Information protection method for supply chain purchase relation chain data

Abstract

The invention relates to the technical field of information protection, in particular to an information protection method for supply chain purchase relation chain data, which comprises the following steps: creating an alternative enterprise name set; according to a MinHash algorithm, generating an alternative enterprise name signature set and a to-be-evaluated enterprise name signature, and creating a first confusion signature set; generating a buyer enterprise name signature according to a MinHash algorithm, and creating a second confusion signature set; combining the alternative enterprise name signatures in the second confusion signature set to generate a purchaser enterprise combination, and generating a transaction relation combination according to the alternative enterprise name signatures in the first confusion signature set and the purchaser enterprise combination; the transaction relation combinations are sent to a third-party data source, the credit score of each transaction relation combination is obtained, the score of the transaction relation combination of the to-be-evaluated enterprise and the buyer enterprise is selected, and the problem that in the prior art, the protection effect on enterprise transaction relation information needing to be protected is poor is solved.

Description

Technical Field

[0001] The present invention relates to the field of information protection technology, and in particular to an information protection method for supply chain procurement relationship chain data. Background Art

[0002] With the development of supply chain finance and credit financing, financial institutions are increasingly relying on data-driven methods to conduct credit assessments on corporate customers. In this process, the procurement relationship chain data of enterprises has become a key factor in assessing the credit qualifications of enterprises. For example, if the transaction buyers of an enterprise include high-quality or core enterprises, this will help improve the credit score of the enterprise and may bring it more favorable financing conditions.

[0003] Although enterprises may be willing to authorize financial institutions to use their procurement relationship chain data, they usually do not want this information to be leaked to other third parties because it may affect the business confidentiality and market competitiveness of the enterprise. While protecting the enterprise procurement relationship chain information, financial institutions also need to actively use third-party data to enhance the risk assessment capabilities of enterprises and their buyers, which is crucial to improving the information protection capabilities of the supply chain procurement relationship chain data.

[0004] Chinese Patent Publication No.: CN115943411A discloses a noise transaction for protecting data, including a processor configured to generate a blockchain request including data for a blockchain, create a tag value identifying whether the blockchain request is a noise request including false data or a noise-free request including non-noise data, and store the tag value in the blockchain request; and a network interface configured to send the blockchain request to one or more blockchain peers.

[0005] It can be seen that the existing technology has the following problems: due to the small size of the set of alternative enterprise names and the lack of a reasonable way to expand the set, the obfuscation effect on the transaction relationship combination information that needs to be protected is poor, resulting in poor protection of the enterprise transaction relationship information that needs to be protected. Summary of the invention

[0006] To this end, the present invention provides a method for protecting supply chain procurement relationship chain data, so as to overcome the problem of poor protection effect on enterprise transaction relationship information that needs to be protected in the prior art.

[0007] To achieve the above-mentioned purpose, the present invention provides a method for protecting information of supply chain procurement relationship chain data, comprising:

[0008] Create a candidate enterprise name set including the name of the enterprise to be evaluated, the names of n purchasing enterprises that have transaction relations with the enterprise to be evaluated, and the names of other enterprises in the same industry as the enterprise to be evaluated;

[0009] Processing the candidate enterprise name set according to the MinHash algorithm to generate a candidate enterprise name signature set;

[0010] Processing the name of the enterprise to be evaluated according to the MinHash algorithm to generate a signature of the enterprise name to be evaluated, and determining M qualified candidate enterprise name signatures according to the first semantic similarity between the candidate enterprise name signatures in the candidate enterprise name signature set and the signature of the enterprise name to be evaluated, and creating a first obfuscated signature set including the M candidate enterprise name signatures;

[0011] Processing the names of n purchasing companies that have a transaction relationship with the company to be evaluated according to the MinHash algorithm, generating n purchasing company name signatures, and determining qualified N candidate company name signatures according to the second semantic similarity between the candidate company name signatures in the candidate company name signature set and the single purchasing company name signature, and creating a second obfuscated signature set including n×N candidate company name signatures;

[0012] Combine and generate the candidate enterprise name signatures in the second obfuscated signature set A combination of buyer companies, and based on the candidate company name signatures in the first obfuscated signature set Generation of a portfolio of acquiring companies Combination of transaction relationships;

[0013] The The transaction relationship combinations are sent to a third-party data source to obtain the credit score of each transaction relationship combination, and the score of the transaction relationship combination between the enterprise to be evaluated and the purchasing enterprise is selected;

[0014] When the creation of the candidate enterprise name set is completed, the expansion of the candidate enterprise name set is determined according to the number of enterprise names in the candidate enterprise name set;

[0015] When the candidate enterprise name set, the name of the enterprise to be evaluated, and the name of the purchasing enterprise having a transaction relationship with the enterprise to be evaluated are processed according to the MinHash algorithm, the number of hash functions in the MinHash algorithm is determined according to the number of enterprise names in the candidate enterprise name set;

[0016] when After the transaction relationship combination is generated, according to the number of transaction relationship combinations Determines adjustments to preset semantic similarity.

