CN115796572A

CN115796572A - Risk enterprise identification method, apparatus, device and medium

Info

Publication number: CN115796572A
Application number: CN202211348885.5A
Authority: CN
Inventors: 王泽皓; 林文辉; 王志刚; 杨军; 钱剑林; 闫凯; 马谊骏
Original assignee: Anhui Aisino Corp; Aisino Corp
Current assignee: Anhui Aisino Corp; Aisino Corp
Priority date: 2022-10-31
Filing date: 2022-10-31
Publication date: 2023-03-14

Abstract

The invention discloses a method, a device, equipment and a medium for identifying a risk enterprise, wherein the method comprises the steps of determining each enterprise and each enterprise employee which are in two-degree association with a blacklist enterprise and each blacklist enterprise employee, screening possible risk enterprises from an enterprise relationship graph, and determining the similarity of a first vector and a second vector respectively corresponding to the blacklist enterprise and the blacklist enterprise employee according to a first vector after vectorization representation of a node of each enterprise in a sub-enterprise relationship graph, so that the risk enterprises which are highly similar in all aspects are predicted, and accurate identification of the risk enterprises is realized.

Description

Risk enterprise identification method, apparatus, device and medium

Technical Field

The invention relates to the technical field of risk prediction, in particular to a risk enterprise identification method, device, equipment and medium.

Background

With the rapid development of economy in China, tax administration becomes more and more important, tax administration departments need to accurately attack risk enterprises with tax risks and comprehensively mine the tax risks of the enterprises, so that how to identify the risk enterprises becomes a valuable problem.

In the existing enterprise risk identification method and system based on enterprise sales relationship maps, the technical scheme mainly comprises three steps, wherein the first step is to collect target tax data and construct an enterprise relationship map; and the third step is to judge whether the industry attributes of the enterprises with similar purchase and sale are consistent or not through the comparison of the industry attributes of the enterprises, and further identify the risk enterprises.

In the prior art, only by means of industry attribute comparison, the risk enterprise cannot be accurately identified, so how to improve the accuracy of risk enterprise identification becomes a technical problem to be solved urgently.

Disclosure of Invention

The invention provides a method, a device, equipment and a medium for identifying an enterprise at risk, which are used for solving the problem that the enterprise at risk cannot be accurately identified in the prior art.

The invention provides a risk enterprise identification method, which comprises the following steps:

determining attribute labels as each first node of blacklist enterprises and blacklist enterprise employees and a sub-enterprise relation map of each second node which is in second-degree association with each first node based on a pre-constructed enterprise relation map;

obtaining a first vector, which is output by the risk model and takes the attribute label in the sub-enterprise relational graph as each second node of the enterprise, according to the attribute label corresponding to each second node in the sub-enterprise relational graph and a pre-stored risk model;

and for each second node of which the attribute label is an enterprise, determining the similarity between the first vector of the second node and each second vector according to the second vectors respectively corresponding to the first vector and the predetermined blacklist enterprise and blacklist enterprise staff, and if any similarity is greater than a preset similarity threshold, determining that the enterprise corresponding to the second node of the first vector is an inauguration enterprise.

Further, the method further comprises:

according to the identified first vectors corresponding to each inauguration enterprise, carrying out normalization processing on each first vector corresponding to each inauguration enterprise to obtain each normalized first vector;

and outputting the normalized module value of each first vector and the identification information of each inauguration enterprise contained in each corresponding second node according to the sequence of the normalized module value of each first vector from large to small.

Further, the construction process of the enterprise relationship graph comprises the following steps:

according to the identification information of enterprises and the identification information of the enterprise employees in the tax data which are obtained in advance, nodes of the enterprises and the enterprise employees in the relational graph are constructed, the corresponding identification information is used as the node identification information, if any one of the enterprises is a blacklist enterprise in the blacklist information, the attribute label of the node corresponding to the enterprise is set as the blacklist enterprise, otherwise, the attribute label of the node corresponding to the enterprise is set as the enterprise, if any one of the enterprise employees is the blacklist enterprise employee in the blacklist information, the attribute label of the node corresponding to the enterprise employee is set as the blacklist enterprise employee, and otherwise, the attribute label of the node corresponding to the enterprise employee is set as the enterprise employee;

aiming at each invoice information in the tax data, connecting nodes of two target enterprises according to identification information of the two target enterprises contained in the invoice information;

and connecting the node of the target enterprise employee with the node of the incumbent enterprise according to the information of the target enterprise employee and the incumbent enterprise in the tax data.

Further, the connecting the nodes of the two target enterprises according to the identification information of the two target enterprises contained in the invoice information includes:

according to the identification information of two target enterprises contained in the invoice information and the preset keywords which are correspondingly stored, wherein the preset keywords comprise a buyer and a seller, the node which takes the preset keywords as the seller in the two target enterprises points to the node which takes the preset keywords as the buyer, one side of the node which takes the seller as a sale side is set as a sale side, one side of the node which takes the buyer side as an entry side, and the transaction amount, the transaction time and the commodity name in the invoice information are used as attribute information of the corresponding side.

Further, the connecting the node of the target enterprise employee with the node of the incumbent enterprise according to the information of the target enterprise employee and the incumbent enterprise in the tax data includes:

and connecting nodes of the target enterprise employees to nodes of the job-holding enterprise in the relation graph according to the first identification information of the target enterprise employees, the first identification information of the job-holding enterprise and the job information in the tax data, and taking the job information as edge attribute information.

