CN114120091A

CN114120091A - Group pedigree identification method, device, equipment and medium

Info

Publication number: CN114120091A
Application number: CN202111437275.8A
Authority: CN
Inventors: 胡逸天; 黄珊珊; 何子龙; 楼华
Original assignee: OneConnect Financial Technology Co Ltd Shanghai
Current assignee: OneConnect Financial Technology Co Ltd Shanghai
Priority date: 2021-11-30
Filing date: 2021-11-30
Publication date: 2022-03-01

Abstract

The invention relates to the field of data processing, and provides a group pedigree identification method, a device, equipment and a medium, wherein the method comprises the following steps: acquiring a group pedigree image to be identified, wherein the group pedigree image comprises a topological relation among enterprises; identifying each enterprise and corresponding topological relation in the pedigree image of the group one by one from bottom to top, and obtaining controlled enterprises and corresponding highest control parties in the pedigree of the group according to the topological relation among the enterprises, wherein at least one highest control party is provided; performing primary processing along the stock right control path by taking the highest control party as a starting point in the group pedigree image to obtain a control coefficient between the highest control party and a controlled enterprise; performing secondary processing according to the control coefficient by combining the stock holding ratio between the highest control party and the controlled enterprise, and determining the highest control party with the highest stock holding ratio corresponding to the controlled enterprise in the group pedigree image as the final control party, so that the group pedigree identification efficiency is improved; and the accuracy rate of group pedigree identification is also improved.

Description

Group pedigree identification method, device, equipment and medium

Technical Field

The invention relates to the field of data processing, and provides a method, a device, equipment and a medium for identifying a group pedigree.

Background

Various multinational enterprises, multiindustrial enterprises and transregional enterprises are increasing, and the status and the proportion of group customers in commercial banks are increasing. Compared with a single enterprise, the group customers have more powerful economic strength, better integrity than that of the single enterprise, diversified requirements and great benefits for banks. But the internal associated transactions of group customers are increasingly complex, and bank-enterprise information asymmetry brought by cross-industry and cross-regional operation increases the potential risk of bank credit assets. Once these risks are exposed, domino effect will be generated, many debt banks will be involved, which not only has an influence on the banking enterprises, but also has a huge impact on the financial systems of the whole country. Therefore, the commercial bank must effectively identify the group customers, control and prevent the credit risk of the group customers, and strengthen the risk management of the credit business of the group customers, so as to promote the healthy and robust development of each business.

However, the existing identification mode related to the pedigree of the group is not good, and workers are required to assist in identification, so that on one hand, the workload of the workers is increased; on the other hand, because each enterprise is sometimes complicated in the group pedigree, the final control party in the group pedigree cannot be accurately identified.

Disclosure of Invention

The invention provides a group pedigree identification method, a device, equipment and a medium, which mainly aim at identifying control parties in a group pedigree in a bidirectional identification mode, identifying the control parties by utilizing a stock right relation and sequentially penetrating upwards according to a principle from bottom to top, and identifying the highest control party at a vertex; penetrating and identifying layer by layer along the principle of the highest control party from top to bottom to obtain the control coefficients of all enterprises of the group pedigree; and determining the stock control proportion of each company in the group pedigree in a top-down identification mode, judging the enterprise attribution to determine the final control party of the enterprise, and obtaining the association relationship of each enterprise in the group pedigree.

In order to achieve the above object, the present invention provides a method for identifying a group pedigree, comprising:

acquiring a group pedigree image to be identified, wherein the group pedigree image comprises a topological relation among enterprises;

identifying each enterprise and corresponding topological relation in the pedigree image of the group one by one from bottom to top, and obtaining controlled enterprises in the pedigree of the group and the highest control party for controlling the controlled enterprises according to the topological relation among the enterprises, wherein the number of the highest control parties is at least one;

processing once along an equity control path by taking the highest control party as a starting point in the group pedigree image to obtain a control coefficient between the highest control party and a controlled enterprise;

and performing secondary processing according to the control coefficient by combining the stock holding ratio between the highest control party and the controlled enterprise, and determining the highest control party with the highest stock holding ratio corresponding to the controlled enterprise in the group pedigree image as a final control party.

Optionally, before identifying each enterprise and the corresponding topological relation in the family pedigree image one by one from the bottom up, the method further includes:

converting the obtained group pedigree image to be identified into a group pedigree image with a preset specification size;

performing normalization processing on the group pedigree image with a preset specification to obtain a group pedigree image with uniform pixel colors;

and performing data enhancement processing on the group pedigree image with uniform pixel colors to obtain a preprocessed group pedigree image, wherein the data enhancement processing is performed on the group pedigree image with uniform pixel colors.

Optionally, the identifying, from bottom to top, each enterprise and the corresponding topological relation in the group pedigree image one by one, and obtaining a controlled enterprise in the group pedigree and a highest controlling party controlling the controlled enterprise according to the topological relation between the enterprises, includes:

identifying each enterprise and corresponding topological relation in the group pedigree image;

determining a controlled enterprise according to an enterprise node corresponding to a controlled object in a topological relation formed by enterprise nodes in the group pedigree image from bottom to top;

identifying the controlled enterprise from bottom to top, and determining a last-level controller of the controlled enterprise, wherein when the controlled enterprise has one or more corresponding enterprise nodes for controlling stock, the enterprise node with the highest stock control right value is determined as the last-level controller of the controlled enterprise;

carrying out layer-by-layer penetration identification according to the stock right control path of the topological relation, and obtaining the superior control party corresponding to each layer of enterprise node one by one;

and when detecting that the enterprise node of a certain layer does not have a previous-level control party, determining the enterprise node of the certain layer as a highest control party, wherein at least one highest control party corresponding to the controlled enterprise is in the topological relation.