[0017] Furthermore, according to the comparison result that the number of enterprise names in the candidate enterprise name set is less than or equal to the first preset number of enterprise names, it is determined to expand the candidate enterprise name set.

[0018] Further, according to the comparison result that the first relative difference is less than or equal to the first preset relative difference, it is determined that the set of candidate enterprise names is expanded in a first expansion manner, and according to the comparison result that the first relative difference is greater than the first preset relative difference, it is determined that the set of candidate enterprise names is expanded in a second expansion manner, the first expansion manner is to add the names of all enterprises covered by the to-be-evaluated enterprise as the center and the preset distance as the radius to the set of candidate enterprise names, and the second expansion manner is to expand the set of candidate enterprise names according to the business scope text of the to-be-evaluated enterprise;

[0019] The first relative difference is determined by the number of enterprise names and a first preset number of enterprise names.

[0020] Further, the preset distance is calculated according to the following formula, setting

[0021]

[0022] Wherein, J represents a preset distance, and ΔL represents the first relative difference.

[0023] Further, the cosine similarity between the name of the company to be expanded in the set of company names to be expanded and the business scope text of the company to be evaluated is calculated according to the following formula:

[0024]

[0025] Wherein, Y represents the cosine similarity, A represents the feature vector of the enterprise name to be expanded, and B represents the feature vector of the business scope text of the enterprise to be evaluated;

[0026] According to the comparison result that the cosine similarity is greater than a preset cosine similarity, it is determined that the company name to be expanded is added to the set of candidate company names.

[0027] Furthermore, based on the comparison result that the number of company names in the candidate company name set is less than or equal to the second preset company name number, the number of hash functions in the MinHash algorithm is determined to be the first number of hash functions; based on the comparison result that the number of company names in the candidate company name set is greater than the second preset company name number, the number of hash functions in the MinHash algorithm is determined to be the second number of hash functions.

[0028] Further, the first semantic similarity between the candidate enterprise name signatures in the candidate enterprise name signature set and the enterprise name signature to be evaluated is calculated according to the following formula, and X1=D1 / D2 is set, wherein X1 represents the first semantic similarity, D1 represents the number of identical hash values ​​at the same position in the candidate enterprise name signature and the enterprise name signature to be evaluated, and D2 represents the total number of hash values ​​of the candidate enterprise name signature and the enterprise name signature to be evaluated;

[0029] According to the comparison result that the first semantic similarity is less than or equal to the first preset semantic similarity, it is determined that the candidate enterprise name signature is unqualified; according to the comparison result that the first semantic similarity is greater than the first preset semantic similarity, it is determined that the candidate enterprise name signature is qualified.

[0030] Further, the second semantic similarity between the candidate enterprise name signature and the single purchaser enterprise name signature in the candidate enterprise name signature set is calculated according to the following formula, and X2=E1 / E2 is set, wherein X2 represents the second semantic similarity, E1 represents the number of identical hash values ​​at the same position in the candidate enterprise name signature and the single purchaser enterprise name signature, and E2 represents the total number of hash values ​​of the candidate enterprise name signature and the single purchaser enterprise name signature;

[0031] According to the comparison result that the second semantic similarity is less than or equal to the second preset semantic similarity, it is determined that the alternative enterprise name signature is unqualified; according to the comparison result that the second semantic similarity is greater than the second preset semantic similarity, it is determined that the alternative enterprise name signature is qualified.

[0032] Furthermore, according to the number of transaction relationship combinations The comparison results that are less than or equal to the preset number of transaction relationship combinations determine that the preset semantic similarity is adjusted.

[0033] Further, when it is determined to adjust the preset semantic similarity, the preset semantic similarity is adjusted in a first adjustment manner according to the comparison result that the second relative difference is less than or equal to the second preset relative difference, and the preset semantic similarity is adjusted in a second adjustment manner according to the comparison result that the second relative difference is greater than the second preset relative difference, wherein the first adjustment manner is to adjust the preset semantic similarity according to the first adjustment coefficient The first preset semantic similarity is adjusted, and the second adjustment method is based on the second adjustment coefficient The second preset semantic similarity is adjusted, wherein ΔP represents the second relative difference, which is determined by the number of transaction relationship combinations. The quantity is determined by the combination with the preset transaction relationship.

