Enterprise security risk identification and assessment method and system based on identity resolution data

Abstract

The present invention proposes a method and system for enterprise security risk identification and assessment based on identity resolution data, the method comprising: obtaining historical industrial Internet enterprise identity resolution data and corresponding historical industrial Internet enterprise information; establishing an information database, and using all data in the information database as sample data; constructing an enterprise security risk identification and assessment classification model based on an improved CART decision tree model and a support vector machine model, respectively, and using the sample data for training to obtain a final classification model; obtaining the identity resolution data of the industrial Internet enterprise to be verified and its industrial Internet enterprise information, inputting them into the final classification model for classification, constructing a multivariate linear regression model, and performing security risk identification and assessment on the classified industrial Internet enterprise to be verified. The present invention realizes the effective assessment of the security risks of industrial Internet enterprises in a unified and integrated industrial Internet identity resolution system, thereby improving the level of enterprise safety production supervision.

Description

Technical Field

[0001] The present invention relates to the field of risk assessment technology, and in particular to a method and system for identifying and assessing enterprise security risks based on identity resolution data. Background Art

[0002] Today, with the application and development of new-generation electronic information technologies such as 5G, big data, and artificial intelligence, networking, intelligence, and digitalization have become the guiding principles of social development. Society is gradually entering the era of the Internet of Everything, where people, things, and everything can be connected through the internet. In the interconnected Industrial Internet, every industry, enterprise, item, component, and even every piece of information has its own globally unique "identity card," known as an identifier. The Industrial Internet identifier code consists of an identifier prefix and an identifier suffix, separated by the UTF-8 character " / ." The identifier prefix consists of the country code, industry code, and enterprise code, and uniquely identifies the enterprise entity. The identifier suffix consists of an object code and a security code. The security code is optional and uniquely identifies the identified object. The identifier prefix should comply with the identifier resolution system plan. To promote the high-quality development of the Industrial Internet, it is particularly important to establish an identifier resolution standard system.

[0003] Although through the unified and integrated industrial Internet identification resolution system, governments, enterprises or users can use identification to access enterprises and products in various links such as design, production, logistics, sales and use, and intelligently associate relevant information data under different managers, different locations and different data structures, it can achieve the precise connection between the global supply chain system and the enterprise production system, as well as the premise and foundation for the full life cycle management and intelligent services of products. However, this system cannot identify and evaluate the security risks of industrial Internet enterprises. In order to maintain the stability of production operations, ensure enterprise safety and reduce enterprise property losses, there is an urgent need for a scientific, reasonable and highly accurate enterprise security risk identification and assessment method. Summary of the Invention

[0004] In view of this, this application proposes a method and system for identifying and assessing enterprise security risks based on identity resolution data, which is used to solve the problem that the unified and integrated industrial Internet identity resolution system cannot identify and assess the security risks of industrial Internet enterprises.

[0005] The technical solution of the present invention is achieved as follows:

[0006] The present invention proposes a method for identifying and assessing enterprise security risks based on identity resolution data, the method comprising:

[0007] S1, obtain historical industrial Internet enterprise identification resolution data and corresponding historical industrial Internet enterprise information;

[0008] S2, establish an information database, store historical industrial Internet enterprise identification resolution data and corresponding historical industrial Internet enterprise information in the information database, and use all data in the information database as sample data;

[0009] S3, constructing enterprise security risk identification and assessment classification models based on the improved CART decision tree model and support vector machine model respectively, and inputting sample data into the enterprise security risk identification and assessment classification model for training to obtain the final classification model;

[0010] S4, obtain the identification resolution data of the industrial Internet enterprises to be verified and their industrial Internet enterprise information, input them into the final classification model for classification, build a multivariate linear regression model, conduct security risk identification and assessment on the classified industrial Internet enterprises to be verified, and obtain the corresponding security risk identification and assessment report.

[0011] Based on the above technical solution, preferably, step S1 specifically includes:

[0012] The historical industrial Internet enterprise information includes corresponding enterprise legal litigation information, enterprise operating information and management information.

[0013] Based on the above technical solution, preferably, step S2 specifically includes:

[0014] Several risk identification and assessment categories are set up, and the TF-IDF method is used to extract feature information data related to several risk identification and assessment categories from historical industrial Internet enterprise information, and the feature information data corresponding to the historical industrial Internet identifier resolution data are stored one by one in the information database.

