparison result between the first platform and the second platform and asset flow information, determining an asset risk score and a data processing process risk score of the first platform, the asset flow information including a maximum reachable hop count of the first platform, a node degree of each host in the first platform, and a historical interaction entropy between the first platform and the second platform; and according to the asset risk scores of the first platform and the second platform, the data processing process scores of the first platform and the second platform, and the risk value and the weight of the to-be-shared data, determining a forward feedback quantitative score shared by the to-be-shared data between the first platform and the second platform. The method is higher in evaluation precision.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of data processing, and particularly relates to a forward feedback data element cross-platform risk quantitative evaluation method and system. BACKGROUND

[0002] With the increasing value of data elements, high-frequency and high-sensitivity flow of data between different big data platforms (such as cloud platforms and cross-institution platforms) has become the norm. However, before each specific data flow behavior occurs, the current industry generally lacks a quantifiable dynamic security evaluation mechanism for this flow event, that is, it is impossible to determine whether the behavior of transmitting a data table from an upstream big data platform A to a downstream big data platform B is safe without exposing sensitive information of each party.

[0003] Traditional technologies mostly adopt static evaluation modes and lack dynamic risk quantification capabilities for cross-platform data flow scenarios. Specifically, these methods mostly score based on static internal asset scanning or historical vulnerability libraries of the platform, only reflect the inherent risk state of the platform itself, and cannot integrate dynamic factors such as data characteristics of this transmission, sender environment, receiver architecture, and transmission path exposure into a quantifiable risk indicator, resulting in a disconnection between the evaluation results and the actual flow risk.

[0004] Therefore, the evaluation results obtained based on traditional technologies have low accuracy. SUMMARY

[0005] Embodiments of the present application provide a forward feedback data element cross-platform risk quantitative evaluation method and system, which can solve the above technical problems.

[0006] In a first aspect, a forward feedback data element cross-platform risk quantitative evaluation method is provided, which is applied to a first platform sharing data. The method comprises: determining a risk value of the data to be shared according to an attribute distribution and a sensitive information content of the data to be shared and a weight ; determining an asset risk score and a data processing process risk score of the first platform according to a sensitive level comparison result between the first platform and a second platform and asset flow information, wherein the second platform is a platform receiving the data to be shared, and the asset flow information comprises a maximum reachable hop count of the first platform, a node degree of each host in the first platform, and a historical interaction entropy between the first and second platforms ; ​​Based on the asset risk score of the first platform The data processing score of the first platform The asset risk score of the second platform Risk score of the data processing process of the second platform The risk value of the data to be shared and weight The forward feedback quantization score of the data to be shared between the first and second platforms is determined.

[0007] Secondly, embodiments of the present invention provide a forward-feedback data element cross-platform risk quantification assessment system, including a first platform for sharing data and a second platform for receiving data, wherein the first platform is used for: The risk value of the data to be shared is determined based on the attribute distribution and the content of sensitive information, respectively. and weight ; Based on the sensitivity level comparison results between the first platform and the second platform Asset process information is used to determine the asset risk score of the first platform. Risk score of data processing process The second platform is the platform that receives the data to be shared, and the asset process information includes: the maximum number of hops that the first platform can reach. The node degree of each host in the first platform Historical interaction entropy between the first and second platforms ; Based on the asset risk score of the first platform The data processing score of the first platform The asset risk score of the second platform Risk score of the data processing process of the second platform The risk value of the data to be shared and weight The forward feedback quantization score of the data to be shared between the first and second platforms is determined.

[0008] The beneficial effects of this invention compared to the prior art are: this invention reduces the inherent risks of the data to be shared. , Asset and environmental risks ), processing risks ( ) and sensitivity comparison differences ( ) integrated after calculation data sharing process forward feedback quantification score, to evaluate the risk degree of data processing process; can realize the comprehensive quantification of the whole link of data flow; and these parameters are calculated according to the performance parameters of all platforms participating in sharing, such as 、 And the interaction parameters of the platform, such as , so that the risk assessment method provided by the application can adapt to the specific flow scene in real time, and comprehensively consider the multi-dimensional dynamic information of data itself, sender, receiver and transmission path, so as to overcome the limitations of traditional static methods, and significantly improve the accuracy of evaluation results and the reflection ability to actual risk. BRIEF DESCRIPTION OF DRAWINGS