[0034] Compared with the prior art, the beneficial effect of the present invention lies in that the present invention expands the set of alternative enterprise names to increase the number of enterprises similar to the enterprise name to be evaluated, provides more noise data when performing MinHash operations and creating obfuscated signature sets, thereby improving the degree of obfuscation of the enterprise to be evaluated and its purchasing enterprise, better protecting their business secrets, and improving the strength of data protection. By setting a first preset number of enterprise names, the size of the set of alternative enterprise names can be flexibly adjusted according to actual conditions and protection needs, thereby avoiding the effect of information protection affected by too few enterprise names and avoiding too many enterprise names resulting in too long calculation time and waste of computing resources.

[0035] Furthermore, the present invention sets different expansion methods and accurately selects the most appropriate expansion strategy, thereby effectively improving the anonymity level and protection effect of the enterprise to be evaluated and its purchasing enterprise. By setting the first relative difference to flexibly determine the expansion method according to actual conditions and protection needs, it avoids the situation where too few enterprise names affect the information protection effect while avoiding the situation where too many enterprise names lead to too long calculation time and waste of computing resources.

[0036] Furthermore, the present invention accurately calculates the preset distance according to the first relative difference, so that the determination of the preset distance is more efficient and accurate, ensuring the expansion effect while avoiding excessive expansion of the set of candidate enterprise names.

[0037] Furthermore, the present invention measures the similarity between the feature vector of the enterprise name to be expanded and the feature vector of the business scope text of the enterprise to be evaluated through cosine similarity, thereby ensuring that the enterprises added to the set have a certain business relevance to the enterprise to be evaluated so as to more accurately query enterprises similar to the enterprise to be evaluated. Only when the cosine similarity exceeds the preset cosine similarity is the enterprise name to be expanded added to the set of candidate enterprise names, thereby protecting information and blurring potential supply chain relationships. At the same time, it prevents enterprise names that are completely unrelated to the enterprise to be evaluated from being added to the set of candidate enterprise names, avoiding the introduction of too much noise data, and helping to maintain the quality and practicality of the set of candidate enterprise names.

[0038] Furthermore, the number of hash functions in the present invention affects the accuracy of the MinHash signature. By selecting an appropriate number of hash functions, while maintaining computational efficiency, it is ensured that the generated signature can accurately reflect the similarity between names. When the set of candidate enterprise names is small, using fewer hash functions is sufficient to provide a good signature effect, thereby saving computing resources. When the set of candidate enterprise names is large, more hash functions are used to provide a more detailed signature to cope with the complexity of the set of candidate enterprise names.

[0039] Furthermore, the present invention sets a first preset semantic similarity to select alternative enterprise name signatures that are highly similar to the enterprise to be evaluated, thereby ensuring that the enterprise added to the first obfuscation signature set can effectively obfuscate the true identity of the enterprise to be evaluated, thereby improving the quality of information protection, and making it more difficult for a third-party data source to distinguish which are real transaction relationships when receiving information, thereby better protecting corporate privacy.

[0040] Furthermore, the present invention sets a second preset semantic similarity X02 to ensure that only candidate enterprise name signatures above a certain similarity level are considered qualified to be added to the second obfuscated signature set, which helps to protect the privacy of purchasing enterprises that have a transaction relationship with the enterprise to be evaluated and reduce potential information leakage risks.

[0041] Furthermore, the present invention controls the scale of generated transaction relationship combinations by comparing the number of transaction relationship combinations with the preset number of transaction relationship combinations. Too many generated combinations will lead to reduced efficiency of data processing and analysis, while too few combinations will not be enough to provide sufficient anonymization protection. The preset semantic similarity is adjusted to help achieve an appropriate scale of obfuscated signature set.

[0042] Furthermore, the present invention calculates the second relative difference and compares it with the second preset relative difference to finely manage the adjustment of the preset semantic similarity, ensuring that the generated transaction relationship combinations are neither too many nor too few, maintaining the rationality of the obfuscated signature set, and thus accurately adjusting the number of transaction relationship combinations.