Optionally, the performing, in the group pedigree image, a first processing along an equity control path with the highest controlling party as a starting point to obtain a control coefficient between the highest controlling party and a controlled enterprise includes:

sequentially processing the next controlled enterprise layer from top to bottom along the equity control path by taking the highest control party as a starting point in the group pedigree image to obtain a control coefficient between the highest control party and the controlled enterprise;

when detecting that the share of the current controlled enterprise by the previous-level control party reaches a first preset proportion, the previous-level control party reaching the first preset proportion and the controlled enterprise form actual control, and the generated control coefficient is 1;

and when detecting that the share of the current controlled enterprise by the previous-level control party does not reach the first preset proportion, the previous-level control party and the controlled enterprise which do not reach the first preset proportion do not form actual control, and the generated control coefficient is 0.

Optionally, the performing, according to the control coefficient, a secondary processing in combination with the stock holding ratio between the highest control party and the controlled enterprise, and determining the highest control party with the largest stock holding ratio corresponding to the controlled enterprise in the group pedigree image as a final control party includes:

dividing each enterprise to form stockholder groups according to an equity control path by taking the highest control party as a starting point in the group pedigree image;

determining the stock holding ratio between the highest control party and the controlled enterprise in each shareholder group from top to bottom according to the stock right control path;

and determining the highest control party corresponding to the stockholder group with the highest stock holding ratio between the highest control party and the controlled enterprise as the final control party.

Optionally, the group pedigree identification method further includes:

detecting whether a control enterprise at the upper level corresponding to the controlled enterprise has an action actor or not;

when detecting that a consistent action person exists in a previous-level control enterprise corresponding to a controlled enterprise, judging whether the share occupation of the previous-level control enterprise exceeds a first preset proportion; if the upper-level control enterprises do not exceed the first preset proportion, determining the related enterprises as a common control relation according to the relation corresponding to the consistent actor;

detecting the highest control party belonging to a common control relation in the group pedigree image;

when the situation that the highest control party of the group pedigree image belonging to the common control relationship is any one of a limited partner enterprise, a vacant shell enterprise or a stock holding platform is detected, enterprise nodes occupying the largest stock holding proportion among the controlled enterprises are determined by screening layer by layer from top to bottom according to the current situation, and the current enterprise nodes are determined as the actual control parties in the group pedigree image.

Optionally, the group pedigree identification method further includes:

identifying a full-scale pedigree between a controlled enterprise and a final control party according to the pedigree image of the group;

fusing enterprise relationships and enterprise nodes determined in the full-scale pedigree by using priority ranking;

and performing local calculation based on the group pedigree image, updating the determined enterprise relation and enterprise nodes to obtain an incremental pedigree formed by penetrating the group pedigree image, and realizing small-batch updating of the group pedigree.

In addition, to achieve the above object, the present invention provides a group pedigree identification apparatus, comprising:

the system comprises an acquisition module, a processing module and a processing module, wherein the acquisition module is used for acquiring a group pedigree image to be identified, and the group pedigree image comprises a topological relation among enterprises;

the identification module is used for identifying each enterprise and the corresponding topological relation in the pedigree image of the group one by one from bottom to top, and obtaining controlled enterprises in the pedigree of the group and the highest control party for controlling the controlled enterprises according to the topological relation among the enterprises, wherein the number of the highest control parties is at least one;

the control coefficient determining module is used for carrying out primary processing along the right-of-stock control path by taking the highest control party as a starting point in the group pedigree image to obtain a control coefficient between the highest control party and a controlled enterprise;

and the final control party determining module is used for performing secondary processing according to the control coefficient by combining the stock holding ratio between the highest control party and the controlled enterprise, and determining the highest control party with the highest stock holding ratio corresponding to the controlled enterprise in the group pedigree image as the final control party.

Furthermore, to achieve the above object, the present invention further provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the method according to any one of the above embodiments.

Furthermore, to achieve the above object, the present invention further provides a computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, implements the steps of the method according to any one of the above embodiments.

The group pedigree identification method, the device, the equipment and the medium provided by the invention identify the control party in the group pedigree in a bidirectional identification mode, identify the control party by utilizing the stock right relation and sequentially penetrating upwards through the control party according to the principle from bottom to top, and identify the highest control party at the top point; penetrating and identifying layer by layer along the principle of the highest control party from top to bottom to obtain the control coefficients of all enterprises of the group pedigree; determining the stock control proportion of each company in the group pedigree in a top-down identification mode, judging the enterprise attribution to determine the final control party of the enterprise, and obtaining the incidence relation of each enterprise in the group pedigree; on one hand, the group pedigree is automatically identified, so that the identification efficiency is improved; on the other hand, the two-way identification mode is adopted to respectively carry out penetration identification, so that the identification accuracy is improved.

Drawings

FIG. 1 is a schematic flow chart illustrating a method for identifying a pedigree of a group according to an embodiment of the present invention;

FIG. 2 is a schematic diagram illustrating a flow of identifying a highest controlling party in the method for identifying a pedigree of a group according to an embodiment of the present invention;

FIG. 3 is a schematic diagram illustrating a flow of determining control coefficients in the method for identifying a pedigree of a clique according to an embodiment of the present invention;

FIG. 4 is a schematic diagram illustrating a process of determining a final controller in the method for identifying a pedigree of a group according to an embodiment of the present invention;

FIG. 5 is a schematic diagram illustrating an actual control determination process in the group pedigree identification method according to an embodiment of the present invention;

fig. 6 is a schematic diagram illustrating a flow of updating a small batch in the group pedigree identification method according to an embodiment of the present invention;

fig. 7 is a schematic structural diagram of a group pedigree identification apparatus according to an embodiment of the present invention;

FIG. 8 is a schematic diagram of an architecture for a computing device according to an embodiment of the invention;

FIG. 9 is a group pedigree image provided in an embodiment of the present invention;

FIG. 10 is another clique pedigree image provided in an embodiment of the invention.