[0009] Figure 1 The structural schematic diagram of a forward feedback data element cross-platform risk quantification evaluation system provided by the embodiment of the application is shown in the figure. Figure 2 The implementation flowchart of a forward feedback data element cross-platform risk quantification evaluation method provided by the embodiment of the application is shown in the figure. Figure 3 The implementation flowchart of a method for obtaining a sensitive level comparison result provided by the embodiment of the application is shown in the figure. Figure 4 The implementation flowchart of a method for obtaining the weight of each participating host in the second platform provided by the embodiment of the application is shown in the figure. Figure 5 The schematic diagram of an asset topology graph provided by the embodiment of the application is shown in the figure. DETAILED DESCRIPTION

[0010] In the following description, specific details are set forth, such as particular system configurations, techniques, etc., in order to provide a thorough understanding of the embodiments of the application. However, persons skilled in the art should understand that the application can be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the application with unnecessary details.

[0011] It should be understood that when used in the specification and the appended claims of the application, the term "comprising" indicates the presence of the described features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0012] It should also be understood that the term "and / or" used in the specification and the appended claims of the application means any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0013] As used in the description of the application and the appended claims, the term "if' can be interpreted to mean "when" or "upon" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrase "if it is determined" or "if [a described condition or event] is detected" can be interpreted to mean "upon determining" or "in response to determining" or "upon detecting [the described condition or event]" or "in response to detecting [the described condition or event]," depending on the context.

[0014] In addition, the description in the specification of the application and the appended claims, the terms "first", "second", "third", etc. are used only to distinguish descriptions and cannot be understood as indicating or implying relative importance.

[0015] In the description of the application, the reference to "one embodiment" or "some embodiments" and the like means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the application. Thus, the appearances of the phrases "in one embodiment", "in some embodiments", "in other embodiments", "in additional embodiments", and so on, in various places in the specification are not necessarily all referring to the same embodiment, unless otherwise specifically noted. The terms "comprise", "comprising", "have", "having", "include", "including", and "contain", "containing", or variants thereof, mean "including but not limited to", unless otherwise specifically noted.

[0016] The application will be further described in the following detailed description with reference to specific embodiments. The embodiments of the application are not limited to this.

[0017] Embodiment 1 Figure 1 The structure of a forward feedback data element cross-platform risk quantitative evaluation system provided by an embodiment of the application is shown. As an example but not limitation, see Figure 1 , the system 100 can include a first platform A sharing data and a second platform B receiving data to be shared.

[0018] As an example, see Figure 1 , the first platform A and the second platform B each include at least one host, and the data to be shared, such as the data table k, can be stored in a host in the first platform A, such as the host KA. The first platform A can retrieve the data table k from the host KA and forward it to the boundary host IA, and then the boundary host IA in the first platform A forwards the data table k to the boundary host IB of the second platform B, and then the boundary host IB forwards it to the host KB in the second platform B for storing data.

[0019] Exemplarily, before actually starting to share the data, the first platform A needs to determine a forward feedback quantification score of the current sharing data operation, so as to evaluate the risk level of the current sharing data operation.

[0020] Specifically, the first platform A can determine a risk value of the data to be shared and a weight according to the attribute distribution and the sensitive information content of the data to be shared respectively , and then determine an asset risk score and a data processing process risk score of itself according to the comparison result of the sensitive levels between the two platforms , asset flow information , and a data processing process risk score , and finally calculate a forward feedback quantification score of the data to be shared between the first platform and the second platform according to the asset risk score of itself , the data processing process risk score of itself , the asset risk score of the second platform B , and the data processing process risk score of the second platform B .

[0021] The present application integrates the risk of the data to be shared itself , , the asset environment risk , the processing flow risk , and the sensitive comparison difference , and calculates the forward feedback quantification score of the data sharing process, so as to evaluate the risk degree of the data processing process; the present application can realize comprehensive quantification of the whole link of data flow, and the parameters are calculated according to the performance parameters of all platforms participating in sharing, such as , and the interaction parameters of the platforms, such as , so that the risk evaluation method provided by the present application can adapt to specific flow scenarios in real time, comprehensively consider multi-dimensional dynamic information of the data itself, the sender, the receiver and the transmission path, thereby overcoming the limitations of traditional static methods, and significantly improving the accuracy of the evaluation results and the reflection ability to the actual risk.

