Enterprise risk detection method and device based on enterprise credit big data knowledge graph

Abstract

The present invention discloses an enterprise risk detection method and device based on enterprise credit big data knowledge graph, wherein the method comprises: constructing an enterprise credit big data unified information model through dispersed data subdomain data; constructing a first enterprise credit big data domain ontology in a top-down manner based on the enterprise credit big data unified information model; and extracting entities and relationships from data in the enterprise credit big data domain through a bottom-up construction method, selecting high-quality new words to expand the scale of the first enterprise credit big data domain ontology, so as to construct a second enterprise credit big data domain ontology; based on the constructed ontology, constructing an enterprise credit big data knowledge graph using enterprise credit big data, acquiring features through the knowledge graph, inputting the acquired feature data into a trained risk control model to output classification results, and using them to classify enterprises. The present invention improves the accuracy of the knowledge graph ontology in the enterprise credit field and improves the performance of the risk control model.

Description

Technical Field

[0001] The present invention relates to the field of enterprise risk detection, and in particular to an enterprise risk detection method and device based on enterprise credit big data knowledge graph. Background Art

[0002] At present, in the enterprise risk detection method based on knowledge graph, the mainstream method is to extract the enterprise node attributes in the knowledge graph as basic attribute features, and extract the relationship between the enterprise and other enterprise entities in the knowledge graph as the association relationship features, and input the basic attribute features and association relationship features of the enterprise as the features of the subsequent risk control model. Some people extract the characteristic information of the enterprise in the network, including the number and proportion of defaulting enterprises in the first-order and second-order neighbor relationships in the network as the relationship features of the enterprise, and input the gradient boosting decision tree classification model in combination with the basic attribute features of the enterprise. Some people define three types of knowledge graphs related to enterprise risks based on business and data background. The knowledge graph network is respectively the upstream and downstream of the enterprise, investment and financing, and closely related knowledge graphs, and uses the community discovery algorithm to obtain the close relationship between enterprises. Some people use equity relations, personnel relations and other data to comprehensively mine enterprise associations and build enterprise credit knowledge graphs. Based on the graph, two models are built, namely the enterprise association analysis model and the enterprise group association risk model, to help identify enterprise risks in the entire credit process of commercial banks.

[0003] As mentioned above, in the current enterprise risk detection methods based on knowledge graphs, the features used in the methods are mainly divided into two categories. The first category is basic attribute features (mainly enterprise data in the financial and judicial fields), and the second category is association relationship features (reflecting the close relationship between enterprise entities in the knowledge graph and other enterprise entities).

[0004] However, due to the strong privacy of credit data, different industries cannot share credit data, and credit data faces challenges of incompleteness and information silos. The basis for building a corporate credit map is corporate credit data. Therefore, the existing corporate credit maps at this stage have problems such as missing information. The corporate entity attributes in the corporate credit map only come from the financial, judicial and other fields, which makes it difficult to fully represent the credit status of the company. The data dimension needs to be increased and the model effect needs to be improved. Summary of the invention

[0005] The present invention aims to solve one of the technical problems in the related art at least to a certain extent.

[0006] To this end, the purpose of the present invention is to solve the problem of missing information in the corporate credit map, improve the accuracy of prediction of defaulting enterprises, and propose an enterprise risk detection method based on the corporate credit big data knowledge map.

[0007] Another object of the present invention is to propose an enterprise risk detection device based on the enterprise credit big data knowledge graph.

[0008] To achieve the above objectives, the present invention proposes, on one hand, a method for detecting enterprise risk based on a knowledge graph of enterprise credit big data, including:

[0009] A unified information model of enterprise credit big data is obtained based on multiple scattered data subdomains; wherein the unified information model of enterprise credit big data includes a hierarchical enterprise information architecture and a hierarchical key personnel architecture; the relationship between key figures and enterprises is extracted through the enterprise information of the hierarchical key personnel architecture and the enterprise personnel information of the hierarchical enterprise information architecture to achieve cross-domain connection of enterprise credit big data; based on the unified information model of enterprise credit big data that achieves the cross-domain connection, a first enterprise credit big data domain ontology is constructed in a top-down manner; and in a bottom-up construction manner, entity extraction and relationship extraction are performed on the data in the enterprise credit big data domain, and high-quality new words are selected to expand the scale of the first enterprise credit big data domain ontology to construct a second enterprise credit big data domain ontology; based on the second enterprise credit big data domain ontology, an enterprise credit big data knowledge graph is constructed using enterprise credit big data and stored in a graph database; enterprise feature data is acquired using the enterprise credit big data knowledge graph, and the acquired enterprise feature data is input into a trained risk control model for calculation and classification and the classification results are output.

