CN-121998753-A - Enterprise risk updating method, enterprise risk updating device, enterprise risk updating equipment, enterprise risk updating medium and enterprise risk updating program product

CN121998753ACN 121998753 ACN121998753 ACN 121998753ACN-121998753-A

Abstract

The application provides an enterprise risk updating method which can be applied to the technical field of big data. The method comprises the steps of responding to the dynamic data of a monitored target enterprise, retrieving a target entity, obtaining temporal attribute data of the target entity, carrying out comparison analysis based on the temporal attribute data and the dynamic data to obtain a change index, retrieving an association graph taking the target entity as a central entity based on a preset hop count, carrying out information transmission based on the association graph through a preset information transmission model, transmitting risk information in the change index from the central entity to other enterprise entities in the association graph to obtain transmission information of the other enterprise entities, carrying out risk prediction based on the change index or the transmission information of the enterprise to be processed through a multidimensional risk assessment model to obtain a risk increment, and calculating the sum of the risk increment and the current risk probability to obtain updated risk probability. The application also provides an enterprise risk updating device, equipment, a storage medium and a program product.

Inventors

WANG RUIBING
CAI SHIYUAN
YAN WENQIAN

Assignees

中国工商银行股份有限公司

Dates

Publication Date: 20260508
Application Date: 20260123

Claims (14)