Correspondingly, the invention provides an inauguration enterprise identification device, which comprises:

the determining module is used for determining attribute labels as each first node of blacklist enterprises and blacklist enterprise employees and a sub-enterprise relationship graph of each second node which is in two-degree association with each first node based on the enterprise relationship graph which is constructed in advance;

the processing module is used for obtaining a first vector, which is output by the risk model and takes the attribute label in the sub-enterprise relational graph as each second node of the enterprise, according to the attribute label corresponding to each second node in the sub-enterprise relational graph and a pre-stored risk model;

and the identification module is used for determining the similarity between the first vector and each second vector according to the second vector respectively corresponding to the first vector of each second node and the pre-determined blacklist enterprise and blacklist enterprise employee, and if any similarity is greater than a preset similarity threshold, determining that the enterprise corresponding to the second node of the first vector is an inauguration enterprise.

Further, the apparatus comprises:

the output module is used for carrying out normalization processing on each first vector corresponding to each inauguration enterprise according to each first vector corresponding to each identified inauguration enterprise to obtain each normalized first vector; and outputting the normalized module value of each first vector and the identification information of each inauguration enterprise contained in each corresponding second node according to the descending order of the normalized module value of each first vector.

Further, the apparatus comprises:

the model building module is used for building nodes of enterprises and enterprise employees in a relational graph according to enterprise identification information and enterprise employee identification information in the tax data acquired in advance, taking the corresponding identification information as node identification information, if any enterprise is a blacklist enterprise in the blacklist information, setting attribute labels of the nodes corresponding to the enterprises as the blacklist enterprise, otherwise, setting the attribute labels of the nodes corresponding to the enterprises as the enterprises, if any enterprise employee is the blacklist enterprise employee in the blacklist information, setting the attribute labels of the nodes corresponding to the enterprise employees as the blacklist enterprise employees, and otherwise, setting the attribute labels of the nodes corresponding to the enterprise employees as the enterprise employees; aiming at each invoice information in the tax data, connecting nodes of two target enterprises according to identification information of the two target enterprises contained in the invoice information; and connecting the node of the target enterprise employee with the node of the arbitrary enterprise according to the information of the target enterprise employee and the arbitrary enterprise in the tax data.

Further, the model building module is specifically configured to, according to identification information of two target enterprises included in the invoice information and preset keywords that are stored correspondingly, where the preset keywords include a buyer and a seller, point a node in the two target enterprises where the preset keyword is the buyer to a node where the preset keyword is the seller, set one side of the node where the seller is located as a sale edge, set one side of the node where the buyer is located as an entry edge, and use the transaction amount, the transaction time, and the commodity name in the invoice information as attribute information of the corresponding edge.

Further, the model building module is specifically configured to connect nodes from the nodes of the target enterprise employees to the nodes of the incumbent enterprises in the relationship graph according to the first identification information of the target enterprise employees, the first identification information of the incumbent enterprises, and the job information in the tax data, and use the job information as edge attribute information.

Accordingly, the present invention provides an electronic device comprising: the system comprises a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory complete mutual communication through the communication bus;

the memory has stored therein a computer program which, when executed by the processor, causes the processor to perform the steps of any of the above-described inauguration enterprise identification methods.

Accordingly, the present invention provides a computer readable storage medium storing a computer program which, when executed by a processor, performs the steps of any of the above-mentioned inauguration enterprise identification methods.

The invention provides a risk enterprise identification method, a device, equipment and a medium, wherein the method comprises the steps of determining each first node with attribute labels of blacklist enterprises and blacklist enterprise employees and a sub-enterprise relationship graph of each second node with two-degree association with each first node from an enterprise relationship graph based on a pre-constructed enterprise relationship graph, inputting the sub-enterprise relationship graph into a risk model to obtain a first vector with the attribute labels of each second node of the enterprise, determining the similarity between the first vector and each second vector according to the first vector of each second node and second vectors respectively corresponding to the blacklist enterprises and the blacklist enterprise employees which are determined in advance, and determining the enterprise corresponding to the second node of the first vector as a risk enterprise if any similarity is greater than a preset similarity threshold; according to the method, each enterprise and each enterprise employee which are in two-degree association with the blacklist enterprise and the blacklist enterprise employee are determined, the possible risk enterprises are screened out from the enterprise relation graph, and the similarity of the first vector value and the second vectors respectively corresponding to the blacklist enterprise and the blacklist enterprise employee is determined according to the first vector value of each enterprise after vectorization representation of the node in the sub-enterprise relation graph, so that the risk enterprises which are highly similar in all aspects are predicted, and accurate identification of the risk enterprises is achieved.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without inventive exercise.

Fig. 1 is a schematic process diagram of an inauguration enterprise identification method according to an embodiment of the present invention;

fig. 2 is a process diagram of an inauguration enterprise identification method according to an embodiment of the present invention;

FIG. 3 is a flowchart of a microservice architecture provided by an embodiment of the present invention;

fig. 4 is a schematic diagram of a process of identifying an inauguration enterprise according to an embodiment of the present invention;

fig. 5 is a schematic structural diagram of an inauguration enterprise identification apparatus according to an embodiment of the present invention;

fig. 6 is a schematic structural diagram of an electronic device provided in the present application.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.