[0015] Based on the above technical solution, preferably, step S3 specifically includes:

[0016] S301, normalizing a number of risk identification and assessment categories and feature information data to obtain a category set and a feature set respectively;

[0017] S302: Constructing a first enterprise safety risk identification and assessment classification model based on the CART decision tree model, using the category set as the dependent variable and the feature set as the independent variable, recursively splitting the training samples into the first enterprise safety risk identification and assessment classification model for training, obtaining a first classification output result, and calculating the posterior probability of each data in the training sample;

[0018] S303: Build a second enterprise security risk identification and assessment classification model based on the support vector machine model, input the category set and training samples for training, obtain a second classification output result, and calculate the posterior probability of each data in the training sample;

[0019] S304: Combine the first classification output result and the second classification output result to obtain a final classification output result, and combine them into a final classification model.

[0020] Based on the above technical solution, preferably, in step S302, the training samples are input into the first enterprise security risk identification and assessment classification model in a recursive segmentation manner for training, which specifically includes:

[0021] Create a root node, input the training samples into the root node, and calculate the Gini coefficient G of all training samples in the root node. The root node contains several feature attribute vectors.

[0022] For each feature attribute vector in the root node, a segmentation threshold calculation is performed. The calculation process includes: specifying a feature attribute vector as a partition feature, setting several segmentation thresholds for the partition feature, dividing the training samples into two groups, and calculating the Gini coefficient G' of all training samples after the partition; calculating the Gini coefficient reduction ΔG before and after the partition, ΔG = GG'; calculating the Gini coefficient reduction corresponding to different segmentation thresholds of the partition feature, comparing the Gini coefficient reductions corresponding to different segmentation thresholds, and obtaining the maximum reduction G B And the corresponding segmentation threshold, and use it as the optimal segmentation threshold;

[0023] Perform segmentation threshold calculation on all feature attribute vectors in the root node to obtain the optimal segmentation feature attribute vector and its corresponding optimal segmentation threshold, and segment the root node based on the optimal segmentation feature attribute vector and its corresponding optimal segmentation threshold;

[0024] Determine whether the node after division is a leaf node. If not, use the node as the root node of the subtree and re-divide it. If so, determine whether all nodes have reached leaf nodes at this time. When all nodes have reached leaf nodes, the training process is completed.

[0025] Based on the above technical solution, preferably, in step S302, calculating the posterior probability of each data in the training sample specifically includes:

[0026] The category set contains several output categories, the training sample contains a feature set, and the feature set contains several feature attribute vectors. When the output category is Y k When the characteristic attribute vector Z j The posterior probability is:

[0027]

[0028] Where k = 1, 2, ..., m, m represents the total number of categories in the category set, Y kIndicates that the output category is the k-th category, β(k) represents the training accuracy of the first enterprise safety risk identification and assessment classification model for the k-th output category, j = 1, 2, ..., n, n is the total number of feature attribute vectors, k = t represents when the feature attribute vector Z j When the output category is equal to the true category t, k≠t represents the feature attribute vector Z j When the output category is not equal to the true category t.

[0029] Based on the above technical solution, preferably, in step S303, calculating the posterior probability of each data in the training sample specifically includes:

[0030] The Bayesian formula is used to calculate the posterior probability P(Y k |Z) SVM , when the output category is Y k When the characteristic attribute vector Z j The posterior probability is:

[0031]

[0032] Where k = 1, 2, ..., m, m represents the total number of categories in the category set, i = 1, 2, ..., m, Y k Indicates that the output category is the kth category, Y i Indicates that the output category is the i-th category, j = 1, 2, ..., n, and n is the total number of feature attribute vectors.

[0033] Based on the above technical solution, preferably, step S304 specifically includes:

[0034] The posterior probability P(Y k |Z j ) CART and the posterior probability P(Y k |Z j ) SVM Perform weighted synthesis to obtain:

[0035]

[0036] Among them, A CART represents the classification accuracy of the first enterprise safety risk identification and assessment classification model, A SVM Represents the classification accuracy of the second enterprise safety risk identification and assessment classification model.