[0010] According to the enterprise risk detection method based on the enterprise credit big data knowledge graph of the embodiment of the present invention, the accuracy of the knowledge graph ontology in the field of enterprise credit is greatly improved through strict top-down concept definition restrictions and relationship restrictions, and the integration of bottom-up methods to expand the ontology scale, laying a solid foundation for the subsequent generation of high-quality knowledge graphs, and innovatively introducing the characteristics of enterprise R&D and innovation capabilities as the input of the risk control model, which improves the accuracy of the knowledge graph ontology in the field of enterprise credit and also improves the performance of the risk control model.

[0011] In addition, the enterprise risk detection method based on the enterprise credit big data knowledge graph according to the above embodiment of the present invention also includes:

[0012] Furthermore, the hierarchical enterprise information architecture of the unified information model of enterprise credit investigation big data includes: multiple sub-domains of enterprise basic information, enterprise personnel information, enterprise operating information, enterprise asset information, enterprise intellectual property information, enterprise financial information, enterprise equity information, judicial data, enterprise risk information and auxiliary reference information.

[0013] Furthermore, the bottom-up construction method is used to perform entity extraction and relationship extraction on the data in the enterprise credit big data field, and select high-quality new words to expand the scale of the first enterprise credit big data field ontology to construct the second enterprise credit big data field ontology, including: performing entity extraction and relationship extraction on the data in the enterprise credit big data field through a bottom-up construction method; based on the entity extraction and relationship extraction, identifying the named entities and relationship instances in the data, and performing quality judgment on the named entities and relationship instances that cannot be identified; determining the quality ranking based on the quality judgment, selecting high-quality new words and expanding the first enterprise credit big data field ontology to construct the second enterprise credit big data field ontology.

[0014] Furthermore, the enterprise characteristic data acquisition includes: obtaining the basic attribute characteristics, association relationship characteristics and R&D innovation capability characteristics of the enterprise; wherein, the basic attribute characteristics of the enterprise and the R&D innovation capability characteristics of the enterprise are obtained from the enterprise credit big data knowledge graph; and, enterprise relationship characteristics are extracted through four types of relationships, and network characteristics in the enterprise credit big data knowledge graph are extracted through the shortest path algorithm and the community discovery algorithm to obtain the association relationship characteristics of the enterprise; wherein, the four types of relationships include equity participation relationship, investment relationship, transaction relationship and litigation relationship.

[0015] Furthermore, the risk control model includes: data preprocessing, feature processing engineering and result classification.

[0016] Furthermore, the data preprocessing includes: preprocessing the acquired enterprise characteristic data, converting date data into character variables, then converting all character variables to obtain numerical data, and extracting IV value, WOE, efficiency and rate of the numerical data.

[0017] Furthermore, the formulas for the IV value, WOE, efficiency and rate are:

[0018]

[0019]

[0020]

[0021]

[0022] Among them, Good i and Bad i Indicates the number of non-defaulting enterprises and defaulting enterprises in each bin, Good T and BadT Represent the total number of non-defaulting enterprises and the number of defaulting enterprises respectively.

[0023] Furthermore, the feature processing project includes: deleting features with more than 50% missing values, features containing only unique values, features with correlation with other features higher than 60%, features with feature importance of 0.0 in the gradient enhancer, and low-importance features that do not contribute to the cumulative feature importance of 99% from the gradient enhancer.