In order to improve accuracy of inauguration enterprise identification, embodiments of the present invention provide an inauguration enterprise identification method, apparatus, device and medium.

Example 1:

fig. 1 is a schematic process diagram of an inauguration enterprise identification method according to an embodiment of the present invention, where the process includes the following steps:

s101: and determining the attribute labels as each first node of the blacklist enterprise and the blacklist enterprise employee and the sub-enterprise relationship maps of each second node which has two-degree association with each first node based on the pre-constructed enterprise relationship maps.

In order to improve accuracy of inauguration enterprise identification, the embodiment of the present invention provides an inauguration enterprise identification method, which is applied to an electronic device, where the electronic device may be an intelligent terminal device such as a host, a tablet computer, a notebook computer, a smart phone, or a server, and the server may be a local server or a cloud server, and the embodiment of the present invention is not limited thereto.

In order to determine a possible risk enterprise, in the embodiment of the present invention, an enterprise relationship graph is constructed in advance, where the enterprise relationship graph refers to a directed graph constructed based on relationships between enterprises and employees of an enterprise, and the enterprise relationship graph is used to represent relationships between an enterprise and other enterprises and between employees of an enterprise, respectively.

Based on a pre-constructed enterprise relationship graph, the electronic equipment determines that the attribute label is each first node of a blacklisted enterprise and the attribute label is each first node of a blacklisted enterprise employee according to the attribute label of each node in the enterprise relationship graph, and according to each first node in the enterprise relationship graph, the electronic equipment performs multi-level logic query of enterprise relationship by adopting a graph query statement, and determines each second node which is in two-degree association with each first node by taking each first node as a center.

Specifically, in order to determine each second node which has a second degree association with each first node of the blacklisted enterprise, the electronic device adopts a graph query statement g.v (). HasLabel ('black _ nsr'). BothE (). OtherV (). SimplePath (). ToList (), wherein the black _ nsr indicates that the attribute tag is the first node of the blacklisted enterprise;

in order to determine that each second node with two-degree association exists in each first node of the blacklisted enterprise employee, the electronic device adopts a graph query statement g.v (). HasLabel ('black _ person'). InE (). OtherV (). OutE (). OtherV (). SimplePath (). ToList ().

The graph Query statement may be a graph traversal statement (Gremlin) or a structured Query Language (CQL), and the second degree association includes direct association and indirect association separated by one node, that is, any second node is directly connected to the first node or is connected to the first node at an interval of one second node.

After determining each first node and each second node, the electronic device determines a sub-enterprise relationship graph containing each first node and each second node in the enterprise relationship graph, wherein the sub-enterprise relationship graph is a partial graph of the enterprise relationship graph.

S102: and obtaining a first vector, which is output by the risk model and takes the attribute label in the sub-enterprise relational graph as each second node of the enterprise, according to the attribute label corresponding to each second node in the sub-enterprise relational graph and a pre-stored risk model.

In order to identify the risk enterprise, the electronic device vectorizes and represents the nodes of each enterprise in the sub-enterprise relational graph, and the electronic device pre-stores a risk model, wherein the risk model may be a graph neural network model or other integrated risk identification models, and is used for implementing vectorization and representation of each node in the input sub-enterprise relational graph.

The electronic equipment inputs the sub-enterprise relational graph into a pre-stored risk model, the risk model determines that the attribute label is each second node of the enterprise based on the attribute label of each second node in the sub-enterprise relational graph, and the risk model performs graph calculation according to each second node, the attribute labels connected with each second node are the second nodes of the enterprise employees and the connected edges, and determines that the attribute label is the first vector of each second node of the enterprise.

S103: and for each second node of which the attribute label is an enterprise, determining the similarity between the first vector of the second node and each second vector according to the second vectors respectively corresponding to the first vector and the predetermined blacklist enterprise and blacklist enterprise staff, and if any similarity is greater than a preset similarity threshold, determining that the enterprise corresponding to the second node of the first vector is an inauguration enterprise.

The electronic equipment prestores second vectors corresponding to blacklisted enterprises and blacklisted enterprise employees respectively aiming at each second node of an enterprise with an attribute label in a sub-enterprise relational graph, wherein the second vectors corresponding to the blacklisted enterprises and the blacklisted enterprise employees are prestores for inputting the sub-enterprise relational graph into a risk model, and graph calculation is carried out to determine the second vector of the first node of the blacklisted enterprise with the attribute label in the sub-enterprise relational graph and the second vector of the first node of the blacklisted enterprise employees.

According to the first vector of the second node and the predetermined second vectors corresponding to the blacklist enterprise and the blacklist enterprise staff, the electronic device determines the similarity between the first vector and each second vector through vector calculation, and also pre-stores a preset similarity threshold, wherein the similarity threshold is preset by a user, if the accuracy of the identification of the risky enterprise is expected to be improved, the similarity threshold can be set to be larger, and if the robustness of the identification of the risky enterprise is expected to be improved, the similarity threshold can be set to be smaller; according to the similarity threshold and the determined similarity between the first vector and each second vector, the electronic equipment compares each similarity with the similarity threshold, and if any similarity is greater than the similarity threshold, the enterprise corresponding to the second node of the first vector is determined to be an inauguration enterprise.