[0043] Furthermore, the present invention accurately calculates the adjustment coefficient according to the second relative difference to finely adjust the preset semantic similarity, ensuring that the generated transaction relationship combinations are neither too many nor too few, maintaining the rationality of the obfuscated signature set, and thus accurately adjusting the number of transaction relationship combinations, ensuring the effect of information protection while avoiding the waste of computing resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a flow chart of a method for protecting information of supply chain procurement relationship chain data according to an embodiment of the present invention;

[0045] Figure 2 It is a logic block diagram of the process of expanding the set of candidate enterprise names in the information protection method of the supply chain procurement relationship chain data in an embodiment of the present invention;

[0046] Figure 3 It is a logic block diagram of a preset semantic similarity adjustment process in the information protection method of supply chain procurement relationship chain data in an embodiment of the present invention; DETAILED DESCRIPTION

[0047] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0048] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0049] See also Figure 1-3 As shown, Figure 1 It is a flow chart of a method for protecting information of supply chain procurement relationship chain data according to an embodiment of the present invention; Figure 2 It is a logic block diagram of the process of expanding the set of candidate enterprise names in the information protection method of the supply chain procurement relationship chain data in an embodiment of the present invention; Figure 3 It is a logic block diagram of a preset semantic similarity adjustment process in the information protection method for supply chain procurement relationship chain data according to an embodiment of the present invention.

[0050] The information protection method of the supply chain procurement relationship chain data of the embodiment of the present invention includes:

[0051] Step S1, creating a candidate enterprise name set including the name of the enterprise to be evaluated, the names of n purchasing enterprises that have transaction relations with the enterprise to be evaluated, and the names of other enterprises in the same industry as the enterprise to be evaluated;

[0052] Step S2, processing the candidate enterprise name set according to the MinHash algorithm to generate a candidate enterprise name signature set;

[0053] Step S3, processing the name of the enterprise to be evaluated according to the MinHash algorithm to generate a signature of the enterprise name to be evaluated, and determining M qualified candidate enterprise name signatures according to the first semantic similarity X1 between the candidate enterprise name signatures in the candidate enterprise name signature set and the signature of the enterprise name to be evaluated, and creating a first obfuscated signature set including the M candidate enterprise name signatures;

[0054] Step S4, processing the names of n purchasing enterprises that have a transaction relationship with the enterprise to be evaluated according to the MinHash algorithm, generating n purchasing enterprise name signatures, and determining qualified N candidate enterprise name signatures according to the second semantic similarity X2 between the candidate enterprise name signatures in the candidate enterprise name signature set and the single purchasing enterprise name signature, and creating a second obfuscated signature set including n×N candidate enterprise name signatures;

[0055] Step S5: Combine and generate the candidate enterprise name signatures in the second obfuscated signature set A combination of buyer companies, and based on the candidate company name signatures in the first obfuscated signature set Generation of a portfolio of acquiring companies Combination of transaction relationships;

[0056] Step S6, the The transaction relationship combinations are sent to a third-party data source to obtain the credit score of each transaction relationship combination, and the score of the transaction relationship combination between the enterprise to be evaluated and the purchasing enterprise is selected.

[0057] In an embodiment of the present invention, the credit score is provided by a third-party data source. The third-party data source collects relevant data about each enterprise in the transaction relationship combination, and the relevant data includes public financial reports, credit records, industry rankings, and historical transaction data. Features are extracted from the collected data. The third-party data source uses a credit scoring model to process the extracted features. The credit scoring model preferably uses a neural network. The extracted features are input into the credit scoring model. The model outputs a credit score that reflects the credit risk of the enterprise in the transaction relationship combination.

[0058] Specifically, when the candidate enterprise name set is created, it is determined whether to expand the candidate enterprise name set according to the comparison result between the number L of enterprise names in the candidate enterprise name set and the first preset number L1 of enterprise names;

[0059] When L≤L1, it is determined to expand the candidate enterprise name set;

[0060] When L>L1, it is determined not to expand the candidate enterprise name set;

[0061] In the embodiment of the present invention, the first preset number of enterprise names L1 is set to 1000, and those skilled in the art may adjust the first preset number of enterprise names L1 according to specific circumstances.

[0062] Specifically, the present invention expands the set of candidate enterprise names to increase the number of enterprises similar to the enterprise name to be evaluated, provides more noise data when performing MinHash operations and creating obfuscated signature sets, thereby improving the degree of obfuscation of the enterprise to be evaluated and its purchasing enterprise, better protecting their business secrets, and improving the strength of data protection. By setting a first preset number of enterprise names, the size of the set of candidate enterprise names can be flexibly adjusted according to actual conditions and protection needs, thereby avoiding the situation where too few enterprise names affect the effect of information protection, and avoiding the situation where too many enterprise names lead to too long calculation time and waste of computing resources.