[0037] Based on the above technical solution, preferably, in step S4, a multivariate linear regression model is constructed to perform security risk identification and assessment on the classified industrial Internet enterprises to be verified, specifically including:

[0038] Based on several risk identification and assessment categories and sample data training, a multivariate linear regression model y is constructed, and its expression is:

[0039] y=β0+β1x1+β2x2+…+β p x p

[0040] Among them, β i is the regression coefficient, x i is the risk identification and assessment category, i=1,2,…,p.

[0041] The present invention also proposes an enterprise security risk identification and assessment system based on identity resolution data, the system comprising:

[0042] Enterprise data acquisition module, which obtains historical industrial Internet enterprise identification resolution data and corresponding historical industrial Internet enterprise information;

[0043] An information database establishment module establishes an information database, stores historical industrial Internet enterprise identification resolution data and corresponding historical industrial Internet enterprise information in the information database, and uses the data in the information database as sample data;

[0044] The classification module builds enterprise security risk identification and assessment classification models based on the improved CART decision tree model and the support vector machine model, and inputs sample data into the enterprise security risk identification and assessment classification model for training to obtain the final classification model;

[0045] The security risk identification and assessment module obtains the identification resolution data and industrial Internet enterprise information of the industrial Internet enterprises to be verified, inputs them into the final classification model for classification, constructs a multivariate linear regression model, conducts security risk identification and assessment on the classified industrial Internet enterprises to be verified, and obtains the corresponding security risk identification and assessment report.

[0046] The enterprise security risk identification and assessment method and system based on identity resolution data of the present invention have the following beneficial effects compared with the prior art:

[0047] (1) Combining the historical industrial Internet identification resolution data and the corresponding enterprise information to construct an enterprise security risk identification and assessment classification model, first classify the identification resolution data of the industrial Internet enterprise to be verified and the corresponding historical industrial Internet enterprise information, and then input them into the pre-trained multivariate linear regression model for enterprise security risk identification and assessment, thus realizing the unified and integrated industrial Internet identification resolution system to effectively assess the security risks of industrial Internet enterprises, thereby improving the level of enterprise safety production supervision.

[0048] (2) By merging the classification results of the two algorithms using the weighted synthesis method of the posterior probability of the CART decision tree model and the posterior probability of the support vector machine model, the classification accuracy and interpretability of the model are maximized, and the accuracy of enterprise security risk identification and assessment is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0050] Figure 1 This is a flowchart of the steps of a method for identifying and assessing enterprise security risks based on identity resolution data in an embodiment of the present invention;

[0051] Figure 2 This is a flowchart of the steps for training the final classification model in an embodiment of the present invention;

[0052] Figure 3 This is a structural diagram of an enterprise security risk identification and assessment system based on identity resolution data in an embodiment of the present invention. DETAILED DESCRIPTION

[0053] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0054] See also Figure 1 The present embodiment provides a method for identifying and assessing enterprise security risks based on identity resolution data, including the following steps:

[0055] S1, obtain historical industrial Internet enterprise identification resolution data and corresponding historical industrial Internet enterprise information.

[0056] Specifically, the historical industrial Internet enterprise information includes the corresponding enterprise legal proceedings information, enterprise operating information and management information. It should be understood that the industrial Internet enterprise identifier is actually the enterprise's identity identifier, including enterprise identity, product identity and equipment identity, etc. The specific manifestation of the industrial Internet enterprise identifier is a string of characters, including a series of information such as country, region, enterprise, industry category, factory, production batch number and location; enterprise legal proceedings information includes enterprise litigation information, information on the person subject to execution, information on dishonesty, information on restrictions on high consumption, information on finalized cases, information on judgment documents, court announcement information, court opening announcement information, service announcement information, judicial auction information, equity freezing information, case filing information, bankruptcy reorganization information, and information on restrictions on exit from the country; enterprise operating information includes abnormal operations, serious violations of law, equity pledge, equity pledge (ordinary enterprises), equity pledge (IPO listed companies), equity pledge (New Third Board companies), administrative penalties, environmental penalties, tax violations, movable mortgages, liquidation, land mortgages, simplified cancellations, cancellation filings, public notices, tax arrears announcements, shell companies, account opening due diligence and other information; management information includes the history of directors, supervisors and senior managers serving as legal representatives, the history of directors, supervisors and senior managers' overseas investments, the history of directors, supervisors and senior managers' overseas appointments, directors, supervisors and senior managers who have been executed for breach of trust, directors, supervisors and senior managers who have been executed, restrictions on high consumption by directors, supervisors and senior managers, equity freezes of directors, supervisors and senior managers, equity pledges of directors, supervisors and senior managers, judicial documents on directors, supervisors and senior managers, equity pledges of directors, supervisors and senior managers, final cases involving directors, supervisors and senior managers, handling of violations by directors, supervisors and senior managers, case filing information for directors, supervisors and senior managers, court announcements of directors, supervisors and senior managers, announcements of service of documents to directors, supervisors and senior managers, price inquiries and evaluations of directors, supervisors and senior managers, restrictions on exit from the country by directors, supervisors and senior managers, historical breach of trust by directors, supervisors and senior managers, historical persons subject to execution, and historical restrictions on high consumption by directors, supervisors and senior managers.