[0022] Embodiment 2 Figure 2 An implementation flowchart of a forward feedback data element cross-platform risk quantification evaluation method provided by an embodiment of the present application is shown. As an example but not limitation, the method can be applied to the first platform A in the system 100 described above. The method can include steps S201-S203, which will be described below.

[0023] S201, respectively according to the attribute distribution and the sensitive information content of the data to be shared, determine a risk value of the data to be shared and a weight​ .

[0024] In some embodiments, the risk value of the data to be shared can be calculated according to the objective possibility of the data to be shared being successfully attacked according to the inherent risk of the data to be shared in the first platform A; and then the weight of the data to be shared is calculated according to the sensitivity and structural complexity of the data to be shared.

[0025] For example, the first platform can be a school hospital physical examination system, the second platform can be a school student information management system, and the data to be shared can be a student physical examination result table. The first platform and the second platform can be management systems of two different hospitals, and the first platform can send a senior disease sample data table as the data to be shared to the second platform.

[0026] In a possible implementation, a plurality of rounds of differential attack simulation can be performed on the data to be shared to obtain a differential attack risk score of the data to be shared, then statistical inference attack simulation is performed on the data to be shared to obtain a statistical inference risk value of the data to be shared, and finally a weighted sum of the differential attack risk score and the statistical inference risk value is calculated to obtain the risk value of the data to be shared .

[0027] In an example, the data to be shared can be sampled at a preset sampling rate, such as 10%, 30%, 50%, 70%, 90%, etc. For each sampling set, the average value of all numerical fields and the 95% confidence interval thereof are calculated, and the weighted average value of the widths of all 95% confidence intervals is taken as the differential attack risk score .

[0028] In an example, each record in the data to be shared can be traversed to attempt to match in public or auxiliary data sets by using its quasi-identifier, such as “gender + zip code + birth year”. If the matching succeeds, the statistical inference risk value of the record is recorded as 1. If the matching fails, the success probability of recovering the sensitive attribute of the record by statistical inference (such as Bayesian inference) is calculated based on the attribute distribution of the record, and taken as the statistical inference risk value of the record.

[0029] In an example, the arithmetic average value of the statistical inference risk value of each record and the differential attack risk score are weighted and averaged, and then normalized to the interval [0, 1] to obtain the risk value of the data to be shared .

[0030] In a possible implementation, the weight of the data to be shared can be determined according to the number of sensitive fields in the data to be shared and the total number of fields . .

[0031] In one example, the weight of the data to be shared may be calculated by the formula .

[0032] S202, according to the comparison result of the sensitivity level between the first platform and the second platform , asset flow information, determine the asset risk score of the first platform and the data processing process risk score .

[0033] In some embodiments, before calculating the asset risk score and the data processing process risk score , the sensitivity level of the second platform B can be exchanged first, and the weight of each participating host in the second platform can be obtained .

[0034] For example, if the sensitivity level of the second platform B is directly exchanged, the internal security policy of the first platform A may be leaked, therefore, the method provided in Embodiment 3 below can be used to adopt the classic solution of Yao's Garbled Circuit, by constructing a secure two-party computing circuit (i.e. Yao's circuit), so that both parties can only obtain the 1-bit information of "who has a higher level", and cannot infer the specific level value of the other party.

[0035] For example, for privacy considerations, the CKKS fully homomorphic encryption scheme can be used to obtain the weight from the second platform based on the method provided in Embodiment 4 below.

[0036] In some embodiments, the asset impact weight of each participating host in the first platform A can be determined first according to the weight , then the data processing process weight of the first platform A is determined, and finally the asset risk score and the data processing process risk score are determined according to the asset impact weight and the data processing process weight respectively.

[0037] In one possible implementation, the asset impact weight of each participating host can be determined according to the node degree and weight of each participating host in the first platform A.

[0038] In one example, the asset impact weight ​The data processing process risk score can be calculated by the following formula:

[0039] wherein, , are two preset weight parameters.