[0063] Specifically, when it is determined to expand the candidate enterprise name set, a first relative difference △L between the number of enterprise names L and the first preset number of enterprise names L1 is calculated, and an expansion method for expanding the candidate enterprise name set is determined according to a comparison result of the first relative difference △L and the first preset relative difference △L1, △L=(L1-L) / L1;

[0064] When △L≤△L1, it is determined to expand the candidate enterprise name set in a first expansion manner;

[0065] When △L>△L1, it is determined to expand the candidate enterprise name set in a second expansion manner;

[0066] Among them, the first expansion method is to add the names of all enterprises covered by the preset distance J as the radius with the enterprise to be evaluated as the center to the set of alternative enterprise names, and the second expansion method is to expand the set of alternative enterprise names according to the business scope text of the enterprise to be evaluated.

[0067] In the embodiment of the present invention, the first preset relative difference △L1 is set to 0.4. The first preset relative difference △L1 is obtained when the number of enterprise names L is 600. Those skilled in the art may adjust the first preset relative difference △L1 according to specific circumstances.

[0068] Specifically, the present invention sets different expansion methods and accurately selects the most appropriate expansion strategy, effectively improving the anonymity level and protection effect of the enterprise to be evaluated and its purchasing enterprise. By setting the first relative difference, the expansion method is flexibly determined according to the actual situation and protection needs, avoiding the fact that too few enterprise names affect the information protection effect while avoiding too many enterprise names resulting in too long calculation time and waste of computing resources.

[0069] Specifically, when it is determined to expand the candidate enterprise name set in the first expansion mode, the preset distance J is calculated according to the following formula:

[0070]

[0071] Specifically, the present invention accurately calculates the preset distance according to the first relative difference, so that the determination of the preset distance is more efficient and accurate, ensuring the expansion effect while avoiding excessive expansion of the candidate enterprise name set.

[0072] Specifically, when it is determined to expand the candidate enterprise name set in the second expansion mode, the cosine similarity Y between the enterprise name to be expanded in the enterprise name set to be expanded and the business scope text of the enterprise to be evaluated is calculated according to the following formula:

[0073]

[0074] Wherein, A represents the feature vector of the enterprise name to be expanded, and B represents the feature vector of the business scope text of the enterprise to be evaluated.

[0075] Specifically, when the cosine similarity Y is calculated, it is determined whether to add the enterprise name to be expanded to the candidate enterprise name set according to the comparison result of the cosine similarity Y and the preset cosine similarity Y0;

[0076] When Y≤Y0, it is determined not to add the enterprise name to be expanded to the candidate enterprise name set;

[0077] When Y>Y0, it is determined to add the enterprise name to be expanded to the candidate enterprise name set;

[0078] In the embodiment of the present invention, the set of enterprise names to be expanded is created by randomly selecting 60% of the names of all enterprises in the country, and the preset cosine similarity Y0 is 0.6. Those skilled in the art can adjust the preset cosine similarity Y0 according to specific circumstances.

[0079] Specifically, the present invention measures the similarity between the feature vector of the enterprise name to be expanded and the feature vector of the business scope text of the enterprise to be evaluated through cosine similarity, ensures that the enterprises added to the set have a certain business relevance with the enterprise to be evaluated so as to more accurately query the enterprises similar to the enterprise to be evaluated, and only when the cosine similarity exceeds the preset cosine similarity is the enterprise name to be expanded added to the set of candidate enterprise names, thereby protecting information and blurring potential supply chain relationships, while preventing enterprise names that are completely unrelated to the enterprise to be evaluated from being added to the set of candidate enterprise names, avoiding the introduction of too much noise data, and helping to maintain the quality and practicality of the set of candidate enterprise names.

[0080] Specifically, when the candidate enterprise name set, the name of the enterprise to be evaluated, and the name of the purchasing enterprise having a transaction relationship with the enterprise to be evaluated are processed according to the MinHash algorithm, the number of hash functions in the MinHash algorithm is determined according to the comparison result of the number of enterprise names L in the candidate enterprise name set and the second preset number of enterprise names L2;

[0081] When L≤L2, the number of hash functions in the MinHash algorithm is determined to be the first number of hash functions H1;

[0082] When L>L2, the number of hash functions in the MinHash algorithm is determined to be the second number of hash functions H2;

[0083] In the embodiment of the present invention, the first hash function number H1 is preferably 100, the second hash function number H2 is preferably 300, and the second preset company name number L2 is 1200. Those skilled in the art can adjust the first hash function number H1, the second hash function number H2 and the second preset company name number L2 according to specific circumstances.