[0024] Furthermore, the result classification includes: obtaining the enterprise feature data samples and enterprise labels; using the enterprise feature data samples and enterprise labels to supervise the training of the LightGBM classification model to obtain a trained LightGBM classification model; inputting the features processed by the feature processing project into the trained LightGBM classification model to perform calculation and classification to obtain classification results; wherein the classification results are divided into default and normal.

[0025] To achieve the above-mentioned purpose, the present invention proposes, on the other hand, an enterprise risk detection device based on enterprise credit big data knowledge graph, comprising:

[0026] An information acquisition module, used to obtain a unified information model of enterprise credit big data based on multiple dispersed data subdomains; wherein the unified information model of enterprise credit big data includes a hierarchical enterprise information architecture and a hierarchical key personnel architecture;

[0027] A relationship connection module, used to extract the relationship between key persons and enterprises through the enterprise information of the hierarchical key personnel architecture and the enterprise personnel information of the hierarchical enterprise information architecture, so as to realize the cross-domain connection of enterprise credit investigation big data;

[0028] An ontology construction module is used to determine the enterprise credit big data domain and construct a first enterprise credit big data domain ontology in a top-down manner based on the unified information model of the enterprise credit big data that realizes the cross-domain connection; and to extract entities and relationships from the data in the enterprise credit big data domain, select high-quality new words and expand the scale of the first enterprise credit big data domain ontology in a bottom-up manner to construct a second enterprise credit big data domain ontology;

[0029] A graph construction module, for constructing a knowledge graph of enterprise credit big data based on the second enterprise credit big data domain ontology and using the enterprise credit big data, and storing it in a graph database;

[0030] The calculation and classification module is used to obtain enterprise characteristic data using the enterprise credit big data knowledge graph, input the obtained enterprise characteristic data into the trained risk control model for calculation and classification, and output the classification results.

[0031] The enterprise risk detection device based on the enterprise credit big data knowledge graph of the embodiment of the present invention greatly improves the accuracy of the knowledge graph ontology in the field of enterprise credit through strict top-down concept definition restrictions and relationship restrictions, and integrates the bottom-up approach to expand the ontology scale, laying a solid foundation for the subsequent generation of high-quality knowledge graphs, and innovatively introduces the characteristics of enterprise R&D and innovation capabilities as the input of the risk control model, which improves the accuracy of the knowledge graph ontology in the field of enterprise credit and also improves the performance of the risk control model.

[0032] Beneficial effects of the present invention:

[0033] (1) The enterprise credit big data knowledge graph construction technology proposed in this invention solves the problem of information missing in existing enterprise credit graphs;

[0034] (2) The risk control model proposed in the present invention, which introduces the characteristics of enterprise R&D and innovation capabilities, outperforms the traditional risk control model based on enterprise credit knowledge graph, making it easier to identify defaulting enterprises in advance and reduce risks.

[0035] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0037] Figure 1 Schematic diagram of enterprise risk detection architecture based on enterprise credit big data knowledge graph according to an embodiment of the present invention;

[0038] Figure 2 Flow chart of a method for detecting enterprise risk based on enterprise credit big data knowledge graph according to an embodiment of the present invention;

[0039] Figure 3 A schematic diagram of the architecture of hierarchical enterprise information of a unified information model of enterprise credit investigation big data according to an embodiment of the present invention;

[0040] 4(a) and 4(b) are schematic diagrams of a secondary architecture of enterprise financial information of a unified information model of enterprise credit investigation big data according to an embodiment of the present invention;

[0041] Figure 5 A schematic diagram of the architecture of hierarchical key personnel of the unified information model of enterprise credit investigation big data according to an embodiment of the present invention;

[0042] Figure 6A schematic diagram of the process of the enterprise credit investigation big data knowledge graph ontology with top-down and bottom-up methods as supplementary methods according to an embodiment of the present invention;

[0043] Figure 7 A schematic diagram of a flow chart of a risk control model designed according to an embodiment of the present invention;

[0044] Figure 8 Schematic diagram of the structure of an enterprise risk detection device based on enterprise credit big data knowledge graph according to an embodiment of the present invention. DETAILED DESCRIPTION

[0045] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0046] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments 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 creative work should fall within the scope of protection of the present invention.