[0040] In a possible implementation, the asset risk score can be calculated by the following formula:

[0041] wherein, is the process exposure of the current data processing process, represents the hop count from the data source to the host where the current data processing process is located, is the maximum hop count from the data source to the last step of the data processing process, is the maximum process exposure.

[0042] In an example, the process exposure can be determined by the rules defined in Table 1 as follows, which is used to evaluate the opening direction and external interaction risk of the data processing process, such as whether it involves external data exchange, whether there is cross-platform data access, etc.

[0043] Exemplarily, since the flow of the data to be shared definitely exists a data sharing stage, therefore = 4.

[0044] Table 1 Process exposure level of each stage of the processing process

[0045] In a possible implementation, the asset risk score can be calculated by the following formula: .

[0046] Exemplarily, the risk score of the participating host and the risk score of each interface thereon can be preset values.

[0047] In a possible implementation, the data processing process risk score can be calculated by the following formula: .

[0048] S203, according to the first platform asset risk score , the data processing process score of the first platform , the asset risk score of the second platform , the data processing process risk score of the second platform , the risk value of the data to be shared and the weight determining a forward feedback quantization score of the to-be-shared data shared between the first platform and the second platform.

[0049] In one example, the forward feedback quantization score can be calculated by the following formula:

[0050] wherein, is the forward feedback quantization score.

[0051] Optionally, the evaluation method can further include steps S204 and S205.

[0052] S204, determining a risk level of sharing the to-be-shared data according to the forward feedback quantization score.

[0053] For example, different risk level thresholds can be set, and the data sharing process can be automatically classified as low risk, medium risk or high risk according to the thresholds and the forward feedback quantization score.

[0054] If the risk level is not low risk, it is medium risk or high risk, and the following step S205 can be performed.

[0055] S205, performing an intervention operation corresponding to the risk level.

[0056] In one example, if the risk level is medium risk, an alarm can be generated to inform the relevant management personnel, and appropriate measures can be suggested, such as increasing the data table sensitivity level in the second platform B, increasing access control, etc.

[0057] In another example, if the risk level is high risk, the data transmission can be blocked.

[0058] For example, the user can also customize the intervention operation.

[0059] Optionally, after performing the intervention operation, the result of the intervention operation can also be recorded, such as enhancing the policy execution situation to the log, and the platform can be re-applied for calculation score after rectification.

[0060] The present application integrates the risk of the to-be-shared data itself ( , ), asset environment risk ( ), processing flow risk ( ) and sensitivity comparison difference ( ) to calculate the forward feedback quantization score of the data sharing process, so as to evaluate the risk degree of the data processing process; the full link of data flow can be quantified; and these parameters are calculated according to the performance parameters of all platforms participating in sharing, such as , and the interaction parameters of the platform, such as , so that the risk assessment method provided by the application can adapt to specific flow scenarios in real time, comprehensively consider the multi-dimensional dynamic information of the data itself, the sender, the receiver and the transmission path, thereby overcoming the limitations of traditional static methods and significantly improving the accuracy of the evaluation results and the reflection ability to the actual risk.

[0061] Further, the application updates the risk factors in real time through data-driven mode, automatically triggers intervention measures such as blocking and alarm according to the risk threshold, and adapts to the real-time response requirements of high-frequency data transmission.

[0062] Embodiment 3 Figure 3 The implementation flowchart of the method for obtaining the sensitive level comparison result provided by the embodiment of the application is shown. As an example but not limitation, the method can be applied to the system 100 described above. The method can include steps S301-S310, which will be described below.

[0063] S301, the first platform and the second platform generate a temporary mapping rule of the sensitive level.

[0064] In one example, the big data platform (i.e. the first platform A and the second platform B) can negotiate a one-time temporary mapping rule of the sensitive level, which will be destroyed immediately after the end of this agreement.

[0065] For example, the temporary mapping rule can map different sensitive levels to different temporary level values. For example, the platform sensitive level is "low" when the value is 100, "medium" when the value is 200, and "high" when the value is 300.

[0066] S302, the first platform generates a local random disturbance factor and calculates the temporary level value of the first platform after disturbance.

[0067] For example, the temporary level value of the first platform after disturbance may be the sum of its temporary level value and the local random disturbance factor .