In the embodiment of the invention, each enterprise and each enterprise employee which are in two-degree association with the blacklist enterprise and the blacklist enterprise employee are determined, the possible risk enterprises are screened out from the enterprise relationship graph, and the similarity of the first vector and the second vectors respectively corresponding to the blacklist enterprise and the blacklist enterprise employee is determined according to the first vector value of each enterprise after vectorization representation of the node in the sub-enterprise relationship graph, so that the risk enterprises which are highly similar in all aspects are predicted, and the accurate identification of the risk enterprises is realized.

Example 2:

in order to show the risk level of each inauguration enterprise, on the basis of the above embodiment, in an embodiment of the present invention, the method further includes:

and outputting the normalized module value of each first vector and the identification information of each inauguration enterprise contained in each corresponding second node according to the descending order of the normalized module value of each first vector.

In order to show the risk level of each inauguration enterprise, after identifying each inauguration enterprise, the electronic device normalizes each first vector according to the first vector corresponding to each inauguration enterprise, specifically, performs processing by using an existing normalization processing algorithm, thereby obtaining each normalized first vector.

The normalization is to limit the preprocessed data within a certain range, so as to eliminate adverse effects caused by singular data.

According to each normalized first vector, determining the identification information of each inauguration enterprise contained in each second node corresponding to each first vector, performing modular operation on each first vector to obtain the modular value of each first vector, and outputting the normalized modular value of each first vector and the identification information of each inauguration enterprise contained in each corresponding second node according to the sequence of the modular values of each first vector from large to small.

Specifically, the electronic device may display and output the modulus value of each first vector and the corresponding identification information of each inauguration enterprise on a display screen of the electronic device; or existing voice synthesis software can be adopted to output the voice of the modulus value of each first vector and the corresponding identification information of each inauguration enterprise in a voice output device of the electronic equipment.

Fig. 2 is a schematic process diagram of an inauguration enterprise identification method provided by an embodiment of the present invention, and as shown in fig. 2, the process includes the following steps:

s201: and determining the attribute labels as the sub-enterprise relationship maps of each first node of the blacklist enterprise and the blacklist enterprise employee and each second node which has two-degree association with each first node based on the pre-constructed enterprise relationship maps.

S202: and inputting the sub-enterprise relational graph into the DGL graph neural network model, and obtaining a first vector of each second node of the enterprise, wherein the attribute label in the output sub-enterprise relational graph is the attribute label corresponding to each second node of the enterprise.

S203: and aiming at each second node of which the attribute label is an enterprise, determining the similarity of the first vector and each second vector according to the first vector of the second node and the predetermined second vectors respectively corresponding to the blacklist enterprise and the blacklist enterprise staff.

S204: and if any similarity is greater than a preset similarity threshold, determining that the enterprise corresponding to the second node of the first vector is an inauguration enterprise.

S205: and according to each first vector corresponding to each identified inauguration enterprise, normalizing each first vector corresponding to each inauguration enterprise to obtain each normalized first vector, and outputting the modulus value of each normalized first vector and the identification information of each inauguration enterprise contained in each corresponding second node according to the descending order of the modulus value of each normalized first vector.

Example 3:

in order to construct an enterprise relationship graph, on the basis of the foregoing embodiments, in an embodiment of the present invention, a construction process of the enterprise relationship graph includes:

according to the identification information of enterprises and the identification information of enterprise employees in the tax data acquired in advance, nodes of the enterprises and the enterprise employees in the relational graph are constructed, the corresponding identification information is used as node identification information, if any one of the enterprises is a blacklist enterprise in the blacklist information, the attribute label of the node corresponding to the enterprise is set as the blacklist enterprise, otherwise, the attribute label of the node corresponding to the enterprise is set as the enterprise, if any one of the enterprise employees is the blacklist enterprise employee in the blacklist information, the attribute label of the node corresponding to the enterprise employee is set as the blacklist enterprise employee, and otherwise, the attribute label of the node corresponding to the enterprise employee is set as the enterprise employee;

In order to construct an enterprise relationship graph, in the embodiment of the present invention, the electronic device obtains and stores tax data in advance, and obtains identification information of an enterprise and identification information of an enterprise employee in the tax data, where the tax data includes enterprise information, enterprise employee information, commodity information, blacklist information, invoice information, and the like, and the identification information of the enterprise may be a tax number of the enterprise, or an enterprise name of the enterprise; the identification information of the enterprise employee may be an identity card number of the enterprise employee.

Specifically, the electronic device extracts tax data according to key fields stored in advance, acquires field information of each key field, and stores the field information in a pure text format (Comma-Separated Values, csv); the field information of the key field in the enterprise information comprises certificate numbers, job information and credit information, wherein the job information comprises job enterprises and duties, the field information of the key field in the enterprise information comprises blacklist enterprises and blacklist enterprise employees, and the field information of the key field in the invoice information comprises tax numbers of sellers and buyers, enterprise names, transaction commodities, transaction amounts and transaction time.