[0057] S2, establish an information database, store the historical industrial Internet enterprise identification resolution data and the corresponding historical industrial Internet enterprise information in the information database, and use all the data in the information database as sample data.

[0058] Specifically, several risk identification and assessment categories are set up, and the TF-IDF method is used to extract feature information data related to several risk identification and assessment categories from historical industrial Internet enterprise information, and the feature information data corresponding to the historical industrial Internet identifier resolution data are stored one by one in the information database.

[0059] It is important to understand that risk identification and assessment categories include corporate credit risk, corporate market risk, corporate management risk, corporate legal risk, corporate reputation risk, corporate operational risk, corporate product risk, corporate personnel risk, corporate compliance risk, corporate property risk, corporate liability risk, etc.

[0060] The specific implementation steps of the TF-IDF method include:

[0061] Perform word segmentation and stop word processing on historical industrial Internet enterprise information, and extract feature information data of the processed historical industrial Internet enterprise information;

[0062] Calculate the target word segmentation t i The frequency TF that appears in historical industrial Internet enterprise information is calculated as follows:

[0063]

[0064] Among them, n i,j Represents the target word t i In the historical industrial internet enterprise information document d j The number of times it appears in k n k,j Represents document d j The total number of times all words appear in ;

[0065] Calculate the target word segmentation t i The inverse document frequency IDF is calculated as follows:

[0066]

[0067] Where D represents the total number of documents in the historical industrial Internet enterprise information, |{j:t i ∈d j}| indicates that the target word t is included i The number of documents;

[0068] Calculate the weighted value of each target word. The calculation formula is:

[0069] TF-IDF=TF*IDF

[0070] Arrange the weighted values ​​of each target word from large to small, and output the first T words, which are the feature information data.

[0071] S3, builds an enterprise security risk identification and assessment classification model based on the improved CART decision tree model and the support vector machine model respectively, and inputs the sample data into the enterprise security risk identification and assessment classification model for training to obtain the final classification model.

[0072] See also Figure 2 , the implementation steps of step S3 specifically include:

[0073] S301 , normalizing a number of risk identification and assessment categories and feature information data to obtain a category set and a feature set respectively.

[0074] S302, constructing the first enterprise safety risk identification and assessment classification model based on the CART decision tree model, taking the category set as the dependent variable and the feature set as the independent variable, inputting the training samples into the first enterprise safety risk identification and assessment classification model for training using a recursive segmentation method, obtaining the first classification output result, and calculating the posterior probability of each data in the training sample.

[0075] It should be understood that constructing the first enterprise safety risk identification and assessment classification model based on the CART decision tree model is an existing technology, so it will not be described in detail.

[0076] Specifically, in step S302, the training samples are input into the first enterprise security risk identification and assessment classification model in a recursive segmentation manner for training, which specifically includes:

[0077] Create a root node, input the training samples into the root node, and calculate the Gini coefficient G of all training samples in the root node. The root node contains several feature attribute vectors.