[0068] Correspondingly, the second platform generates a local random disturbance factor and calculates the temporary level value of the second platform after disturbance .

[0069] S303, the first platform constructs a Yao circuit.

[0070] For example, the first platform A can use an open-source privacy computing framework (such as ObliVM) to construct a Yao circuit for comparing the size of two 3-bit integers.

[0071] S304, the first platform for Yao's circuit, and temporary level values ​​after disturbance. Obfuscation is performed to generate the first obfuscation table and encryption key.

[0072] S305, the first platform sends the first obfuscation table and encryption key to the second platform.

[0073] Accordingly, the second platform receives the first obfuscation table and the encryption key.

[0074] S306, the second platform determines its own obfuscation label based on the encryption key.

[0075] For example, the second platform B can securely obtain the temporary rating value after its own perturbation using the encryption key provided by the first platform A through the Oblivious Transfer (OT) protocol. The corresponding obfuscation tags.

[0076] S307, the second platform determines the comparison result based on the first obfuscation table and its own obfuscation label.

[0077] For example, the second platform B can evaluate Yao's circuit logic layer by layer according to the first confusion table and confusion label, and finally obtain a 1-bit output bit result (i.e. comparison result).

[0078] S308, the second platform sends the comparison results to the first platform.

[0079] Accordingly, the first platform receives the comparison results.

[0080] S309, the first platform determines the sensitivity level comparison results. .

[0081] For example, if the comparison result is 1, it means It can make It equals the first preset value, for example, 0.8; if the comparison result is 0, it means... It can make It equals the second preset value, for example, 0.4.

[0082] Correspondingly, the second platform B can simultaneously perform this step to determine... .

[0083] S310, first platform and second platform clean up interaction data.

[0084] For example, after the protocol is executed, the first platform A and the second platform B can completely clear all intermediate data such as temporary mapping rules, random perturbation factor r, and obfuscation table in memory and temporary storage, thus completely blocking the inference path.

[0085] Embodiment 4 FIG. 4 shows a flowchart of an implementation of a method for obtaining weights of participating hosts in a second platform according to an embodiment of the present application. By way of example and without limitation, the method can be applied to the system 100 described above. The method can include steps S401-S405, which are described below.

[0086] S401, the first platform determines asset flow information.

[0087] In one possible implementation, the asset flow information can include a maximum reachable hop count of the first platform , a node degree of each host in the first platform , and a historical interaction entropy between the first and second platforms .

[0088] In one example, the first platform A and the second platform B can each maintain an internal asset topology in the form of a directed graph (see FIG. 1), where nodes represent hosts and edges represent network connections or data dependency relationships. Figure 5

[0089] By way of example, the first platform A can use a breadth-first search algorithm to calculate the shortest path length between a host KA storing data to be shared and a host IA participating in the data flow, obtaining a reachable hop count between the node IA and the node KA. The maximum reachable hop count refers to the reachable hop count of the data storage host KA to the boundary host in the first platform A.

[0090] Specifically, a participating host refers to a host participating in the process of sharing data.

[0091] In one example, the node degree of a host can be the sum of the out-degree and the in-degree of the host in the asset topology graph, reflecting the potential influence of the host as a risk spreading node .

[0092] In one example, the historical interaction entropy between the first platform A and the second platform B can be determined according to the amount of data involved in the transmission records related to the data to be shared .

[0093] By way of example, all transmission records related to the data to be shared can be extracted from the de-identified historical interaction logs of the first platform A and the second platform B, the transmission frequency probability of each data amount level can be calculated according to the amount of data involved in the transmission records, the Shannon entropy and the KL divergence can be calculated according to the probability, and finally the historical interaction entropy of the first platform A and the second platform B can be obtained by weighting the Shannon entropy and the KL divergence .

[0094] ​For example, the transmission record can include transmission time, data volume, high-sensitive data volume, and the like.

[0095] For example, the preset data volume level can be: 0-1GB, 1-10GB, 10-50GB, 50GB+.

[0096] For example, the Shannon entropy can satisfy the following formula:

[0097] wherein, is the Shannon entropy, is the transmission frequency probability of the i-th data volume level.

[0098] For example, the KL divergence for measuring sensitivity bias can satisfy the following formula:

[0099] wherein, is the KL divergence between the first platform A and the second platform B, is the proportion of high-sensitive data in the n-th data volume level, is the preset reference proportion in the n-th data volume level.