The electronic equipment acquires the corresponding relation between the tax number and the enterprise name of an enterprise from enterprise business information, wherein the corresponding relation between the tax number and the enterprise name of the enterprise has one-to-many corresponding relation between the tax number and the enterprise name, one-to-many corresponding relation between the tax number and the enterprise name and one-to-one corresponding relation between the tax number and the enterprise name; because outdated error information exists in the tax number and the enterprise name when the tax number and the enterprise name are in one-to-many or many-to-one correspondence, the electronic equipment screens out the one-to-one correspondence between the tax number and the enterprise name, and takes the tax number or the enterprise name in the one-to-one correspondence as identification information of a corresponding enterprise.

The electronic equipment constructs nodes of enterprises in a relation graph according to the identification information of the enterprises, takes the identification information of the enterprises as the node identification information of the nodes of the enterprises, constructs the nodes of the enterprise employees in the relation graph according to the identification information of the enterprise employees, and takes the identification information of the enterprise employees as the node identification information of the nodes of the enterprise employees.

According to pre-stored blacklist information, the electronic equipment creates an attribute label for each node, wherein the blacklist information comprises identification information of blacklist enterprises and identification information of blacklist enterprise employees; specifically, the electronic device determines that the enterprise is the blacklisted enterprise in the blacklisted information according to the identification information of the blacklisted enterprise in the blacklisted information, if the identification information of the enterprise is the same as the identification information of any blacklisted enterprise, and sets the attribute tag of the node corresponding to the enterprise as the blacklisted enterprise, otherwise, sets the attribute tag of the node corresponding to the enterprise as the enterprise; the electronic equipment determines that the enterprise employee is the blacklist enterprise employee in the blacklist information according to the identification information of the blacklist enterprise employee in the blacklist information, and if the identification information of the enterprise employee is the same as the identification information of the blacklist enterprise employee, the electronic equipment sets the attribute label of the node corresponding to the enterprise employee as the blacklist enterprise employee, and otherwise, sets the node attribute label corresponding to the enterprise employee as the enterprise employee.

For example, the electronic device sets the attribute tag of the node corresponding to the blacklist enterprise in the blacklist information to black _ nsr, and sets the attribute tags of the nodes corresponding to the other enterprises to nsr; and setting the attribute labels of the nodes corresponding to the blacklist enterprise employees in the blacklist information as black _ person, and setting the attribute labels of the nodes corresponding to the other enterprise employees as person.

The electronic equipment determines nodes marked by the identification information of the two target enterprises in a relation graph according to the identification information of the two target enterprises contained in the invoice information and connects the nodes of the two target enterprises aiming at each invoice information in the tax data, wherein the invoice information comprises the identification information of the enterprise of a seller and the identification information of the enterprise of a buyer.

The electronic equipment determines the nodes corresponding to the target enterprise employees and the nodes of the incumbent enterprises in the relation graph according to the information of the target enterprise employees and the incumbent enterprises in the tax data, and connects the nodes of the target enterprise employees and the nodes of the incumbent enterprises.

After the electronic device creates the enterprise relationship graph, the enterprise relationship graph is stored in a database, wherein the database can be a distributed graph database (janussgraph) or a network-oriented database (Neo 4 j).

As a possible implementation manner, in the embodiment of the present invention, after the node of the enterprise in the relationship graph is constructed, the electronic device may further use, as attribute information of the node of the enterprise, other information, except the tax number and the enterprise name, in the enterprise information in the tax data; after the nodes of the enterprise employees in the relationship graph are constructed, the electronic equipment can also be used as attribute information of the nodes of the enterprise employees according to other information of the enterprise employees except the certificate numbers, for example, the other information of the enterprise employee information except the certificate numbers includes the names of the enterprise employees, the duties of the enterprise employees, the time of job entry and the like.

Example 4:

to construct the enterprise relationship graph, on the basis of the foregoing embodiments, in an embodiment of the present invention, the node that connects two target enterprises according to identification information of the two target enterprises included in the invoice information includes:

And correspondingly storing preset keywords according to the identification information of the two target enterprises contained in the invoice information and the identification information of each target enterprise, wherein the preset keywords comprise three words of 'buyer' and three words of 'seller'.

Determining a preset keyword as identification information of a target enterprise of a buyer, and determining a node of the buyer of which the preset keyword is identified by the identification information of the target enterprise of the buyer in a relation graph; and determining the identification information of the target enterprise of which the preset keyword is the seller, determining the seller node of which the preset keyword is the identification information of the target enterprise of the seller in the relation map, and pointing to the buyer node from the seller node to obtain the edge connecting the seller node and the buyer node.

Setting the side of the seller node among the sides connecting the seller node and the buyer node as a selling side, setting the side of the buyer node among the sides as an entering side, and taking the transaction amount, the transaction time and the commodity name in the invoice information as the attribute information of the sides, specifically taking the transaction amount and the transaction time as the attribute information of the selling side and taking the commodity name as the attribute information of the entering side.

Example 5:

in order to construct an enterprise relationship graph, on the basis of the foregoing embodiments, in an embodiment of the present invention, the connecting the node of the target enterprise employee and the node of the incumbent enterprise according to the information of the target enterprise employee and the incumbent enterprise in the tax data includes:

and connecting nodes of the target enterprise employees to nodes of the incumbent enterprises in the relationship graph according to the first identification information of the target enterprise employees, the first identification information of the incumbent enterprises and the position information in the tax data, and taking the position information as side attribute information.