[0078] For each feature attribute vector in the root node, a segmentation threshold calculation is performed. The calculation process includes: specifying a feature attribute vector as a partition feature, setting several segmentation thresholds for the partition feature, dividing the training samples into two groups, and calculating the Gini coefficient G' of all training samples after the partition; calculating the Gini coefficient reduction ΔG before and after the partition, ΔG = GG'; calculating the Gini coefficient reduction corresponding to different segmentation thresholds of the partition feature, comparing the Gini coefficient reductions corresponding to different segmentation thresholds, and obtaining the maximum reduction G B And the corresponding segmentation threshold, and use it as the optimal segmentation threshold;

[0079] Perform segmentation threshold calculation on all feature attribute vectors in the root node to obtain the optimal segmentation feature attribute vector and its corresponding optimal segmentation threshold, and segment the root node based on the optimal segmentation feature attribute vector and its corresponding optimal segmentation threshold;

[0080] Determine whether the node after division is a leaf node. If not, use the node as the root node of the subtree and re-divide it. If so, determine whether all nodes have reached leaf nodes at this time. When all nodes have reached leaf nodes, the training process is completed.

[0081] Furthermore, in step S302, calculating the posterior probability of each data in the training sample specifically includes:

[0082] The category set contains several output categories, the training sample contains a feature set, and the feature set contains several feature attribute vectors. When the output category is Y k When the characteristic attribute vector Z j The posterior probability is:

[0083]

[0084] Where k = 1, 2, ..., m, m represents the total number of categories in the category set, Y k Indicates that the output category is the k-th category, β(k) represents the training accuracy of the first enterprise safety risk identification and assessment classification model for the k-th output category, j = 1, 2, ..., n, n is the total number of feature attribute vectors, k = t represents when the feature attribute vector Z j When the output category is equal to the true category t, k≠t represents the feature attribute vector Z j When the output category is not equal to the true category t.

[0085] S303: Build a second enterprise security risk identification and assessment classification model based on the support vector machine model, input the category set and training samples for training, obtain a second classification output result, and calculate the posterior probability of each data in the training sample.

[0086] It should be understood that the method of constructing the second enterprise security risk identification and assessment classification model based on the support vector machine model is an existing technology and will not be elaborated here.

[0087] In step S303, calculating the posterior probability of each data in the training sample specifically includes:

[0088] The Bayesian formula is used to calculate the posterior probability P(Y k |Z) SVM , when the output category is Y k When the characteristic attribute vector Z j The posterior probability is:

[0089]

[0090] Where k = 1, 2, ..., m, m represents the total number of categories in the category set, i = 1, 2, ..., m, Y k Indicates that the output category is the kth category, Y i Indicates that the output category is the i-th category, j = 1, 2, ..., n, and n is the total number of feature attribute vectors.

[0091] S304: Combine the first classification output result and the second classification output result to obtain a final classification output result, and combine them into a final classification model.

[0092] Step S304 specifically includes:

[0093] The posterior probability P(Y k |Z j ) CARTand the posterior probability P(Y k |Z j ) SVM Perform weighted synthesis to obtain:

[0094]

[0095] Among them, A CART represents the classification accuracy of the first enterprise safety risk identification and assessment classification model, A SVM Represents the classification accuracy of the second enterprise safety risk identification and assessment classification model.

[0096] S4, obtain the identification resolution data of the industrial Internet enterprises to be verified and their industrial Internet enterprise information, input them into the final classification model for classification, build a multivariate linear regression model, conduct security risk identification and assessment on the classified industrial Internet enterprises to be verified, and obtain the corresponding security risk identification and assessment report.

[0097] Furthermore, in step S4, a multivariate linear regression model is constructed to conduct security risk identification and assessment on the classified industrial Internet enterprises to be verified, specifically including:

[0098] Based on several risk identification and assessment categories and sample data training, a multivariate linear regression model y is constructed, and its expression is:

[0099] y=β0+β1x1+β2x2+…+β p x p

[0100] Among them, β i is the regression coefficient, x i is the risk identification and assessment category, i=1,2,…,p.

[0101] It should be understood that the method for safety risk identification and assessment is: determine the enterprise safety risk threshold Z based on the multivariate linear regression model, and then derive the conditional probability P of the enterprise safety risk; if the P value is lower than 0.5, it indicates that the probability of enterprise safety risk is relatively large, that is, the enterprise is judged to be a risk type that is about to form; if the P value is not lower than 0.5, it indicates that the enterprise has no safety risk and operates normally.