[0047] The following describes the enterprise risk detection method and device based on the enterprise credit big data knowledge graph proposed in accordance with an embodiment of the present invention with reference to the accompanying drawings. First, the enterprise risk detection method based on the enterprise credit big data knowledge graph proposed in accordance with an embodiment of the present invention will be described with reference to the accompanying drawings.

[0048] The enterprise risk detection method based on the enterprise credit big data knowledge graph of the embodiment of the present invention has the following overall process: Figure 1 As shown. Based on the original risk control model, the embodiment of the present invention adds the enterprise R&D innovation capability feature to increase the level and dimension of the feature. In addition to the addition of features, the enterprise risk control model in the enterprise risk monitoring method of the embodiment of the present invention uses LightGBM, because LightGBM actually adopts a gradient boosting framework based on the decision tree algorithm. Therefore, LightGBM can also obtain the importance of features to the model during the training process, which is used to evaluate the influence of different features on whether the enterprise defaults.

[0049] Figure 2 It is a flow chart of an enterprise risk detection method based on enterprise credit big data knowledge graph according to an embodiment of the present invention.

[0050] like Figure 2 As shown, the enterprise risk detection method based on the enterprise credit big data knowledge graph includes the following steps:

[0051] Step S1, obtaining a unified information model of enterprise credit big data based on multiple dispersed data subdomains; wherein the unified information model of enterprise credit big data includes a hierarchical enterprise information architecture and a hierarchical key personnel architecture.

[0052] Specifically, the embodiment of the present invention utilizes expert knowledge, studies a series of relevant enterprise credit data standard systems, investigates papers and patents related to enterprise credit knowledge graphs, extracts an "enterprise-key person" joint framework from existing scattered data sub-domains such as government affairs, industry and commerce, justice, and public opinion, and designs a hierarchical enterprise information architecture and key personnel architecture for enterprise credit big data scenarios. The relationships between various entities are used as connecting edges to achieve global entity association of enterprise credit big data.

[0053] The hierarchical enterprise information architecture of the unified information model of enterprise credit big data is supported by 10 information subdomains, including enterprise basic information, enterprise personnel information, enterprise operation information, enterprise asset information, enterprise intellectual property information, enterprise financial information, enterprise equity information, judicial data, enterprise risk information, and auxiliary reference. The hierarchical enterprise information architecture of the unified information model of enterprise credit big data is as follows: Figure 3 shown.

[0054] As shown in Figure 4(a) and Figure 4(b), taking enterprise financial data as an example, a fine-grained view of the enterprise information architecture is presented.

[0055] Step S2, extracting the relationship between key persons and enterprises through the enterprise information of the hierarchical key personnel architecture and the enterprise personnel information of the hierarchical enterprise information architecture, so as to realize the cross-domain connection of enterprise credit big data.

[0056] It is understandable that the view of the hierarchical key personnel architecture of the unified information model of enterprise credit big data is composed of four information sub-domains: basic information, work information, social relations, and historical risks. With the help of the enterprise information in the key personnel architecture and the enterprise personnel information in the enterprise information architecture, the association barriers between the architecture and the enterprise architecture can be broken through to form a mapping relationship between entity objects, thereby realizing the hierarchical and associated nature of the "enterprise-key personnel" of credit big data, and preliminarily solving the problem of cross-domain connection difficulties of enterprise credit big data.

[0057] like Figure 5 The figure shows a view of the hierarchical key personnel architecture of the unified information model of enterprise credit reporting big data.

[0058] The hierarchical enterprise information architecture and key person information architecture for the enterprise credit big data scenario hope to achieve the global entity association of enterprise credit big data in a "dual-core" manner, and it is necessary to define the relationship between entities. The entity relationship settings are shown in Table 1.

[0059] Table 1: Entity relationship design table

[0060]

[0061]

[0062] Step S3, based on the unified information model of corporate credit big data that realizes cross-domain connection, a first corporate credit big data domain ontology is constructed using a top-down approach; and through a bottom-up construction method, entity extraction and relationship extraction are performed on the data in the corporate credit big data field, and high-quality new words are selected to expand the scale of the first corporate credit big data field ontology to construct a second corporate credit big data field ontology.