[0084] Specifically, the number of hash functions in the present invention affects the accuracy of the MinHash signature. By selecting an appropriate number of hash functions, while maintaining computational efficiency, it is ensured that the generated signature can accurately reflect the similarity between names. When the set of candidate enterprise names is small, using fewer hash functions is sufficient to provide a good signature effect, thereby saving computing resources. When the set of candidate enterprise names is large, more hash functions are used to provide a more detailed signature to cope with the complexity of the set of candidate enterprise names.

[0085] Specifically, when creating the first obfuscated signature set, the first semantic similarity X1 between the candidate enterprise name signatures in the candidate enterprise name signature set and the enterprise name signature to be evaluated is calculated, and whether the candidate enterprise name signature is qualified is determined based on the comparison result of the first semantic similarity X1 and the first preset semantic similarity X01, and X1=D1 / D2 is set, wherein D1 represents the number of identical hash values ​​at the same position in the candidate enterprise name signature and the enterprise name signature to be evaluated, and D2 represents the total number of hash values ​​of the candidate enterprise name signature and the enterprise name signature to be evaluated;

[0086] When X1≤X01, it is determined that the candidate enterprise name signature is unqualified;

[0087] When X1>X01, it is determined that the candidate enterprise name signature is qualified;

[0088] In the embodiment of the present invention, the first preset semantic similarity X1 is set to 0.9, and those skilled in the art may adjust the first preset semantic similarity X1 according to specific circumstances.

[0089] Specifically, the present invention sets a first preset semantic similarity to select alternative enterprise name signatures that are highly similar to the enterprise to be evaluated, thereby ensuring that the enterprise added to the first obfuscation signature set can effectively obfuscate the true identity of the enterprise to be evaluated, improving the quality of information protection, and making it more difficult for a third-party data source to distinguish which are real transaction relationships when receiving information, thereby better protecting corporate privacy.

[0090] Specifically, when creating the second obfuscated signature set, the second semantic similarity X2 between the candidate enterprise name signature and the single purchaser enterprise name signature in the candidate enterprise name signature set is calculated, and the comparison result between the second semantic similarity X2 and the second preset semantic similarity X02 determines whether the candidate enterprise name signature is qualified, and sets X2=E1 / E2, wherein E1 represents the number of identical hash values ​​at the same position in the candidate enterprise name signature and the single purchaser enterprise name signature, and E2 represents the total number of hash values ​​of the candidate enterprise name signature and the single purchaser enterprise name signature;

[0091] When X2≤X02, it is determined that the candidate enterprise name signature is unqualified;

[0092] When X2>X02, it is determined that the candidate enterprise name signature is qualified;

[0093] In the embodiment of the present invention, the second semantic similarity X2 is set to 0.8, and those skilled in the art may adjust the second semantic similarity X2 according to specific circumstances.

[0094] Specifically, the present invention sets a second preset semantic similarity X02 to ensure that only candidate enterprise name signatures above a certain similarity level are considered qualified to be added to the second obfuscated signature set, which helps to protect the privacy of purchasing enterprises that have a transaction relationship with the enterprise to be evaluated and reduce potential information leakage risks.

[0095] Specifically, when After the transaction relationship combination is generated, according to the number of transaction relationship combinations The comparison result with the preset transaction relationship combination quantity P determines whether to adjust the preset semantic similarity;

[0096] when , it is determined to adjust the preset semantic similarity;

[0097] when , it is determined that the preset semantic similarity is not adjusted;

[0098] In the embodiment of the present invention, the preset transaction relationship combination quantity P is set to 10,000. Those skilled in the art may adjust the preset transaction relationship combination quantity P according to specific circumstances.

[0099] Specifically, the present invention controls the scale of generated transaction relationship combinations by comparing the number of transaction relationship combinations with the preset number of transaction relationship combinations. Too many generated combinations will lead to reduced efficiency of data processing and analysis, while too few combinations will not be enough to provide sufficient anonymization protection. The preset semantic similarity is adjusted to help achieve an appropriate scale of obfuscated signature set.

[0100] Specifically, when it is determined to adjust the preset semantic similarity, the number of transaction relationship combinations is calculated. The preset transaction relationship combination quantity P and the second relative difference △P are combined, and the adjustment method for adjusting the preset semantic similarity is determined according to the comparison result of the second relative difference △P and the second preset relative difference △P0, and the setting

[0101] When △P≤△P0, it is determined to adjust the preset semantic similarity in a first adjustment manner;

[0102] When △P>△P0, it is determined to adjust the preset semantic similarity in a second adjustment manner;

[0103] The first adjustment method is to adjust the first preset semantic similarity X01 according to the first adjustment coefficient T1, and the second adjustment method is to adjust the second preset semantic similarity X02 according to the second adjustment coefficient T2.