[0102] See also Figure 3 , this embodiment also provides an enterprise security risk identification and assessment system based on identity resolution data, the system comprising: an enterprise data acquisition module 10, an information database establishment module 20, a classification module 30 and a security risk identification and assessment module 40;

[0103] Enterprise data acquisition module 10, obtains historical industrial Internet enterprise identification resolution data and corresponding historical industrial Internet enterprise information;

[0104] An information database establishment module 20 establishes an information database, stores historical industrial Internet enterprise identification resolution data and corresponding historical industrial Internet enterprise information in the information database, and uses the data in the information database as sample data;

[0105] The classification module 30 constructs an enterprise security risk identification and assessment classification model based on the improved CART decision tree model and the support vector machine model, and inputs the sample data into the enterprise security risk identification and assessment classification model for training to obtain a final classification model;

[0106] The security risk identification and assessment module 40 obtains the identification resolution data of the industrial Internet enterprise to be verified and its industrial Internet enterprise information, inputs it into the final classification model for classification, constructs a multivariate linear regression model, conducts security risk identification and assessment on the classified industrial Internet enterprise to be verified, and obtains the corresponding security risk identification and assessment report.

[0107] It should be noted that the system embodiments described above are merely illustrative and do not limit the scope of protection of the present invention. In implementation and application, those skilled in the art may select some or all of the modules to achieve the purpose of the present embodiment according to actual needs, and no limitations are imposed here. Furthermore, the specific implementation of the enterprise security risk identification and assessment system based on identity resolution data in this embodiment is substantially consistent with the implementation of the above-mentioned method and will not be further elaborated here.

[0108] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for identifying and assessing enterprise security risks based on identity resolution data, characterized in that: The method comprises: S1, obtain historical industrial Internet enterprise identification resolution data and corresponding historical industrial Internet enterprise information; S2, establish an information database, store historical industrial Internet enterprise identification resolution data and corresponding historical industrial Internet enterprise information in the information database, and use all data in the information database as sample data; S3, constructing enterprise security risk identification and assessment classification models based on the improved CART decision tree model and support vector machine model respectively, and inputting sample data into the enterprise security risk identification and assessment classification model for training to obtain the final classification model; S4: Obtain the identification resolution data and industrial internet enterprise information of the industrial internet enterprise to be verified, input them into the final classification model for classification, construct a multivariate linear regression model, conduct security risk identification and assessment on the classified industrial internet enterprise to be verified, and obtain the corresponding security risk identification and assessment report; Step S1 specifically includes: the historical industrial Internet enterprise information includes corresponding enterprise legal litigation information, enterprise operating information and management information; Step S3 specifically includes: S301, normalizing several risk identification and assessment categories and feature information data to obtain a category set and a feature set respectively; S302, constructing a first enterprise safety risk identification and assessment classification model based on the CART decision tree model, taking the category set as the dependent variable and the feature set as the independent variable, and inputting the training samples into the first enterprise safety risk identification and assessment classification model in a recursive segmentation manner for training to obtain a first classification output result, and calculating the posterior probability of each data in the training sample; S303, constructing a second enterprise safety risk identification and assessment classification model based on the support vector machine model, inputting the category set and the training samples for training to obtain a second classification output result, and calculating the posterior probability of each data in the training sample; S304, combining the first classification output result and the second classification output result to obtain a final classification output result, and combining them into a final classification model; Step S304 specifically includes: converting the posterior probability P(Y k |Z j ) CART and the posterior probability P(Y k |Z j ) SVM Perform weighted synthesis to obtain: Among them, A CART represents the classification accuracy of the first enterprise safety risk identification and assessment classification model, A SVM represents the classification accuracy of the second enterprise safety risk identification and assessment classification model; In step S4, a multivariate linear regression model is constructed to conduct security risk identification and assessment on the classified industrial Internet enterprises to be verified, specifically including: Based on several risk identification and assessment categories and sample data training, a multivariate linear regression model y is constructed, and its expression is: y=β0+β1x1+β2x2+…+β p x p Among them, β i is the regression coefficient, x i is the risk identification and assessment category, i=1,2,…,p.