[0063] Specifically, the first step in building a high-quality corporate credit big data knowledge graph is to define an accurate and clear knowledge model (schema), that is, to give an ontology that describes the basic cognitive framework in the field of corporate credit. However, the traditional construction method that only focuses on the "top-down method" is highly dependent on domain experts. The "bottom-up method" and massive, multi-source, and heterogeneous data are huge challenges for bottom-up construction of the ontology and subsequent knowledge fusion.

[0064] Based on the defects of a single knowledge graph ontology construction method, a corporate credit big data knowledge graph ontology construction method based on "top-down as the main and bottom-up as the auxiliary" is used. The concepts and relationships are constrained by the top-down method, and the scale of the ontology is expanded by integrating the bottom-up method. This greatly improves the accuracy and sophistication of the knowledge graph ontology, laying a solid foundation for the subsequent generation of high-quality knowledge graphs. The specific construction process is as follows: Figure 6 shown.

[0065] To form a domain ontology in a top-down manner, it is necessary to mine the knowledge of the domain knowledge base and listen to the advice of domain experts to construct the domain ontology. The domain knowledge base includes but is not limited to the Internet knowledge base, encyclopedia websites, industry authoritative guides, metadata national standards and relational databases in the field. For example, the "enterprise-key personnel system" mentioned in the embodiment of the present invention, which is based on the hierarchical enterprise information architecture and the key personnel information architecture, orderly organizes the massive data resources in the field of corporate credit big data. From this label system, high-quality concepts and attributes in the field of corporate credit, as well as the relationships between concepts, can be screened out to construct the prototype of the domain ontology.

[0066] The domain ontology created by the top-down method can already guide the construction of the enterprise credit big data knowledge graph instance library. However, due to the growth of the scale of data resources in the enterprise credit field, the enterprise credit domain ontology model constructed only by the top-down method is limited by scale and cannot meet the needs of subsequent knowledge graph construction technology (such as knowledge extraction and knowledge fusion). If the multi-source, massive, and heterogeneous data resources in the enterprise credit field can be sorted, utilized, and improved, it can generate a huge data driving force for the construction of the knowledge graph in the enterprise credit field. Therefore, the bottom-up construction method is also an important part of the enterprise credit big data knowledge graph ontology and data construction. The bottom-up auxiliary line construction process first extracts entities and relationships from the data in the enterprise credit field, extracts the named entities and relationship instances in the data, and makes quality judgments for the named entities and relationship instances that cannot be identified. Credit experts determine whether the new words with the highest quality ranking are high-quality phrases and expand the current enterprise credit field ontology structure.

[0067] Step S4, based on the second enterprise credit big data domain ontology, use the enterprise credit big data to construct an enterprise credit big data knowledge graph and store it in a graph database.

[0068] Specifically, after constructing the knowledge graph ontology of corporate credit big data using the above method, the existing corporate credit big data is used to construct the knowledge graph and store it in the Neo4j graph database to provide a data basis for subsequent risk control models.

[0069] Step S5, using the corporate credit big data knowledge graph to obtain corporate characteristic data, inputting the obtained corporate characteristic data into the trained risk control model for calculation and classification and outputting the classification results.

[0070] Specifically, in the enterprise risk control model module, the basic attribute characteristics, correlation characteristics, and R&D innovation capability characteristics of the enterprise are obtained from the enterprise credit big data knowledge graph, processed, and used as the input of the risk control model to perform supervised training on the LightGBM classification model. By introducing the enterprise R&D innovation capability characteristics, the performance of the risk control model is improved. The processing flow of the embodiment of the present invention is as follows Figure 7 As shown, including:

[0071] (1) Data acquisition module:

[0072] In the enterprise credit big data knowledge graph, the enterprise's basic attribute capability characteristics and R&D innovation capability characteristics exist in the form of enterprise node attributes, which can be directly exported from the Neo4j graph database. The enterprise association relationship characteristics are to reflect the closeness of the enterprise entity and the defaulting enterprise entity. Due to the variety of node and edge types in heterogeneous networks, the characteristics and difficulty of extracting graphs are increased. Therefore, the proposal limits the enterprise credit big data knowledge graph to a homogeneous network, restricts the nodes at both ends of the relationship to only be enterprises, and folds and reduces the person nodes to reduce the interference of people on the network, ensuring that each relationship is between enterprises. Combining existing data and traditional cognitive logic, four types of high-risk enterprise relationships are retained: equity participation relationship, investment relationship, transaction relationship, and litigation relationship. Based on these four types of relationships, enterprise relationship features are extracted, and the method of extracting network features in the knowledge graph is the shortest path algorithm and the community discovery algorithm.

[0073] The extracted network features are shown in Table 2:

[0074] Table 2: Enterprise relationship table

[0075]

[0076] The characteristics of R&D innovation capabilities are shown in Table 3:

[0077] Table 3: R&D and innovation capabilities

[0078]

[0079] (2) Data preprocessing module:

[0080] The IV value (Information Value), WOE, efficiency, and rate extracted from non-data type data using the credit score card are subsequently processed as new features added to the model.

[0081] Enterprise data contains many attributes in pure string format, such as enterprise type, industry category and other specific length code data. It also contains date data such as establishment date and approval date. For date data, first convert it, convert it into numerical data in seconds, and then convert it into character format. Then convert all character variables into numerical data, and extract their IV value (Information Value), WOE, efficiency, and rate.

[0082] The formulas for WOE, IV, Efficiency, and rate are as follows:

[0083]

[0084]

[0085]

[0086]

[0087] Among themGood i and Bad i Indicates the number of non-defaulting companies and defaulting companies in each bin. T and Bad T Represent the total number of non-defaulting enterprises and the number of defaulting enterprises respectively.

[0088] (3) Feature Engineering Module:

[0089] In the feature engineering phase, in order to deal with the problems of a large number of missing values ​​in the original data and too high correlation between features, the features need to be processed first. The main steps are to delete features with more than 50% missing values, features with only unique values, features with correlations with other features higher than 60%, features with feature importance of 0.0 in the gradient booster (GBM), and low-importance features that do not contribute to the cumulative feature importance of 99% from GBM.

[0090] (4) Classification module:

[0091] This module uses the LightGBM algorithm. The features processed by the feature engineering module are input into the model to obtain the classification results. The results are classified into two categories: default and normal. Since LightGBM actually uses a gradient boosting framework based on the decision tree algorithm, LightGBM can obtain the importance of features to the model during the training process. The importance of features can be used to evaluate the impact of different features on whether the company defaults.

[0092] Through the above steps, through strict top-down concept definition restrictions and relationship restrictions, and integrating the bottom-up approach to expand the ontology scale, the accuracy of the knowledge graph ontology in the field of corporate credit reporting has been greatly improved, laying a solid foundation for the subsequent generation of high-quality knowledge graphs. In addition, the characteristics of corporate R&D and innovation capabilities have been innovatively introduced as the input of the risk control model, which has improved the accuracy of the knowledge graph ontology in the field of corporate credit reporting and the performance of the risk control model.

[0093] In order to implement the above embodiment, Figure 8 As shown, this embodiment also provides an enterprise risk detection device 10 based on the enterprise credit big data knowledge graph, and the device 10 includes: an information acquisition module 100, a relationship connection module 200, an ontology construction module 300, a graph construction module 400, and a calculation classification module 500.