The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.

Detailed Description

It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

To facilitate understanding of the present application, the concepts related to the present application will be explained first.

Corporate lineage identification is a critical task in the financial industry, particularly in the risk management sector and of great importance. The related processes include group unified credit management in pre-loan, group concentration risk management after loan, internal control related transaction and the like. Each process needs service personnel to manually confirm the affiliation group, the workload of the special personnel is increased by wearing a complex equity structure layer by layer, and a plurality of branch company special personnel possibly give different confirmation results, so that the concentration degree after credit and the associated transaction management work are increased. On the other hand, in the whole service life cycle, the group of the client may have changes, and it is also a difficult and arduous task to capture the changes in time and make recognition again. The existing industry technology mainly solves the basic group pedigree automatic identification, but the group identification of individual scenes still has errors.

In one embodiment, a flow chart of a method for identifying a group pedigree is provided, which is shown in fig. 1 and includes the following steps:

step S101, acquiring a group pedigree image to be identified, wherein the group pedigree image comprises a topological relation among enterprises;

wherein, the group pedigree image reflects the equity relationship of each enterprise in the group, and the group pedigree image includes but is not limited to: topological relation among enterprises, node names of the enterprises, properties of the enterprises, equity relation among the enterprises and the like; searching and downloading can be carried out through a sky eye check, an enterprise check and a national industrial and commercial enterprise website, and then the family image of the group to be identified is obtained.

The corporate pedigree image is a map showing organizational relationships and property relationships between enterprises. In some examples, individual clique pedigree images may also be obtained manually.

For example, the format of the lineage images of a clique includes, but is not limited to, bmp, jpg, png, tif, gif, psd, pcd, wnf, raw, etc., and is not limited thereto.

Step S102, identifying each enterprise and corresponding topological relation in the pedigree image of the group one by one from bottom to top, and obtaining controlled enterprises in the pedigree of the group and the highest control party for controlling the controlled enterprises according to the topological relation among the enterprises, wherein the number of the highest control parties is at least one;

the method comprises the steps of performing penetrating identification from a controlled enterprise to the top in a group pedigree image, penetrating and identifying a control party at the upper level of an identified object layer by layer, judging a topological relation among enterprises until the highest control party in the topology is penetrated and identified, wherein the highest control party is located at the tail end of a stock right control path formed by the topological relation of the enterprises, and if one controlled enterprise corresponds to a plurality of stock right control paths, the highest control party is multiple.

Step S103, carrying out primary processing along an equity control path by taking the highest control party as a starting point in the group pedigree image to obtain a control coefficient between the highest control party and a controlled enterprise;

the method comprises the steps that a highest control party is used as a starting point, the whole group pedigree is penetrated and radiated layer by layer from top to bottom, group members in a group pedigree image are identified according to the stock right penetration, if a plurality of highest control parties exist in the group pedigree image, processing is carried out once along a stock right control path for many times, and then the control coefficient between each highest control party and a controlled enterprise in the group pedigree image can be obtained.

For example, the final control party is taken as a starting point, control coefficients are introduced, calculation is sequentially performed downwards along the direction of the stock control relationship from top to bottom, the control relationship between each level of enterprise and the next level of control party is calculated one by one, and then the control coefficient of the final control party to the controlled party is obtained.

And step S104, performing secondary processing according to the control coefficient and the stock holding ratio between the highest control party and the controlled enterprise, and determining the highest control party with the highest stock holding ratio corresponding to the controlled enterprise in the group pedigree image as a final control party.

Specifically, the highest control method of the pedigree vertex of the group found from bottom to top is taken as a starting point, the shareholders of all enterprises are divided into groups according to the group to which the shareholders belong, the share holding ratio of each group is calculated, the shareholder group with the largest share holding ratio is taken as a final control method, the group to which the current enterprise belongs is re-determined, and secondary judgment is carried out through the mode, so that the accuracy of group identification is further improved.

In the embodiment, the final control party and the controlled object of the group pedigree image are automatically identified by utilizing big data, and the topological connection relation between enterprises is determined, so that the labor cost is greatly reduced; meanwhile, a bidirectional identification mode is adopted to identify the control party of the controlled object, so that the identification accuracy of the pedigree of the group is improved; in the top-down identification link, the identification accuracy of the final control party is further improved through secondary calculation judgment and identification.

In other embodiments, in order to obtain a good quality clique pedigree image, the identified clique pedigree image needs to be preprocessed; before identifying each enterprise and corresponding topological relation in the family pedigree image one by one from bottom to top, the step of preprocessing the family pedigree image is detailed as follows:

step S001, converting the acquired group pedigree image to be identified into a group pedigree image with a preset specification size;

step S002, performing normalization processing on the group pedigree images with preset specifications to obtain group pedigree images with uniform pixel colors;

and S003, performing data enhancement processing on the group pedigree image with uniform pixel colors to obtain a preprocessed group pedigree image, wherein the data enhancement processing is performed on the group pedigree image with uniform pixel colors.