1. A method for enterprise risk updating, the method comprising: In response to monitoring dynamic data of a target enterprise, searching a target entity corresponding to the target enterprise in a pre-constructed temporal knowledge graph, acquiring temporal attribute data of the target entity, and carrying out comparative analysis based on the temporal attribute data and the dynamic data to obtain a variation index, wherein the dynamic data indicates variation information of an operating condition or public opinion, and the temporal attribute data comprises a plurality of attributes corresponding to different times; Based on a preset hop count, a correlation diagram taking the target entity as a central entity is searched in the temporal knowledge graph, information is transmitted based on the correlation diagram through a preset information transmission model, so that risk information in the change index is transmitted from the central entity to other business entities in the correlation diagram to obtain transmission information of the other business entities, and the correlation diagram comprises a plurality of business entities associated with the central entity; Under the condition that the enterprise entity of the association diagram is matched with any enterprise to be processed, performing risk prediction based on the change index or the propagation information of the enterprise to be processed through a pre-trained multidimensional risk assessment model to obtain a risk increment; Acquiring the current risk probability of the enterprise to be processed, calculating the sum of the risk increment and the current risk probability, and obtaining updated risk probability so as to determine the processing result of the enterprise to be processed according to the updated risk probability.
2. The method of claim 1, wherein the comparing the temporal attribute data with the dynamic data to obtain a variation index comprises: Under the condition that the dynamic data indicate the operation condition, extracting operation indexes in the dynamic data, calculating a historical average value of the operation indexes corresponding to the temporal attribute data, and calculating the difference between the operation indexes and the historical average value to obtain absolute fluctuation amplitude; Calculating the ratio of the absolute fluctuation amplitude to the historical mean value to obtain a relative fluctuation amplitude, and taking the relative fluctuation amplitude as a change index, wherein the operation index comprises at least one of profit margin, business income increase rate, net profit increase rate and asset liability rate.
3. The method of claim 1, wherein said comparing based on said temporal attribute data and said dynamic data to obtain a variation index further comprises: under the condition that the dynamic data indicate the operation condition, extracting operation indexes in the dynamic data, extracting historical indexes corresponding to the operation indexes in the temporal attribute data, and sorting the historical indexes and the operation indexes according to time to obtain a time sequence index sequence; and carrying out trend prediction based on the time sequence index sequence through a time convolution network to obtain an index trend, and taking the index trend as a variation index.
4. The method of claim 1, wherein the comparing the temporal attribute data with the dynamic data to obtain a variation index comprises: Under the condition that the dynamic data indicate public opinion information, carrying out emotion tendency identification on the dynamic data through a preset public opinion analysis model to obtain emotion tendency scores; based on preset weights, carrying out weighted summation on click quantity and forwarding quantity in the dynamic data to obtain propagation quantity, and normalizing the propagation quantity to obtain propagation heat; And taking the emotion tendency score and the transmission heat as change indexes.
5. The method of claim 1, wherein pre-constructing the temporal knowledge-graph comprises: Performing entity extraction on the collected multidimensional data by utilizing a pre-trained entity extraction model to obtain an entity set, wherein the multidimensional data comprises at least one of enterprise business registration information, enterprise financial reports, enterprise public opinion information and enterprise business reporting materials; For any entity in the entity set, acquiring an entity data set related to the entity in the multidimensional data, dividing the entity data set into a plurality of time sequence subsets according to time information, and respectively extracting features of the time sequence subsets to obtain entity features; And identifying the relationship types among the entities by utilizing a pre-trained relationship identification model to obtain an entity relationship triplet set, and constructing a knowledge graph according to the triplet set to obtain a temporal knowledge graph, wherein the relationship types comprise at least one of supply chain association, guarantee association or stock right association.
6. The method according to claim 1, wherein the performing information propagation based on the association graph through a preset information propagation model to transfer risk information in the change index from the central entity to other business entities in the association graph to obtain propagation information of the other business entities includes: determining a first side weight corresponding to a first relation type in a preset information propagation model based on the first relation type between the central entity and a business entity associated with a first jump in the association diagram; Information transmission is carried out on the change index based on the first side weight, so that transmission information of a first-hop enterprise entity is obtained; And acquiring a second relation type between the current enterprise entity and the enterprise entity associated with the next hop in any hop after the first hop, determining a second side weight corresponding to the second relation type in the information propagation model, and carrying out information propagation on the propagation information of the current enterprise entity based on the second side weight to obtain the propagation information of the enterprise entity of the next hop.
7. The method of claim 6, wherein information dissemination of dissemination information of the current business entity is based on the second edge weights, obtaining the propagation information of the next-hop enterprise entity comprises the following steps: Based on the edge weight, carrying out linear transformation on the propagation information of the current enterprise entity to obtain transformation information; and determining an attenuation factor according to the distance between the next-hop enterprise entity and the central entity, and obtaining the propagation information of the next-hop enterprise entity according to the product of the attenuation factor and the transformation information.
8. The method of claim 1, wherein the obtaining the current risk probability for the enterprise to be processed comprises: Responding to the received risk question and answer questions of an enterprise to be processed, and searching in a preset temporal knowledge graph based on the risk questions to obtain search results, wherein the search results comprise nodes and/or edges related to the risk question and answer questions in the temporal knowledge graph and associated information of the nodes and/or edges; and extracting features of the search result to obtain multi-dimensional risk features, carrying out risk prediction based on the multi-dimensional risk features through the multi-dimensional risk assessment model to obtain initial risk probability of the enterprise to be processed, and taking the initial risk probability as current risk probability.
9. The method of claim 8, wherein pre-training the multi-dimensional risk assessment model comprises: Repeatedly training a preset risk prediction network based on a sample multidimensional risk feature and a corresponding sample risk probability which are obtained in advance until a fusion loss value of the risk prediction network reaches a preset loss threshold value, and taking the optimal risk prediction network as a multidimensional risk assessment model; The fusion loss value acquisition method comprises the following steps: performing risk prediction based on sample multidimensional risk characteristics through the risk prediction network to obtain a first predicted value; Randomly grouping the sample multidimensional risk features to obtain a first sub-feature and a second sub-feature, and carrying out risk prediction based on the first sub-feature through the risk prediction model to obtain a second predicted value; performing risk prediction based on the second sub-feature through the risk prediction model to obtain a third predicted value, and calculating the sum of the second predicted value and the third predicted value to obtain a fourth predicted value; Calculating a loss value used for representing the first predicted value and the sample risk probability to obtain a full-scale predicted loss value; And carrying out weighted summation on the full-quantity predicted loss value and the component predicted loss value to obtain a fusion loss value.
10. The method according to claim 1, wherein the risk prediction by the pre-trained multidimensional risk assessment model based on the change index or the propagation information of the enterprise to be processed, and obtaining the risk increment comprises: based on a preset time interval, information aggregation is carried out on a plurality of propagation information or variation indexes of an enterprise to be processed, so that aggregated information is obtained; and carrying out risk prediction on the aggregation information through the multidimensional risk assessment model to obtain a risk increment.
11. An enterprise risk updating apparatus, the apparatus comprising: The dynamic comparison analysis module is used for responding to the monitored dynamic data of a target enterprise, searching a target entity corresponding to the target enterprise in a pre-constructed temporal knowledge graph, acquiring the temporal attribute data of the target entity, and carrying out comparison analysis based on the temporal attribute data and the dynamic data to obtain a variation index, wherein the dynamic data indicates the variation information of the operating condition or public opinion, and the temporal attribute data comprises a plurality of attributes corresponding to different times; The information propagation module is used for searching an association graph taking the target entity as a central entity in the temporal knowledge graph based on a preset hop count, and performing information propagation based on the association graph through a preset information propagation model so as to transmit risk information in the change index from the central entity to other business entities in the association graph to obtain propagation information of the other business entities, wherein the association graph comprises a plurality of business entities associated with the central entity; A risk increment evaluation module for performing risk prediction based on the change index or the propagation information of any enterprise to be processed through a pre-trained multidimensional risk evaluation model under the condition that the enterprise entity of the association diagram is matched with any enterprise to be processed, to obtain a risk increment, and The risk updating module is used for acquiring the current risk probability of the enterprise to be processed, calculating the sum of the risk increment and the current risk probability, and obtaining updated risk probability so as to determine the processing result of the enterprise to be processed according to the updated risk probability.
12. An electronic device, comprising: One or more processors; a memory for storing one or more computer programs, Characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1-10.
13. A computer-readable storage medium, on which a computer program or instructions is stored, which, when executed by a processor, carries out the steps of the method according to any one of claims 1 to 10.
14. A computer program product comprising a computer program or instructions which, when executed by a processor, implement the steps of the method according to any one of claims 1 to 10.