The electronic equipment searches a node of the target enterprise employee identified by the first identification information of the target enterprise employee and a node of the job-undertaking enterprise identified by the first identification information of the job-undertaking enterprise in a relation graph according to the first identification information of the target enterprise employee, the first identification information of the job-undertaking enterprise of the target enterprise employee and the job information of the target enterprise employee in the tax data, connects the node from the node of the target enterprise employee to the node of the job-undertaking enterprise, and takes the job information as edge attribute information.

For example, when the job information of the target enterprise employee is a corporate person, the edge attribute information is determined as clear _ person, when the job information of the target enterprise employee is a financial person, the edge attribute information is determined as fine _ person, and when the job information of the target enterprise employee is a tax person, the edge attribute information is determined as tax _ person.

Example 6:

in the following, a method for identifying an inauguration enterprise according to the present invention is described with a specific embodiment, where the electronic device employs an application container engine (Docker) containerization technology, implements containerization deployment on the data processing process, the graph database, and the risk model, and stores data in a cloud.

Specifically, after acquiring the tax data, the electronic equipment performs containerization deployment on the data preprocessing process, and field information of a key field in the csv file format is obtained through data preprocessing.

And performing containerization deployment on the construction process of the enterprise relational graph, and performing node construction and edge construction according to the acquired field information to obtain the constructed enterprise relational graph and store the constructed enterprise relational graph in a database.

And performing containerization deployment on a process of vectorizing and representing nodes with attribute labels as enterprises by using a risk model, wherein the risk model can be a graph neural network model or other integrated risk identification models.

And performing containerization representation on the process of identifying the risk enterprises, determining the similarity between the first vector and each second vector according to the first vector after the node vectorization representation of the enterprise by the attribute label and the prestored second vectors corresponding to the blacklist enterprise and the blacklist enterprise staff, and if any similarity is greater than a preset similarity threshold, determining that the enterprise corresponding to the second node of the first vector is the risk enterprise.

And forming a risk enterprise identification system by each container, realizing micro-service architecture of the whole system, realizing environment isolation among the containers, and realizing starting and stopping and daily operation and maintenance of the containers through scripts.

Fig. 3 is a flowchart of a micro service architecture provided in an embodiment of the present invention, where as shown in fig. 3, each box in fig. 3 represents a container, each arrow indicates a data direction, tax data is subjected to a container of data preprocessing to obtain a csv file, the csv file is input into a container constructed by an enterprise relationship graph to obtain a constructed enterprise relationship graph, a sub-enterprise relationship graph having attribute labels of a blacklist enterprise and each first node of a blacklist enterprise employee and each second node having a two-degree association with each first node is obtained from the enterprise relationship graph, the sub-enterprise relationship graph is input into a container for risk model calculation, and an output attribute label is a first vector of each second node of the enterprise; and aiming at each second node of which the attribute label is an enterprise, determining the similarity between the first vector and each second vector according to the first vector of the second node and the predetermined second vectors respectively corresponding to the blacklist enterprise and the blacklist enterprise staff, and if any similarity is greater than a preset similarity threshold value, determining that the enterprise corresponding to the second node of the first vector is an inauguration enterprise.

In the embodiment of the invention, the micro-service architecture of the risk enterprise identification system is realized on the basis of the Docker containerization technology, so that the complex environment conflict among different component modules is solved, the deployment efficiency and stability of the system are greatly improved, and the operation and maintenance cost is saved. Under the condition of the same configured server environment, the deployment efficiency of the system is improved by more than 50%, and the downtime-free duration is improved by more than 100%.

Example 7:

fig. 4 is a schematic diagram of a process for identifying an inauguration enterprise according to an embodiment of the present invention, where as shown in fig. 4, the process includes the following steps:

s401: according to the identification information of enterprises and the identification information of the enterprise employees in the tax data which are obtained in advance, nodes of the enterprises and the enterprise employees in the relational graph are constructed, the corresponding identification information is used as node identification information, if any enterprise is a blacklist enterprise in the blacklist information, the attribute label of the node corresponding to the enterprise is set as the blacklist enterprise, otherwise, the attribute label of the node corresponding to the enterprise is set as the enterprise, if any enterprise employee is the blacklist enterprise employee in the blacklist information, the attribute label of the node corresponding to the enterprise employee is set as the blacklist enterprise employee, and otherwise, the attribute label of the node corresponding to the enterprise employee is set as the enterprise employee.

S402: aiming at each invoice information in the tax data, according to identification information of two target enterprises contained in the invoice information and a preset keyword which is correspondingly stored, wherein the preset keyword comprises a buyer and a seller, a node which takes the preset keyword as the seller in the two target enterprises points to the node which takes the preset keyword as the buyer, one side of the node which takes the seller in the edge is set as a selling edge, one side of the node which takes the buyer in the edge is set as an entering edge, and the transaction amount, the transaction time and the commodity name in the invoice information are taken as attribute information of the corresponding edge.

S403: and connecting nodes of the target enterprise employees to nodes of the functional enterprises in the relation graph according to the first identification information of the target enterprise employees, the first identification information of the functional enterprises and the position information in the tax data, and taking the position information as side attribute information to obtain the constructed enterprise relation graph.