Wherein, the group pedigree images are uniformly scaled to 512 × 512; performing normalization operation, performing Gaussian blur processing on the group pedigree image, reversely superposing the group pedigree image with the original image, and moving the pixel color mean value to 128; performing data enhancement operation on the images, and performing the following operation on each group pedigree image: flipping horizontally with a probability of fifty percent; transposing with a fifty percent probability; random gamma transformation, the gamma value is limited between (50, 150); randomly changing the HUE, saturation and value of the picture with a probability of fifty percent; performing a translation rotation operation with a probability of fifty percent; operating in any one of the following combinations: a constrained contrast histogram equalization (CLAHE) algorithm; an algorithm of overlapping with itself after image sharpening (IAASharpen); an algorithm for performing a certain degree of embossing operation on the image, fusing the result with the image through a certain channel (IAAEmboss); random brightness and contrast adjustment; random blurring and gaussian noise addition.

Through the method, on one hand, the group pedigree images with different formats and sizes can be converted into the preset format through the preprocessing mode, and meanwhile, the group pedigree images with the preset specification are subjected to normalization processing to obtain the group pedigree images with uniform pixel colors; on the other hand, the group pedigree image with uniform pixel color is enhanced, and the group pedigree image with standard brightness, definition and saturation value can be obtained.

In other embodiments, the enterprises and corresponding topological relations in the corporate pedigree image are identified one by one from bottom to top, and the controlled enterprise in the corporate pedigree and the highest controlling party controlling the controlled enterprise are obtained according to the topological relations among the enterprises, which is detailed in fig. 2, and is a schematic diagram of an identification process of the highest controlling party in the corporate pedigree identification method provided in an embodiment of the present invention, which is detailed as follows:

step S201, identifying each enterprise and corresponding topological relation in the group pedigree image;

the node information of each enterprise node in the family pedigree image is identified by using an optical character recognition technology, for example, the node newly comprises but is not limited to an enterprise name, an enterprise type, and share right control between enterprises; and determining the corresponding topological relation of each enterprise according to the connection relation among the enterprise nodes in the group pedigree image.

It should be noted that, the algorithm for identifying each enterprise node in the pedigree image of the group and the topological relationship between each enterprise node further includes, but is not limited to, performing image identification based on a full convolution neural network, and performing image identification based on a convolution neural network.

For example, a full convolution neural network is built, a full convolution neural network training concentrated group pedigree image is used as a training element and input to the full convolution neural network, a determination value tag for whether a topological structure area exists is obtained through prediction of the full convolution neural network, a full convolution neural network loss function is further built by combining with a tag manually marked in the topological structure area in the neural network data set, and the full convolution neural network loss function is optimized and trained until the full convolution neural network loss function reaches Nash equilibrium to obtain the optimized full convolution neural network; furthermore, the full convolution neural network is formed by sequentially cascading an input layer, a plurality of convolution layers, a pooling layer, an anti-convolution layer and a softmax layer.

Step S202, determining a controlled enterprise according to an enterprise node corresponding to a controlled object in a topological relation formed by enterprise nodes in the group pedigree image from bottom to top;

wherein, in acquiring the topological relation formed by the enterprise nodes,

step S203, identifying the controlled enterprise from bottom to top, and determining a last-level controller of the controlled enterprise, wherein when the controlled enterprise has one or more corresponding enterprise nodes for controlling stock, the enterprise node with the largest stock control right value is determined as the last-level controller of the controlled enterprise;

for example, when a corporate enterprise is detected as being and having only one shareholder, the shareholder is the superior controller of the enterprise;

for another example, when a corporate enterprise is detected to have a plurality of shareholders, and the share right of any shareholder exceeds 50%, the shareholder is the upper-level control party of the enterprise;

for another example, according to international practice, it is generally believed that a significant impact is expected when an investor directly or indirectly owns more than 20% of shares of an invested company. Therefore, when the share ratio of the corporate enterprise shareholder is detected to exceed 20% and be less than 50% during the process of passing, and the share ratio of the shareholder is the highest and only share ratio, the shareholder is the upper-level control party.

Step S204, carrying out layer-by-layer penetration identification according to the stock right control path of the topological relation, and obtaining the upper-level controllers corresponding to each layer of enterprise nodes one by one;

step S205, when it is detected that there is no previous-level controller in a certain layer of enterprise nodes, determining the layer of enterprise nodes as a highest controller, where at least one highest controller corresponds to the controlled enterprise in the topological relationship.

(1) The share right ratio of any shareholder of the corporate enterprise is lower than twenty percent

(2) The national sponsor: corporate enterprise shareholder government agency and government-related business unit

Wherein, what the stock right information is the "stock name" and the "payment accepting ratio" that the enterprise corresponds, the principle of considering the enterprise as the first big stock is:

if the enterprise has only one shareholder, the shareholder is the first big shareholder of the enterprise;

if the enterprise has a plurality of shareholders, and the 'payment accepting ratio' of any shareholder exceeds fifty percent, the shareholder is the first large shareholder of the enterprise;

if the "share ratio" of the shareholder of the enterprise exceeds twenty percent and is less than fifty percent, and the "share ratio" of the shareholder is the highest and only shareholder, then the shareholder is the first largest shareholder of the enterprise.

Specifically, through one-by-one recognition from bottom to top, namely, a layer-by-layer penetration recognition mode, the penetration recognition of each layer is carried out from the enterprise at the lowest layer to the enterprise at the highest layer along the equity relationship, and the upper level control party is obtained in sequence until any condition is met, so that the highest control party belonging to the controlled party is obtained.

By the method, the enterprise of the highest control party is identified according to the equity relation through a bottom-up principle, and the highest control party corresponding to the controlled party can be quickly found.