Description

Enterprise risk updating method, enterprise risk updating device, enterprise risk updating equipment, enterprise risk updating medium and enterprise risk updating program product Technical Field The application relates to the technical field of big data, in particular to an enterprise risk updating method, an enterprise risk updating device, enterprise risk updating equipment, enterprise risk updating media and enterprise risk updating program products. Background In the process of applying for transacting financial business by enterprises, efficient risk identification of enterprises by financial institutions is a key barrier for guaranteeing asset quality and preventing financial risks. The transaction mode of some financial services is that an enterprise client submits materials, a financial institution performs preliminary risk detection based on the materials, after the preliminary risk detection is passed, the financial institution still needs to pass through the approval of a plurality of flow nodes, or the approval needs to be delayed for a period of time due to time factors, then the result of the transaction or the result of the non-transaction is finally given, and after the result of the transaction is given, the relevant money of the service is issued to the enterprise client, so that the financial property safety is ensured. At present, under the condition of multiple process node approval or delayed approval, the financial institution finally carries out empirical judgment and gives a judgment result according to the initial static material submitted by the customer by an approver, and the decision mode is difficult to dynamically track the risk change of the customer, for example, some risk events occur after the material is submitted by the enterprise, the existing risk identification method cannot capture the risks in real time, and because the association relationship possibly exists among multiple enterprises applying for transacting the financial service, the dynamic risk event of one enterprise can be affected by other enterprises, the real-time risk caused by the update of the dynamic risk event of the enterprise cannot be effectively identified among the enterprises by the existing risk detection method, so that the problems of key risk signal capture lag, abnormal variation cannot be quickly identified, risk assessment according to the one-piece and the like are generated, the financial institution cannot timely early warn potential risks before issuing service money, and the property safety of the financial institution cannot be effectively ensured. Disclosure of Invention In view of the foregoing, the present application provides an enterprise risk updating method, apparatus, device, medium, and program product. According to the first aspect of the application, an enterprise risk updating method is provided, which comprises the steps of responding to the monitoring of dynamic data of a target enterprise, searching out a target entity corresponding to the target enterprise in a pre-built temporal knowledge graph, acquiring temporal attribute data of the target entity, performing comparative analysis based on the temporal attribute data and the dynamic data to obtain a change index, wherein the dynamic data indicates change information of operating conditions or public opinion, the temporal attribute data comprises a plurality of attributes corresponding to different times, searching out an association graph taking the target entity as a center entity in the temporal knowledge graph based on a preset hop count, performing information propagation based on the association graph through a preset information propagation model to transfer risk information in the change index from the center entity to other enterprise entities in the association graph, obtaining propagation information of the other enterprise entities, wherein the association graph comprises a plurality of enterprise entities associated with the center entity, predicting risks based on the change index or the propagation information to be processed through a pre-trained multidimensional risk assessment model under the condition that the enterprise entities of the association graph are matched with any enterprise to be processed, obtaining a risk to be processed, and obtaining a risk to be processed probability, and updating the probability and a risk to be processed risk increment according to a current risk processing probability. According to the embodiment of the application, the comparison analysis is performed based on the temporal attribute data and the dynamic data, and the obtaining of the change index comprises the steps of extracting the operation index in the dynamic data under the condition that the dynamic data indicates the operation condition, calculating the historical mean value of the corresponding operation index in the temporal attribute data, calculating the difference between the operation index and