S404: and determining the attribute labels as each first node of the blacklist enterprise and the blacklist enterprise staff and the sub-enterprise relation graph of each second node which is in two-degree association with each first node based on the pre-constructed enterprise relation graph.

S405: and obtaining a first vector of each second node of the enterprise, which is output by the risk model and corresponds to the attribute label in the sub-enterprise relational graph, according to the attribute label corresponding to each second node in the sub-enterprise relational graph and the pre-stored risk model.

S406: and aiming at each second node of which the attribute label is an enterprise, determining the similarity between the first vector and each second vector according to the first vector of the second node and the second vectors respectively corresponding to the pre-determined blacklist enterprise and blacklist enterprise staff, and if any similarity is greater than a preset similarity threshold value, determining that the enterprise corresponding to the second node of the first vector is an inauguration enterprise.

Example 8:

fig. 5 is a schematic structural diagram of an inauguration enterprise identification apparatus according to an embodiment of the present invention, and as shown in fig. 5, the apparatus includes:

a determining module 501, configured to determine, based on a pre-constructed enterprise relationship graph, that an attribute label is a sub-enterprise relationship graph of each first node of a blacklist enterprise and blacklist enterprise employee and each second node that has a second degree association with each first node;

a processing module 502, configured to obtain, according to the attribute tag corresponding to each second node in the sub-enterprise relationship graph and a risk model stored in advance, a first vector, output by the risk model, of which the attribute tag is of each second node of an enterprise;

the identifying module 503 is configured to determine, for each second node of the enterprise to which the attribute tag is assigned, a similarity between the first vector of the second node and each second vector corresponding to the blacklisted enterprise and the blacklisted enterprise employee, respectively, and if any similarity is greater than a preset similarity threshold, determine that the enterprise corresponding to the second node of the first vector is an inauguration enterprise.

Further, the apparatus comprises:

the output module 504 is configured to, according to each identified first vector corresponding to each inauguration enterprise, perform normalization processing on each first vector corresponding to each inauguration enterprise to obtain each normalized first vector; and outputting the normalized module value of each first vector and the identification information of each inauguration enterprise contained in each corresponding second node according to the descending order of the normalized module value of each first vector.

Further, the apparatus comprises:

a model building module 505, configured to build nodes of enterprises and enterprise employees in a relational graph according to identification information of the enterprises and identification information of the enterprise employees in the tax data acquired in advance, where the corresponding identification information is used as node identification information, and if any one of the enterprises is a blacklist enterprise in the blacklist information, setting an attribute tag of the node corresponding to the enterprise as the blacklist enterprise, otherwise, setting an attribute tag of the node corresponding to the enterprise as the enterprise, and if any one of the enterprise employees is the blacklist enterprise employee in the blacklist information, setting an attribute tag of the node corresponding to the enterprise employee as the blacklist enterprise employee, otherwise, setting an attribute tag of the node corresponding to the enterprise employee as the enterprise employee; aiming at each invoice information in the tax data, connecting nodes of two target enterprises according to identification information of the two target enterprises contained in the invoice information; and connecting the node of the target enterprise employee with the node of the arbitrary enterprise according to the information of the target enterprise employee and the arbitrary enterprise in the tax data.

Further, the model building module 505 is specifically configured to, according to the identification information of two target enterprises included in the invoice information and the preset keywords correspondingly stored, where the preset keywords include a buyer and a seller, point a node in the two target enterprises where the preset keyword is the buyer to a node where the preset keyword is the seller, set one side of the node in the seller in the edge as a marketing edge, set one side of the node in the buyer in the edge as an entry edge, and use the transaction amount, the transaction time, and the commodity name in the invoice information as attribute information of the corresponding edge.

Further, the model building module 505 is specifically configured to connect, according to the first identification information of the target enterprise employee and the first identification information of the incumbent enterprise in the tax data and the job information, a node from the target enterprise employee to the node of the incumbent enterprise in the relationship graph, and use the job information as edge attribute information.

Example 9:

fig. 6 is a schematic structural diagram of an electronic device provided in the present application, and on the basis of the foregoing embodiments, the present application further provides an electronic device, as shown in fig. 6, including: a processor 601, a communication interface 602, a memory 603 and a communication bus 604, wherein the processor 601, the communication interface 602 and the memory 603 complete the communication with each other through the communication bus 604.

The memory 603 has stored therein a computer program which, when executed by the processor 601, causes the processor 601 to perform the steps of:

determining attribute labels as each first node of blacklist enterprises and blacklist enterprise employees and a sub-enterprise relationship graph of each second node which is in two-degree association with each first node based on a pre-constructed enterprise relationship graph;

Further, the processor 601 is further configured to perform normalization processing on each first vector corresponding to each inauguration enterprise according to each first vector corresponding to each identified inauguration enterprise, so as to obtain each normalized first vector;

Further, the process of the processor 601 specifically for constructing the enterprise relationship graph includes:

Further, the processor 601 is specifically configured to, according to the identification information of the two target businesses included in the invoice information, connect the nodes of the two target businesses include:

according to the identification information of two target enterprises and the corresponding stored preset keywords contained in the invoice information, wherein the preset keywords comprise a buyer and a seller, the node of which the preset keywords are the seller in the two target enterprises points to the node of which the preset keywords are the buyer, one side of the node of which the side is the seller is set as a selling side, one side of the node of which the side is the purchasing side is set as an entering side, and the transaction amount, the transaction time and the commodity name in the invoice information are used as the attribute information of the corresponding side.