In other embodiments, the step of performing a processing along an equity control path from the highest controlling party in the group pedigree image to obtain a control coefficient between the highest controlling party and a controlled enterprise is shown in fig. 3, which is a schematic control coefficient determining flow diagram in the group pedigree identification method provided in an embodiment of the present invention, and is detailed as follows:

step S301, taking the highest control party as a starting point in the pedigree image of the group, and sequentially carrying out the following processing on the next layer of controlled enterprises from top to bottom along a stock right control path to obtain a control coefficient between the highest control party and the controlled enterprises;

and selecting the step S302 or S303 to obtain the control coefficient between the highest control party and the controlled enterprise according to the following conditions.

Specifically, if the control coefficient is greater than a first preset proportion, a control relationship is formed; if the control coefficient is larger than the second preset proportion and smaller than the first preset proportion, forming a major influence relation; if the second predetermined ratio is smaller than the first predetermined ratio, the relationship is normal.

For example, the method of introducing the control coefficient calculates the control coefficient between the controlled enterprise and the control party at the upper level.

According to the provisions of the official act: "the stockholder will make a resolution and must pass through half of the votes held by the stockholders attending the conference". Therefore, when the investor directly owns more than 50% of shares of the invested company, the invested company is considered to be actually controlled, and the control coefficient is 1; if there are two or more investment relations less than 50% in the penetration path, the control coefficient of the path is 0. When the control coefficient is larger than 0.5, determining to form a control relation; greater than 0.2 and less than 0.5 constitute a significant influence relationship; less than 0.2 is a general relationship.

In the process of penetrating from top to bottom, the indirect investment proportion is obtained by multiplication, so the longer the path is, the smaller the indirect investment proportion is, and when the share proportion is less than 20%, the control relation cannot be formed according to the rules. However, in actual services, if each layer forms a control relationship, the highest control party has control capability for each node enterprise of a control path regardless of the length of the path.

Step S302, when detecting that the share occupied by the current controlled enterprise by the previous-level control party reaches a first preset proportion, the previous-level control party reaching the first preset proportion and the controlled enterprise form actual control, and the generated control coefficient is 1;

the first preset proportion can be set according to user requirements, and in this embodiment, the first preset proportion is preferably 50%.

Step S303, when it is detected that the share of the current controlled enterprise by the previous-level controller does not reach the first preset proportion, the previous-level controller and the controlled enterprise which do not reach the first preset proportion do not form actual control, and the generated control coefficient is 0.

Referring to fig. 9, compared to the conventional cumulative multiplication method, for example, the holdup ratio of company a 4 to company a is 13.90%, (52.45% × 51.9808% × 51%), which is less than the lower control limit of 20%, and the former does not form a control relationship with the latter. However, according to the method of controlling the coefficient in the present embodiment, the control coefficient of the former to the control coefficient of the latter is 1, and it is obvious that the company is calculated by using the control coefficients to be more in line with the actual business logic.

According to the rule of the 'official laws', the control coefficient is 1 every time, actual control is formed, namely, the third share proportion is larger than 50%, and the control coefficient of a company corresponding to the former to a company corresponding to the latter is 1, so that the fact that the operator is calculated by using the control coefficient is more consistent with actual business logic obviously. For example, the local nationally owned company A4 accounts for 52.45% of the local nationally owned company A2, more than 50% of the local nationally owned company A, and the control coefficient is 1; similarly, the local nationally owned company A2 accounts for 51.9808% of the local nationally owned company A1, more than 50% of the local nationally owned company A, and the control coefficient is also 1; the local state company A1 accounts for 51 percent of the shares of the company A, more than 50 percent of the shares, and the control coefficient is also 1, so the company A4 actually controls the company A.

By the method, the full-quantity relation splicing calculation is realized by the stock right penetration calculation, and the identification running time of the associated party is effectively reduced; meanwhile, the control coefficient is used as a judgment index, and a classification model can be constructed on the basis of stock right penetration, namely, the stock relation of each node is calculated according to the stock right relation direction in the knowledge graph, so that the classification model for automatically identifying the stock relation among enterprises is realized.

In other embodiments, further comprising: when detecting that a controlled enterprise (a controlled object) and a highest control party belong to the same pedigree in a pedigree image of a group, determining whether an investment relationship exists in the branch organization according to the stock holding proportion of the controlled enterprise and the highest control party, and if the investment relationship does not exist, perfecting a pedigree identification result by utilizing identified pedigree members; otherwise, it is not processed.

Specifically, the controlled enterprise and the highest control party belong to the same pedigree and calculate based on the share proportion, in the actual business, most branches do not have investment relation, and when the pedigree members are identified from top to bottom, the branch relations are brought in, so that the pedigree identification result is perfected.

Through above-mentioned mode, can automatic identification group pedigree image, compare the management mode of the group's map of financial industry, utilize big data automatic identification group pedigree image and generate the group pedigree, greatly reduce artificial work load, improved identification efficiency, simultaneously, also improve the quality of group pedigree management.

In other embodiments, processing relational triple data in the knowledge graph based on a SparkGraphX graph computing architecture to generate a multi-node distributed equity penetration computing engine;

the Spark graph X is a distributed graph processing framework, provides a simple, easy-to-use and rich interface for graph calculation and graph mining based on a Spark platform, and greatly facilitates the requirement on processing of the distributed graph; spark graph x is naturally a distributed graph processing system, since the bottom layer is processed based on Spark. The distributed or parallel processing system of the graph divides the graph into a plurality of subgraphs, then the subgraphs are respectively calculated, and the computation can be respectively iterated to carry out staged computation during computation, namely, the graph is subjected to parallel computation.

Through the distributed graphic processing cross-frame, a multi-node distributed stock right penetration calculation engine is generated, full-scale relation splicing calculation is realized through the multi-node distributed stock right penetration calculation engine, and the identification running time of an associated party is effectively reduced.