[0100] For example, the historical interaction entropy between the first platform and the second platform can be calculated by the formula: , , are the preset weights of the Shannon entropy and the KL divergence, respectively, and the sum is 1. For example, , may each be 0.5.

[0101] Optionally, if the first platform and the second platform are interacting for the first time, the historical interaction entropy between the first platform and the second platform can be set to 1.

[0102] S402, the first platform sends asset flow information ciphertext to the second platform.

[0103] In one example, the first platform A can encrypt its asset flow information based on the CKKS homomorphic encryption scheme, and send the encrypted asset flow information to the second platform B together with the public key PK used during encryption and the relinearization key RLK.

[0104] For example, the asset flow information ciphertext can include: , wherein, represents an encryption operation.

[0105] Correspondingly, the second platform B receives the asset flow information ciphertext.

[0106] S403, the second platform performs homomorphic calculation on the asset flow information ciphertext based on the scaling factor to obtain the weight ciphertext.

[0107] In a possible implementation, the attack risk of the host between two big data platforms, which belongs to the first platform / second platform, is different for the first platform A. Assuming two data transmission paths with the same number of hops, one of which all participating hosts are in the first platform A, and the other of which part of the participating hosts are in the first platform A and the other part are in the second platform B, the degree of influence of the risk is different for the first platform; big data platforms tend to trust their own platforms more, and the first path is obviously safer. Therefore, a scaling factor can be introduced, and homomorphic calculation is performed on the asset flow information ciphertext based on the scaling factor.

[0108] In an example, the scaling factor can be calculated by the formula: , and the range is . Therefore, the ciphertext of the scaling factor can be calculated in the ciphertext domain by the formula: .

[0109] Exemplarily, the preset two weight parameters , may be 0.6 and 0.4 respectively.

[0110] In an example, the weight ciphertext can include a denominator ciphertext , an asset normalization weight ciphertext , and a data processing process ciphertext .

[0111] Exemplarily, the denominator ciphertext may be calculated by the formula: +1, wherein is the maximum reachable hop number of the second platform B.

[0112] Exemplarily, the asset normalization weight ciphertext may be calculated by the formula: , wherein = , represents the hop number of any host between the data storage host and the boundary host of the second platform B to the boundary host of the second platform B, represents the approximate reciprocal.

[0113] Exemplarily, the data processing process ciphertext may be calculated by the formula: , wherein represents the hop number of the data source to the current data processing process. The numerical value is equal to the hop number of the boundary host of the second platform B to the current data processing process j.

[0114] S404, the second platform sends the weight ciphertext to the first platform.

[0115] Correspondingly, the first platform receives the weight ciphertext.

[0116] S405, the first platform uses the private key to decrypt the weight ciphertext to obtain the weight .

[0117] In a possible implementation, the private key SK can be used to decrypt the denominator ciphertext , the asset normalization weight ciphertext to obtain the weight .

[0118] In an example, the weight satisfies the following formula:

[0119] wherein, .

[0120] By the method provided in the above embodiments 3 and 4, the CKKS fully homomorphic encryption and the Yao's garbled circuit are combined to obtain the sensitive level comparison result and the weight , which can respectively solve the privacy leakage problem in the asset parameter collaborative operation and the sensitivity level comparison, realize the safe collaboration under the bidirectional privacy protection, and improve the privacy protection strength.

[0121] In the above embodiments, the description of each embodiment has its own focus, and the part not described or recorded in a certain embodiment can be referred to the related description of other embodiments.

Claims

1. A method for forward-feedback data element cross-platform risk quantification assessment, characterized in that, The method is applied to a first platform sharing data, and the method comprises: determine a risk value of the data to be shared according to an attribute distribution, a sensitive information content of the data to be shared respectively and weights ; According to the comparison result of the sensitive levels between the first platform and the second platform , asset flow information, determining an asset risk score of the first platform and a data processing process risk score , wherein the second platform is a platform receiving the to-be-shared data, and the asset flow information comprises: maximum reachable hop count of the first platform , node degree of each host in the first platform , historical interaction entropy between the first platform and the second platform ; based on the first platform asset risk score , a data processing procedure score of the first platform , an asset risk score of the second platform , a data processing procedure risk score of the second platform , a risk value of the data to be shared , and a weight , determine a forward feedback quantification score of the data to be shared shared between the first platform and the second platform.