Further, the processor 601 is specifically configured to connect the node of the target staff of the enterprise and the node of the incumbent enterprise according to the information of the target staff of the enterprise and the incumbent enterprise in the tax data, and includes:

The communication bus mentioned in the electronic device may be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The communication bus may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is shown, but this is not intended to represent only one bus or type of bus.

The communication interface 602 is used for communication between the above-described electronic apparatus and other apparatuses.

The Memory may include a Random Access Memory (RAM) or a Non-Volatile Memory (NVM), such as at least one disk Memory. Alternatively, the memory may be at least one memory device located remotely from the processor.

The Processor may be a general-purpose Processor, including a central processing unit, a Network Processor (NP), and the like; but may also be a Digital instruction processor (DSP), an application specific integrated circuit, a field programmable gate array or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or the like.

Example 10:

on the basis of the foregoing embodiments, the present application further provides a computer-readable storage medium, in which a computer program executable by a processor is stored, and when the program is run on the processor, the processor is caused to execute the following steps:

and for each second node of which the attribute tag is an enterprise, determining the similarity between the first vector and each second vector according to the first vector of the second node and the predetermined second vectors respectively corresponding to the blacklist enterprise and the blacklist enterprise staff, and if any similarity is greater than a preset similarity threshold, determining that the enterprise corresponding to the second node of the first vector is an inauguration enterprise.

Further, the method further comprises:

according to the identified first vectors corresponding to the risk enterprises, carrying out normalization processing on the first vectors corresponding to the risk enterprises to obtain normalized first vectors;

and connecting the node of the target enterprise employee with the node of the arbitrary enterprise according to the information of the target enterprise employee and the arbitrary enterprise in the tax data.

As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

It will be apparent to those skilled in the art that various changes and modifications may be made in the present application without departing from the spirit and scope of the application. Thus, if such modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is intended to include such modifications and variations as well.

Claims

1. An inauguration enterprise identification method, characterized in that the method comprises:

2. The method of claim 1, further comprising:

3. The method of claim 1, wherein the building of the enterprise relationship graph comprises:

4. The method according to claim 3, wherein the connecting the nodes of the two target businesses according to the identification information of the two target businesses contained in the invoice information comprises:

5. The method of claim 3, wherein the connecting the node of the target enterprise employee with the node of the incumbent enterprise according to the information of the target enterprise employee and the incumbent enterprise in the tax data comprises:

6. An inauguration enterprise identification apparatus, characterized in that said apparatus comprises:

and the identification module is used for determining the similarity between the first vector and each second vector according to the first vector of each second node and the predetermined second vectors respectively corresponding to the blacklist enterprises and the blacklist enterprise employees, and determining the enterprise corresponding to the second node of the first vector as an inauguration enterprise if any similarity is greater than a preset similarity threshold.

7. The apparatus of claim 6, wherein the apparatus comprises:

8. The apparatus of claim 6, wherein the apparatus comprises:

the model building module is used for building nodes of enterprises and enterprise employees in a relational graph according to enterprise identification information and enterprise employee identification information in the tax data acquired in advance, taking the corresponding identification information as node identification information, setting attribute labels of the nodes corresponding to the enterprises as blacklist enterprises if any enterprise is a blacklist enterprise in the blacklist information, otherwise setting the attribute labels of the nodes corresponding to the enterprises as the enterprises, setting the attribute labels of the nodes corresponding to the enterprise employees as the blacklist enterprise employees if any enterprise employee is the blacklist enterprise employee in the blacklist information, and otherwise setting the attribute labels of the nodes corresponding to the enterprise employees as the enterprise employees; aiming at each invoice information in the tax data, connecting nodes of two target enterprises according to identification information of the two target enterprises contained in the invoice information; and connecting the node of the target enterprise employee with the node of the arbitrary enterprise according to the information of the target enterprise employee and the arbitrary enterprise in the tax data.

9. The apparatus according to claim 8, wherein the model building module is specifically configured to, according to identification information of two target enterprises included in the invoice information and corresponding stored preset keywords, where the preset keywords include a buyer and a seller, point a node of the two target enterprises whose preset keyword is the buyer to a node of the two target enterprises whose preset keyword is the seller, set a side of the node of the seller in the edge as a selling edge, set a side of the node of the buyer in the edge as an entering edge, and take a transaction amount, a transaction time, and a commodity name in the invoice information as attribute information of the corresponding edge.

10. The apparatus of claim 8, wherein the model building module is configured to connect nodes from the node of the target enterprise employee to the node of the incumbent enterprise in the relationship graph according to the first identification information of the target enterprise employee, the first identification information of the incumbent enterprise, and the job information in the tax data, and use the job information as edge attribute information.

11. An electronic device, comprising: the system comprises a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory complete mutual communication through the communication bus;

the memory has stored therein a computer program which, when executed by the processor, causes the processor to carry out the steps of the inauguration enterprise identification method according to any of the claims 1 to 5.

12. A computer-readable storage medium, storing a computer program executable by a processor, the program, when executed on the processor, causing the processor to perform the steps of the inauguration enterprise identification method according to any of the claims 1 to 5.