Determining the share occupation relation of each enterprise node in the knowledge graph by using the share penetration calculation engine;

the share relation between the nodes formed by the enterprises in the knowledge graph is calculated, for example, 80% of shares of a certain enterprise corresponding to the point a of the enterprise share a certain enterprise corresponding to the point B of the enterprise, wherein the share penetration calculation is to perform share calculation in the knowledge graph from bottom to top or from bottom to top, so as to obtain the share relation between the nodes.

And determining control coefficients among nodes corresponding to the enterprises according to the share relation among the enterprises in the direction of the stock control relation in the knowledge graph.

In other embodiments, the step of performing secondary processing by combining the stock-holding ratio between the highest controlling party and the controlled enterprise according to the control coefficient to determine the highest controlling party with the largest stock-holding ratio corresponding to the controlled enterprise in the group pedigree image as the final controlling party is described in detail in fig. 4, which is a schematic flow diagram determined by the final controlling party in the group pedigree identification method provided in an embodiment of the present invention, and the detailed description is as follows:

step S401, taking the highest control party as a starting point in the group pedigree image, and dividing each enterprise according to an equity control path to form an equity group;

specifically, the highest controllers in the topology structure are respectively used as starting points, and the enterprise nodes related to the right control paths to the lowest controlled enterprises are respectively used as shareholder groups.

Step S402, determining the stock holding ratio between the highest control party and the controlled enterprise in each shareholder group from top to bottom according to the stock right control path;

specifically, the stock holding ratio between the highest control party in each stockholder group and the controlled enterprise is determined according to the control proportion of the upper-level control party of each controlled enterprise.

And step S403, determining the highest controller corresponding to the stockholder group with the highest stock holding ratio between the highest controller and the controlled enterprise as the final controller.

Specifically, a final control party is found by going through from bottom to top, if the highest control party is not only one enterprise, the stock holding ratio of the highest control party to the controlled party is utilized, wherein the stock holding ratio is determined by the direct stock holding ratio and the indirect stock holding ratio of each enterprise, and the controlled party is classified as the final control party with the largest stock holding ratio. For example, starting from the highest control party, through penetrating downwards to find the final beneficial share right, excluding the data which accounts for less than twenty percent of the shares, finding the group (party) with the largest beneficial share right as the final control party, and circularly fitting until all the end enterprises are found according to the above manner, determining the highest control party corresponding to the stockholder group with the largest share holding ratio between the highest control party and the controlled enterprises to be the final control party.

In other embodiments, if company a goes up to great stockholder company B, on a bottom-up strike-through basis, it belongs to the B-holded group, where the B-holded shares are greater than the shares of company C1 and company C2, but the sum of the shares of company C1 and company C2 is greater than the B-holded shares. In the case where the company C1 and the company C2 belong to the C stock group by performing the secondary certification on the company (controlled object), the C stock group actually holds the largest stock, and the stock far exceeding 50% of the stock also exceeds the B stock group.

Through the method, the highest control party is judged in a secondary identification mode, the final control party in the group pedigree image is determined, and the identification precision of the group pedigree image is improved.

Optionally, in other embodiments, in order to improve the real path in the enterprise association relationship identification method in the family pedigree image, the following details are described:

determining the type of each enterprise node according to the control coefficient among the enterprise nodes in the group pedigree image, wherein the type at least comprises a subsidiary company, a joint company and a co-operation company;

according to the new justice rule, the control types of the companies are divided into sub-companies, joint companies and affiliated companies, and the control types of the companies are determined according to the control coefficient range occupied by the control coefficients among the nodes in the knowledge graph.

And generating a plurality of real paths of the association parties between the enterprises and the association parties according to the types of the enterprise nodes and the stock control relationship.

And generating a plurality of real paths of the correlation parties among the correlation parties according to the relationship among the nodes and the control type among the companies, for example, generating a plurality of real paths of the correlation parties according to the relationship among the nodes and the control type formed by a new judicial law, wherein the real paths of the correlation parties accord with the new company regulation.

For example, if company A makes up more than 50% of the shares of company B, then company B is a subsidiary of company A; if the share of the company A in the company B is just equal to 50%, the company B is a co-operation company of the company A; if the share of the company A in the company B is between 20 and 50 percent, the company B is a joint company of the company A.

Through the mode, according to the provisions of the 'official laws', the subsidiary companies, the affiliated company and the affiliated company of the enterprise are identified according to the control coefficients, and the accuracy of identification of the related party is improved.

In other embodiments, referring to fig. 5 in detail, a schematic diagram of an actual control determination flow in the group pedigree identification method provided in an embodiment of the present invention further includes:

step S501, detecting whether a control enterprise at the upper level corresponding to a controlled enterprise has an action actor or not;

step S502, when detecting that a consistent actor exists in a previous-level control enterprise corresponding to a controlled enterprise, judging whether the share of the previous-level control enterprise exceeds a first preset proportion; if the upper-level control enterprises do not exceed the first preset proportion, determining the related enterprises as a common control relation according to the relation corresponding to the consistent actor;

step S503, detecting the highest control party belonging to the common control relationship in the group pedigree image;

step S504, when the highest control party of the group pedigree image belonging to the common control relationship is detected to be any one of a limited partner enterprise, a vacant shell enterprise or a stock holding platform, according to the current situation, an enterprise node occupying the largest stock holding proportion among the controlled enterprises is determined by means of layer-by-layer screening from top to bottom, and the current enterprise node is determined as an actual control party in the group pedigree image.