2. The enterprise security risk identification and assessment method based on identity resolution data according to claim 1, characterized in that: Step S2 specifically includes: Several risk identification and assessment categories are set up, and the TF-IDF method is used to extract feature information data related to several risk identification and assessment categories from historical industrial Internet enterprise information, and the feature information data corresponding to the historical industrial Internet identifier resolution data are stored one by one in the information database.

3. The enterprise security risk identification and assessment method based on identity resolution data according to claim 1, characterized in that: In step S302, the training samples are input into the first enterprise security risk identification and assessment classification model by recursive segmentation for training, which specifically includes: Create a root node, input the training samples into the root node, and calculate the Gini coefficient G of all training samples in the root node. The root node contains several feature attribute vectors. For each feature attribute vector in the root node, a segmentation threshold calculation is performed. The calculation process includes: specifying a feature attribute vector as a partition feature, setting several segmentation thresholds for the partition feature, dividing the training samples into two groups, and calculating the Gini coefficient G' of all training samples after the partition; calculating the Gini coefficient reduction ΔG before and after the partition, ΔG = GG'; calculating the Gini coefficient reduction corresponding to different segmentation thresholds of the partition feature, comparing the Gini coefficient reductions corresponding to different segmentation thresholds, and obtaining the maximum reduction G B And the corresponding segmentation threshold, and use it as the optimal segmentation threshold; Perform segmentation threshold calculation on all feature attribute vectors in the root node to obtain the optimal segmentation feature attribute vector and its corresponding optimal segmentation threshold, and segment the root node based on the optimal segmentation feature attribute vector and its corresponding optimal segmentation threshold; Determine whether the node after division is a leaf node. If not, use the node as the root node of the subtree and re-divide it. If so, determine whether all nodes have reached leaf nodes at this time. When all nodes have reached leaf nodes, the training process is completed.

4. The enterprise security risk identification and assessment method based on identity resolution data according to claim 1, characterized in that: In step S302, calculating the posterior probability of each data in the training sample specifically includes: The category set contains several output categories, the training sample contains a feature set, and the feature set contains several feature attribute vectors. When the output category is Y k When the characteristic attribute vector Z j The posterior probability is: Where k = 1, 2, ..., m, m represents the total number of categories in the category set, Y k Indicates that the output category is the k-th category, β(k) represents the training accuracy of the first enterprise safety risk identification and assessment classification model for the k-th output category, j = 1, 2, ..., n, n is the total number of feature attribute vectors, k = t represents when the feature attribute vector Z j When the output category is equal to the true category t, k≠t represents the feature attribute vector Z j When the output category of is not equal to the true category t.

5. The enterprise security risk identification and assessment method based on identity resolution data according to claim 4, characterized in that: In step S303, calculating the posterior probability of each data in the training sample specifically includes: The Bayesian formula is used to calculate the posterior probability P(Y k |Z) SVM , when the output category is Y k When the characteristic attribute vector Z j The posterior probability is: Where k = 1, 2, ..., m, m represents the total number of categories in the category set, i = 1, 2, ..., m, Y k Indicates that the output category is the kth category, Y i Indicates that the output category is the i-th category, j = 1, 2, ..., n, and n is the total number of feature attribute vectors.

6. An enterprise security risk identification and assessment system based on identity resolution data, characterized in that: The method according to any one of claims 1 to 5 is performed, wherein the system comprises: Enterprise data acquisition module, which obtains historical industrial Internet enterprise identification resolution data and corresponding historical industrial Internet enterprise information; An information database establishment module establishes an information database, stores historical industrial Internet enterprise identification resolution data and corresponding historical industrial Internet enterprise information in the information database, and uses the data in the information database as sample data; The classification module builds enterprise security risk identification and assessment classification models based on the improved CART decision tree model and the support vector machine model, and inputs sample data into the enterprise security risk identification and assessment classification model for training to obtain the final classification model; The security risk identification and assessment module obtains the identification resolution data and industrial Internet enterprise information of the industrial Internet enterprises to be verified, inputs them into the final classification model for classification, constructs a multivariate linear regression model, conducts security risk identification and assessment on the classified industrial Internet enterprises to be verified, and obtains the corresponding security risk identification and assessment report.

Images