[0104] In the embodiment of the present invention, the second preset relative difference △P0 is set to 0.4. The second preset relative difference △P0 is obtained when △P0 is 6000. Those skilled in the art can adjust the second preset relative difference △P0 according to specific circumstances.

[0105] Specifically, the present invention calculates the second relative difference and compares it with the second preset relative difference to finely manage the adjustment of the preset semantic similarity, ensuring that the generated transaction relationship combinations are neither too many nor too few, maintaining the rationality of the obfuscated signature set, and thus accurately adjusting the number of transaction relationship combinations.

[0106] Specifically, when it is determined to adjust the preset semantic similarity in the first adjustment manner, the first adjustment coefficient T1 is calculated according to the following formula:

[0107]

[0108] The adjusted first preset semantic similarity is set to Xx01=X01×T1.

[0109] Specifically, when it is determined to adjust the preset semantic similarity in the second adjustment manner, the second adjustment coefficient T2 is calculated according to the following formula:

[0110]

[0111] The adjusted second preset semantic similarity is set to Xx02=X02×T2.

[0112] Specifically, the present invention accurately calculates the adjustment coefficient according to the second relative difference to finely adjust the preset semantic similarity, ensuring that the generated transaction relationship combinations are neither too many nor too few, maintaining the rationality of the obfuscated signature set, and thus accurately adjusting the number of transaction relationship combinations, ensuring the effect of information protection while avoiding the waste of computing resources.

[0113] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0114] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for protecting information of supply chain procurement relationship chain data, characterized in that: include: Create a candidate enterprise name set including the name of the enterprise to be evaluated, the names of n purchasing enterprises that have transaction relations with the enterprise to be evaluated, and the names of other enterprises in the same industry as the enterprise to be evaluated; Processing the candidate enterprise name set according to the MinHash algorithm to generate a candidate enterprise name signature set; Process the name of the enterprise to be evaluated according to the MinHash algorithm to generate a signature of the enterprise name to be evaluated, determine M qualified candidate enterprise name signatures according to the first semantic similarity between the candidate enterprise name signatures in the candidate enterprise name signature set and the signature of the enterprise name to be evaluated, and create a first obfuscated signature set including the M candidate enterprise name signatures; Process the names of n purchasing companies that have a transaction relationship with the company to be evaluated according to the MinHash algorithm to generate n purchasing company name signatures, and determine qualified N candidate company name signatures according to the second semantic similarity between the candidate company name signatures in the candidate company name signature set and the single purchasing company name signature, and create a second obfuscated signature set including n×N candidate company name signatures; Combine and generate the candidate enterprise name signatures in the second obfuscated signature set A combination of buyer companies, and based on the candidate company name signatures in the first obfuscated signature set Generation of a portfolio of acquiring companies Combination of transaction relationships; The The transaction relationship combinations are sent to a third-party data source to obtain the credit score of each transaction relationship combination, and the score of the transaction relationship combination between the enterprise to be evaluated and the purchasing enterprise is selected; When the creation of the candidate enterprise name set is completed, the expansion of the candidate enterprise name set is determined according to the number of enterprise names in the candidate enterprise name set; When the candidate enterprise name set, the name of the enterprise to be evaluated, and the name of the purchasing enterprise having a transaction relationship with the enterprise to be evaluated are processed according to the MinHash algorithm, the number of hash functions in the MinHash algorithm is determined according to the number of enterprise names in the candidate enterprise name set; when After the transaction relationship combination is generated, according to the number of transaction relationship combinations Determines adjustments to preset semantic similarity.

2. The information protection method of supply chain procurement relationship chain data according to claim 1 is characterized in that: The expansion of the candidate enterprise name set is to determine to expand the candidate enterprise name set under the condition that the number of enterprise names in the candidate enterprise name set is less than or equal to the first preset number of enterprise names.

3. The information protection method of supply chain procurement relationship chain data according to claim 2 is characterized in that: The determining to expand the candidate enterprise name set includes determining to expand the candidate enterprise name set in a first expansion manner under the condition that the first relative difference is less than or equal to the first preset relative difference, wherein the first expansion manner is to add the names of all enterprises covered by a circle with the enterprise to be evaluated as the center and a preset distance as the radius to the candidate enterprise name set; The first relative difference is determined by the number of enterprise names and a first preset number of enterprise names.