Specifically, referring to fig. 10 as an example, the tie-up logic with the largest holding proportion of company c is identified, and the company c should be tied up to the company d and become the final controller, however, in step S502, since there is an action actor in the stock control relationship of company c (for example, the action actors are formed among company c1, company e, and company z), and the accumulated holding exceeds 50%, the company c is configured to have a common control relationship, and therefore, in this embodiment, the actually corresponding final controller should be the company corresponding to the action actor.

Specifically, in the above step S503, when the highest controlling party belonging to the common controlling relationship is detected to be any one of the limited partner enterprise, the vacant enterprise or the stock supporting platform according to the stock right penetration logic, for example, when the highest controlling party is detected to be the limited partner enterprise and the stock share ratio of the limited partner enterprise is the largest, other types of companies excluded as the limited partner enterprise are determined according to the types of the enterprise (including at least the subsidiary, the affiliated company and the partnering company), and other enterprises (the subsidiary, the affiliated company and the partnering company) next to the limited partner enterprise stock are determined as the actual controlling party (actual final controlling party) of the controlled enterprise.

Similarly, when the highest control party belonging to the common control relationship is an empty enterprise or an inventory holding platform, the final control party (enterprise) corresponding to the empty enterprise or the inventory holding platform is excluded and determined as the actual control party of the controlled enterprise in the same manner as described above.

Through the mode, the identification processes of different final control parties are determined according to different enterprise property types of the identified enterprises, the identification processes of enterprise nodes in the group pedigree images are adaptively adjusted, the group pedigree images under different scenes are adapted, the flexibility of an algorithm is improved according to a scene distinguishing and feedback mechanism, and a set of complete closed-loop group identification scheme is formed, so that the identification accuracy of the group pedigree is improved.

In other embodiments, referring to fig. 6 in detail, a small-lot update flow diagram in the group pedigree identification method provided in an embodiment of the present invention further includes:

step S601, identifying a full-scale pedigree between a controlled enterprise and a final control party according to the pedigree image of the group;

wherein, the full pedigree is the identification result of each group pedigree image of the database, and the relationship between the controlled enterprise and the final control party identified in each group pedigree image can be determined through the full pedigree, such as the controlled enterprise, the highest control party, the final control party, and the like.

Step S602, fusing the enterprise relationships and the enterprise nodes determined in the full-scale pedigree by using priority sequencing;

the priority ranking is preset by a database, for example, priority list enterprises are identified in a mode of mutually supplementing two modes of a unified social credit code and a regular matching (regular expression), wherein the social unified credit code carries out entity classification on the enterprises based on an industrial and commercial coding rule; or clustering according to the similarity of the text topics identified by the full-scale pedigree; performing multi-target sequencing on the clustering results of the full-scale pedigree, and adjusting a sequencing sequence according to the clustering results; mining association rules according to historical execution results of the full-scale pedigree, and dynamically adjusting the sequencing sequence; thereby realizing the sequencing of the priority of the whole pedigree, further re-determining the enterprise relationship and the enterprise node determined in the identified whole pedigree,

step S603, local calculation is carried out based on the group pedigree image, the determined enterprise relation and enterprise nodes are updated, an incremental pedigree formed by penetrating the group pedigree image is obtained, and small-batch updating of the group pedigree is achieved.

The local calculation of the group pedigree image is realized by comparing the local calculation with a group pedigree image stored in the past historical data, namely, comparing the updated group pedigree image with the group pedigree image before updating by capturing the same group pedigree image, determining the updated group pedigree image, identifying a local area of the group pedigree image by using the comparison detection of the updated group pedigree image and the stored group pedigree image before updating, identifying an increment pedigree corresponding to the updated local area by the layer-by-layer penetration identification, and realizing the small-batch updating of the group pedigree if the group pedigree image is only the increment pedigree corresponding to the updated local area.

Specifically, the full pedigree is calculated based on more than 5000 ten thousand relations and more than 1 hundred million nodes, a great deal of time and resources are consumed, and finally, the identification result corresponding to the full pedigree is obtained. And this technique carries out the local calculation of pedigree based on the graph data, realizes pedigree T +0 and updates, and in practical application, the business personnel has formed professional affirmation (enterprise) relation under specific scene based on professional experience, has fused the affirmation relation through priority sequencing based on big data (the identification result that the full pedigree corresponds), and according to full pedigree result delineation affirmation relation influence, utilize relation and node after the re-affirmation, pierce through and form the increment pedigree, realize group pedigree small batch volume and update.

By the method, the relationship and the nodes of the group pedigree are quantitatively updated by adopting a small batch updating mode, a feedback mechanism can be formed, and a set of complete closed-loop group identification scheme is realized.

The embodiment provides a group pedigree identification method, which identifies control parties in a group pedigree in a bidirectional identification mode, identifies the control parties by utilizing a stock right relationship and sequentially penetrating upwards on a bottom-up principle, and identifies the highest control party of a vertex; penetrating and identifying layer by layer along the principle of the highest control party from top to bottom to obtain the control coefficients of all enterprises of the group pedigree; determining the stock control proportion of each company in the group pedigree in a top-down identification mode, judging the enterprise attribution to determine the final control party of the enterprise, and obtaining the incidence relation of each enterprise in the group pedigree; on one hand, the group pedigree is automatically identified, so that the identification efficiency is improved; on the other hand, the two-way identification mode is adopted to respectively carry out penetration identification, so that the identification accuracy is improved.