[0094] The information acquisition module 100 is used to obtain a unified information model of enterprise credit investigation big data based on multiple dispersed data subdomains; wherein the unified information model of enterprise credit investigation big data includes a hierarchical enterprise information architecture and a hierarchical key personnel architecture;

[0095] The relationship connection module 200 is used to extract the relationship between key persons and enterprises through the enterprise information of the hierarchical key personnel architecture and the enterprise personnel information of the hierarchical enterprise information architecture, so as to realize the cross-domain connection of the enterprise credit investigation big data;

[0096] The ontology construction module 300 is used to determine the enterprise credit big data domain and construct a first enterprise credit big data domain ontology in a top-down manner based on the unified information model of enterprise credit big data that realizes cross-domain connection; and to extract entities and relationships from the data in the enterprise credit big data domain, select high-quality new words and expand the scale of the first enterprise credit big data domain ontology through a bottom-up construction method, so as to construct a second enterprise credit big data domain ontology;

[0097] A graph construction module 400 is used to construct a knowledge graph of enterprise credit investigation big data based on the second enterprise credit investigation big data domain ontology and store it in a graph database using enterprise credit investigation big data;

[0098] The calculation and classification module 500 is used to obtain enterprise characteristic data using the enterprise credit big data knowledge graph, input the obtained enterprise characteristic data into the trained risk control model for calculation and classification, and output the classification results.

[0099] According to the enterprise risk detection device based on the enterprise credit big data knowledge graph of the embodiment of the present invention, the accuracy of the knowledge graph ontology in the field of enterprise credit is greatly improved through strict top-down concept definition restrictions and relationship restrictions, and the integration of bottom-up methods to expand the ontology scale, laying a solid foundation for the subsequent generation of high-quality knowledge graphs, and innovatively introducing the characteristics of enterprise R&D and innovation capabilities as the input of the risk control model, which improves the accuracy of the knowledge graph ontology in the field of enterprise credit and also improves the performance of the risk control model.

[0100] It should be noted that the aforementioned explanation of the embodiment of the enterprise risk detection method based on the enterprise credit big data knowledge graph is also applicable to the enterprise risk detection device based on the enterprise credit big data knowledge graph of this embodiment, and will not be repeated here.

[0101] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0102] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0103] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A method for detecting enterprise risk based on enterprise credit big data knowledge graph, characterized in that: The following steps are involved: Obtaining a unified information model of enterprise credit big data based on multiple dispersed data subdomains; wherein the unified information model of enterprise credit big data includes a hierarchical enterprise information architecture and a hierarchical key personnel architecture; Through the enterprise information of the hierarchical key personnel architecture and the enterprise personnel information of the hierarchical enterprise information architecture, the relationship between the key person and the enterprise is extracted to achieve cross-domain connection of enterprise credit big data; Based on the unified information model of the enterprise credit big data that realizes the cross-domain connection, a first enterprise credit big data domain ontology is constructed in a top-down manner; and through a bottom-up construction method, entity extraction and relationship extraction are performed on the data in the enterprise credit big data domain, and high-quality new words are selected to expand the scale of the first enterprise credit big data domain ontology to construct a second enterprise credit big data domain ontology; Based on the second corporate credit big data domain ontology, using the corporate credit big data to construct a corporate credit big data knowledge graph and store it in a graph database; Using the enterprise credit big data knowledge graph to acquire enterprise characteristic data, inputting the acquired enterprise characteristic data into the trained risk control model for calculation and classification and outputting the classification results; The enterprise characteristic data acquisition includes: acquiring the basic attribute characteristics, relationship characteristics and R&D innovation capability characteristics of the enterprise; wherein, Obtaining the basic attribute characteristics of the enterprise and the R&D and innovation capability characteristics of the enterprise from the enterprise credit big data knowledge graph; and extracting enterprise relationship characteristics through four types of relationships, and extracting network characteristics in the enterprise credit big data knowledge graph through the shortest path algorithm and the community discovery algorithm to obtain the enterprise's association relationship characteristics; wherein the four types of relationships include equity participation relationship, investment relationship, transaction relationship and litigation relationship; Feature processing engineering, including: Remove features with more than 50% missing values, features containing only unique values, features with correlation with other features higher than 60%, features with feature importance of 0.0 in the gradient booster, and low importance features that do not contribute to the cumulative feature importance of 99% from the gradient booster.

2. The method according to claim 1, characterized in that: The hierarchical enterprise information architecture of the unified information model of enterprise credit investigation big data includes: Various sub-domains of enterprise basic information, enterprise personnel information, enterprise management information, enterprise asset information, enterprise intellectual property information, enterprise financial information, enterprise equity information, judicial data, enterprise risk information and auxiliary reference information.