4. The information protection method of supply chain procurement relationship chain data according to claim 3 is characterized in that: The determining to expand the set of candidate enterprise names includes determining to expand the set of candidate enterprise names in a second expansion manner under the condition that the first relative difference is greater than a first preset relative difference, and the second expansion manner is to expand the set of candidate enterprise names according to the business scope text of the enterprise to be evaluated.

5. The information protection method of supply chain procurement relationship chain data according to claim 4 is characterized in that: Calculate the preset distance according to the following formula, set Wherein, J represents a preset distance, and ΔL represents the first relative difference.

6. The information protection method of supply chain procurement relationship chain data according to claim 5 is characterized in that: The expanding the candidate enterprise name set according to the business scope text of the enterprise to be evaluated includes determining to add the enterprise name to be expanded to the candidate enterprise name set under the condition that the cosine similarity between the enterprise name to be expanded in the enterprise name set to be expanded and the business scope text of the enterprise to be evaluated is greater than a preset cosine similarity; The cosine similarity between the name of the company to be expanded in the set of company names to be expanded and the business scope text of the company to be evaluated is calculated according to the following formula: Wherein, Y represents the cosine similarity, A represents the feature vector of the enterprise name to be expanded, and B represents the feature vector of the business scope text of the enterprise to be evaluated.

7. The information protection method of supply chain procurement relationship chain data according to claim 6 is characterized in that: The determining of the number of hash functions in the MinHash algorithm includes determining the number of hash functions in the MinHash algorithm to be a first number of hash functions under the condition that the number of company names in the candidate company name set is less than or equal to the second preset number of company names, and determining the number of hash functions in the MinHash algorithm to be a second number of hash functions under the condition that the number of company names in the candidate company name set is greater than the second preset number of company names.

8. The information protection method of supply chain procurement relationship chain data according to claim 7 is characterized in that: The first semantic similarity between the candidate enterprise name signatures in the candidate enterprise name signature set and the enterprise name signature to be evaluated is calculated according to the following formula, setting X1=D1 / D2, wherein X1 represents the first semantic similarity, D1 represents the number of identical hash values ​​at the same position in the candidate enterprise name signature and the enterprise name signature to be evaluated, and D2 represents the total number of hash values ​​of the candidate enterprise name signature and the enterprise name signature to be evaluated; The determination of the qualified M candidate enterprise name signatures includes determining that the candidate enterprise name signatures are unqualified under the condition that the first semantic similarity is less than or equal to the first preset semantic similarity, and determining that the candidate enterprise name signatures are qualified under the condition that the first semantic similarity is greater than the first preset semantic similarity.

9. The information protection method of supply chain procurement relationship chain data according to claim 8 is characterized in that: The second semantic similarity between the candidate enterprise name signatures in the candidate enterprise name signature set and the single purchaser enterprise name signature is calculated according to the following formula, setting X2=E1 / E2, wherein X2 represents the second semantic similarity, E1 represents the number of identical hash values ​​at the same position in the candidate enterprise name signature and the single purchaser enterprise name signature, and E2 represents the total number of hash values ​​of the candidate enterprise name signature and the single purchaser enterprise name signature; The determination of the qualified N candidate enterprise name signatures includes determining that the candidate enterprise name signatures are unqualified under the condition that the second semantic similarity is less than or equal to the second preset semantic similarity, and determining that the candidate enterprise name signatures are qualified under the condition that the second semantic similarity is greater than the second preset semantic similarity.

10. The information protection method of supply chain procurement relationship chain data according to claim 9 is characterized in that: The determination of the adjustment of the preset semantic similarity includes the number of transaction relationship combinations Under the condition that the number of combinations of transaction relationships is less than or equal to the preset number, it is determined to adjust the preset semantic similarity; The determining to adjust the preset semantic similarity includes determining to adjust the preset semantic similarity in a first adjustment manner under the condition that the second relative difference is less than or equal to the second preset relative difference, and determining to adjust the preset semantic similarity in a second adjustment manner under the condition that the second relative difference is greater than the second preset relative difference, wherein the first adjustment manner is to adjust the preset semantic similarity according to the first adjustment coefficient. The first preset semantic similarity is adjusted, and the second adjustment method is based on the second adjustment coefficient The second preset semantic similarity is adjusted, wherein ΔP represents the second relative difference, which is determined by the number of transaction relationship combinations. The quantity is determined by the combination with the preset transaction relationship.

Images