In one embodiment, the present invention further provides a group pedigree identification apparatus 700, referring to fig. 7, comprising:

the system comprises an acquisition module 701, an acquisition module and a processing module, wherein the acquisition module is used for acquiring a group pedigree image to be identified, and the group pedigree image comprises a topological relation among enterprises;

wherein, the obtaining module further comprises, after obtaining the pedigree image of the group:

An identifying module 702, configured to identify each enterprise and the corresponding topological relation in the pedigree image of the group one by one from bottom to top, and obtain a controlled enterprise in the pedigree of the group and a highest controlling party controlling the controlled enterprise according to the topological relation between the enterprises, where at least one highest controlling party is provided;

specifically, the identification module further includes:

A control coefficient determining module 703, configured to perform primary processing along an equity control path in the group pedigree image with the highest controlling party as a starting point, so as to obtain a control coefficient between the highest controlling party and a controlled enterprise;

specifically, the control coefficient determination module further includes:

A final control party determining module 704, configured to perform secondary processing according to the control coefficient in combination with the stock holding ratio between the highest control party and the controlled enterprise, and determine the highest control party with the largest stock holding ratio corresponding to the controlled enterprise in the group pedigree image as the final control party;

specifically, the final controller determining module further includes:

Optionally, based on the above embodiment, the group pedigree identification apparatus further includes:

the actual control determining module is used for detecting whether an action person exists in a previous-level control enterprise corresponding to the controlled enterprise; when detecting that a consistent action person exists in a previous-level control enterprise corresponding to a controlled enterprise, judging whether the share occupation of the previous-level control enterprise exceeds a first preset proportion; if the upper-level control enterprises do not exceed the first preset proportion, determining the related enterprises as a common control relation according to the relation corresponding to the consistent actor; the system is also used for detecting the highest control party belonging to the common control relationship in the group pedigree image; when the situation that the highest control party of the group pedigree image belonging to the common control relationship is any one of a limited partner enterprise, a vacant shell enterprise or a stock holding platform is detected, enterprise nodes occupying the largest stock holding proportion among the controlled enterprises are determined by screening layer by layer from top to bottom according to the current situation, and the current enterprise nodes are determined as the actual control parties in the group pedigree image.

Optionally, based on the above embodiment, the group pedigree identification apparatus further includes: small batch update module for

Identifying a full-scale pedigree between a controlled enterprise and a final control party according to the pedigree image of the group; fusing enterprise relationships and enterprise nodes determined in the full-scale pedigree by using priority ranking; and performing local calculation based on the group pedigree image, updating the determined enterprise relation and enterprise nodes to obtain an incremental pedigree formed by penetrating the group pedigree image, and realizing small-batch updating of the group pedigree.

The embodiment provides a group pedigree identification device, which identifies control parties in a group pedigree in a bidirectional identification mode, identifies the control parties by utilizing a stock right relationship and sequentially penetrating upwards on a bottom-up principle, and identifies the highest control party of a vertex; penetrating and identifying layer by layer along the principle of the highest control party from top to bottom to obtain the control coefficients of all enterprises of the group pedigree; determining the stock control proportion of each company in the group pedigree in a top-down identification mode, judging the enterprise attribution to determine the final control party of the enterprise, and obtaining the incidence relation of each enterprise in the group pedigree; on one hand, the group pedigree is automatically identified, so that the identification efficiency is improved; on the other hand, the two-way identification mode is adopted to respectively carry out penetration identification, so that the identification accuracy is improved.

It should be understood that the group pedigree identification device system substantially has a plurality of modules for executing the group pedigree identification method in any of the above embodiments, and specific functions and technical effects are only referred to the above embodiments, and are not described herein again.

In an embodiment, referring to fig. 8, the embodiment further provides a computer apparatus 800, including a memory 801, a processor 802, and a computer program stored on the memory and executable on the processor, where the processor 802 implements the steps of the method according to any one of the above embodiments when executing the computer program.

In an embodiment, a computer-readable storage medium is also provided, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method according to any of the above embodiments.

The embodiment of the application can acquire and process related data based on an artificial intelligence technology. Among them, Artificial Intelligence (AI) is a theory, method, technique and application system that simulates, extends and expands human Intelligence using a digital computer or a machine controlled by a digital computer, senses the environment, acquires knowledge and uses the knowledge to obtain the best result.

It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, apparatus, article, or method. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, apparatus, article, or method that includes the element.

The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments. Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium (e.g., ROM/RAM, magnetic disk, optical disk) as described above and includes instructions for enabling a terminal device (e.g., a mobile phone, a computer, a server, or a network device) to execute the method according to the embodiments of the present invention.

The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.

Claims

1. A method for identifying a group pedigree, the method comprising:

2. The method of claim 1, wherein before identifying each enterprise and corresponding topological relationship in the corporate lineage image one by one from bottom to top, the method further comprises:

3. The method for identifying corporate pedigrees according to claim 1, wherein the identifying each enterprise and corresponding topological relation in the corporate pedigree image one by one from bottom to top, and obtaining the controlled enterprise in the corporate pedigree and the highest controlling party controlling the controlled enterprise according to the topological relation between the enterprises comprises:

4. The method of any one of claims 1 to 3, wherein the performing a process along an equity control path from the highest controlling party in the corporate lineage image to obtain a control coefficient between the highest controlling party and a controlled enterprise comprises:

5. The method for identifying the clique pedigree as claimed in any one of claims 1 to 3, wherein the performing of the secondary processing according to the control coefficient and the stock holding ratio between the highest control party and the controlled enterprise to determine the highest control party with the highest stock holding ratio corresponding to the controlled enterprise in the clique pedigree image as the final control party comprises:

6. The method of clique pedigree identification according to claim 1, further comprising:

7. The method of clique pedigree identification according to claim 1, further comprising:

8. A group pedigree identification apparatus, the apparatus comprising:

9. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the steps of the method of any of claims 1 to 7 are implemented by the processor when executing the computer program.

10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 7.