2. The method of claim 1, wherein, The risk value of the data to be shared is determined according to the attribute distribution, the sensitive information content, and the weight of the data to be shared respectively and the weight , comprising: performing a plurality of rounds of differential attack simulation and re-identification and statistical inference attack simulation on the data to be shared respectively, to obtain a differential attack risk score and a statistical inference risk score of the data to be shared and statistical inference risk value of the data to be shared computing the differential attack risk score obtaining a risk value for the data to be shared by weighting the statistical inference risk values ; According to the number of sensitive fields in the data to be shared and the total number of fields, the weight of the data to be shared is determined .

3. The method of claim 1, wherein, comparing a sensitivity level of the first platform to a sensitivity level of the second platform asset flow information, determining an asset risk score for the first platform and a data processing procedure risk score prior to determining the asset risk score and the data processing procedure risk score, the method further comprising: generating a temporary mapping rule of a sensitivity level, wherein the temporary mapping rule is used to map the sensitivity level to a temporary level number ; temporary level value after perturbation of the first platform , the Yao circuit performs confusion processing to generate a first confusion table and an encryption key; sending the first confusion table to the second platform and receiving a comparison result from the second platform, wherein the comparison result is obtained by the second platform by comparing a confusion label of the second platform with the first confusion table, the confusion label is determined according to the temporary rank value of the second platform after perturbation determined; determining the sensitivity level comparison result based on the comparison result .

4. The method of claim 1, wherein, The sensitive level comparison result between the first platform and the second platform , asset flow information, determining the asset risk score of the first platform and data processing process risk score , comprising: Acquiring weights of each participating host in the second platform ; According to the weights Determining asset influence weights of each participating host in the first platform ; Determining data processing procedure weights for a first platform ; Based on the asset influence weights respectively The weight of the data processing process Determine the asset risk score Risk classification of the data processing process .

5. The method of claim 4, wherein, The method further includes obtaining weights of the participating hosts in the second platform , comprising: sending asset flow information ciphertext to the second platform; receiving weight ciphertext from the second platform, wherein the weight ciphertext is obtained by homomorphic calculation on the asset flow information ciphertext based on a scaling factor; decrypting the weight ciphertext using a private key to obtain the weight .

6. The method of claim 5, wherein, the weights satisfy the following equation: wherein, is the scaling factor, is the maximum reachable hop count for the first platform, second platform, respectively, , is any host between the second platform data store host and the border host is the hop count to the border host of the second platform B.

7. The method of claim 4, wherein, The asset risk score satisfies the equation: wherein, is the asset impact weight of the participating host in the first platform , the risk score of the participating host and the risk score of each interface thereon can be preset values; The data processing procedure risk score satisfies the following equation: wherein, a data processing procedure executed by the first platform when sharing data a data processing procedure risk score.

8. The method of claim 1, wherein, The forward feedback quantization score satisfies the following formula: wherein, is the forward feedback quantization fraction.

9. The method of claim 1, wherein, The method further comprises: determining a risk level of sharing the data to be shared according to the forward feedback quantization score; if the risk level is not low risk, performing an intervention operation corresponding to the risk level.

10. A forward feedback data element cross-platform risk quantification assessment system, characterized in that, The first platform for sharing data and the second platform for receiving data, the first platform is used for: determine a risk value of the data to be shared according to an attribute distribution, a sensitive information content of the data to be shared respectively and weights ; According to the comparison result of the sensitive levels between the first platform and the second platform , asset flow information, determining an asset risk score of the first platform and a data processing process risk score , wherein the second platform is a platform receiving the data to be shared, and the asset flow information comprises: maximum reachable hop count of the first platform , node degree of each host in the first platform , historical interaction entropy between the first platform and the second platform ; based on the first platform asset risk score , a data processing procedure score of the first platform , an asset risk score of the second platform , a data processing procedure risk score of the second platform , a risk value of the data to be shared , and a weight , a forward feedback quantification score of the data to be shared between the first and second platforms is determined.

Images