3. The method according to claim 1, characterized in that The bottom-up construction method is used to extract entities and relationships from the data in the enterprise credit investigation big data domain, and select high-quality new words to expand the scale of the first enterprise credit investigation big data domain ontology to construct the second enterprise credit investigation big data domain ontology, including: By building from the bottom up, extracting entities and relationships from the data in the enterprise credit big data field; Based on the entity extraction and the relationship extraction, the named entities and relationship instances in the data are identified, and quality judgment is performed on the named entities and relationship instances that cannot be identified; A quality ranking is determined based on the quality determination, high-quality new words are selected and the first corporate credit investigation big data domain ontology is expanded to construct the second corporate credit investigation big data domain ontology.

4. The method according to claim 1, characterized in that: The risk control model includes: data preprocessing, feature processing engineering and result classification.

5. The method according to claim 4, characterized in that The data preprocessing includes: The acquired enterprise characteristic data is preprocessed, the date data is converted into character variables, and then all character variables are converted to obtain numerical data, and IV value, WOE, efficiency and rate of the numerical data are extracted.

6. The method according to claim 5, characterized in that The formulas for IV value, WOE, efficiency and rate are: （1） （2） （3） （4） in, and It means counting the number of non-defaulting enterprises and defaulting enterprises in each bin. and Represent the total number of non-defaulting enterprises and the number of defaulting enterprises respectively.

7. The method according to claim 1, characterized in that The results are classified as follows: Obtaining the enterprise feature data sample and enterprise label; Using the enterprise feature data samples and enterprise labels to supervise the training of the LightGBM classification model, a trained LightGBM classification model is obtained; The features processed by the feature processing project are input into the trained LightGBM classification model to perform calculation and classification to obtain classification results; wherein the classification results are divided into default and normal.

8. An enterprise risk detection device based on enterprise credit big data knowledge graph, characterized in that: include: An information acquisition module, used to obtain a unified information model of enterprise credit big data based on multiple dispersed data subdomains; wherein the unified information model of enterprise credit big data includes a hierarchical enterprise information architecture and a hierarchical key personnel architecture; A relationship connection module, used to extract the relationship between key persons and enterprises through the enterprise information of the hierarchical key personnel architecture and the enterprise personnel information of the hierarchical enterprise information architecture, so as to realize the cross-domain connection of enterprise credit investigation big data; An ontology construction module is used to determine the enterprise credit big data domain and construct a first enterprise credit big data domain ontology in a top-down manner based on the unified information model of the enterprise credit big data that realizes the cross-domain connection; and to extract entities and relationships from the data in the enterprise credit big data domain, select high-quality new words and expand the scale of the first enterprise credit big data domain ontology in a bottom-up manner to construct a second enterprise credit big data domain ontology; A graph construction module, for constructing a knowledge graph of enterprise credit big data based on the second enterprise credit big data domain ontology and using the enterprise credit big data, and storing it in a graph database; A calculation and classification module, which is used to obtain enterprise characteristic data by using the enterprise credit big data knowledge graph, input the obtained enterprise characteristic data into the trained risk control model for calculation and classification, and output the classification result; The enterprise characteristic data acquisition includes: acquiring the basic attribute characteristics, relationship characteristics and R&D innovation capability characteristics of the enterprise; wherein, Obtaining the basic attribute characteristics of the enterprise and the R&D and innovation capability characteristics of the enterprise from the enterprise credit big data knowledge graph; and extracting enterprise relationship characteristics through four types of relationships, and extracting network characteristics in the enterprise credit big data knowledge graph through the shortest path algorithm and the community discovery algorithm to obtain the enterprise's association relationship characteristics; wherein the four types of relationships include equity participation relationship, investment relationship, transaction relationship and litigation relationship; Feature processing engineering, including: Remove features with more than 50% missing values, features containing only unique values, features with correlation with other features higher than 60%, features with feature importance of 0.0 in the gradient booster, and low importance features that do not contribute to the cumulative feature importance of 99% from the gradient booster.

Images