Enterprise association relationship identification method, identification device, equipment and medium
Through fine-grained clustering and threshold verification of enterprise association relationships, the coarse identification and error problems of enterprise association relationship recognition in the prior art are solved, and the accuracy and effectiveness of identification are improved.
Patent Information
- Application Number
- CN202411158664.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-22
- Publication Date
- 2024-11-26
AI Technical Summary
The existing enterprise relationship identification methods can only achieve rough identification, and there are identification omissions and errors, resulting in poor recognition results and it is difficult to accurately identify in complex enterprise relationships.
By determining the target enterprise set to be clustered, extracting the associated attribute data and its importance from the enterprise attribute data according to the association type, performing clustering processing, and verifying the clustering results using preset thresholds to accurately locate the enterprise association relationship.
It realizes more fine-grained enterprise relationship recognition, reduces clustering results deviations, improves recognition effect, and avoids identification omissions and errors.
Smart Images

Figure CN119025942A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of computer technology, and in particular to a method for identifying enterprise association relationships, an identification device, an electronic device, and a computer-readable storage medium. Background Art
[0002] The relationship between enterprises refers to the direct or indirect connection between different enterprises in the process of business decision-making. In actual business scenarios, by effectively identifying the relationship between different enterprises, it is easy to find potential risk points of the business and improve risk control capabilities.
[0003] However, in the existing methods for identifying corporate relationships, only rough identification of corporate relationships can be achieved. Since the relationships between different enterprises are relatively complex, the above identification methods often have problems such as omissions and errors in the identification of corporate relationships, resulting in poor identification effects of corporate relationships. Summary of the invention
[0004] The present application provides a method, an identification device, an electronic device, and a computer-readable storage medium for identifying enterprise association relationships, which can achieve more fine-grained identification of enterprise association relationships and improve the identification effect of enterprise association relationships.
[0005] In a first aspect, the present application provides a method for identifying enterprise association relationships, which includes: determining multiple target enterprise sets to be clustered, wherein each target enterprise set includes multiple target enterprises, and different target enterprise sets correspond to different association types; for each target enterprise set, determining the association attribute data of each target enterprise and the importance of the association attribute data from multiple enterprise attribute data corresponding to each target enterprise in the target enterprise set according to the association type corresponding to the target enterprise set; clustering multiple target enterprises in the target enterprise set according to the association attribute data of each target enterprise and the importance of the association attribute data to obtain a first set clustering result of the target enterprise set; verifying the first set clustering result of the target enterprise set according to a preset clustering threshold, and obtaining a second set clustering result of the target enterprise set according to the verification result; determining the association relationship between multiple target enterprises in the multiple target enterprise sets according to the multiple second set clustering results of the multiple target enterprise sets.
[0006] In a second aspect, the present application provides an identification device for enterprise association relationships, which includes: a set determination module, used to determine multiple target enterprise sets to be clustered, wherein each target enterprise set includes multiple target enterprises, and different target enterprise sets have different corresponding association types; a clustering processing module, used to determine, for each target enterprise set, from multiple enterprise attribute data corresponding to each target enterprise in the target enterprise set according to the association type corresponding to the target enterprise set, the associated attribute data of each target enterprise and the importance of the associated attribute data; clustering the multiple target enterprises in the target enterprise set according to the associated attribute data of each target enterprise and the importance of the associated attribute data to obtain a first set clustering result of the target enterprise set; verifying the first set clustering result of the target enterprise set according to a preset clustering threshold, and obtaining a second set clustering result of the target enterprise set according to the verification result; a relationship identification module, used to determine the association relationship between multiple target enterprises in the multiple target enterprise sets according to multiple second set clustering results of the multiple target enterprise sets.
[0007] In a third aspect, the present application provides an electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores one or more computer programs executable by the at least one processor, and one or more of the computer programs are executed by the at least one processor so that the at least one processor can execute the above-mentioned method for identifying corporate association relationships.
[0008] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the above-mentioned method for identifying enterprise association relationships when executed by a processor / processing core.
[0009] The embodiment provided by the present application first determines multiple target enterprise sets to be clustered, wherein each target enterprise set includes multiple target enterprises, and different target enterprise sets correspond to different association types; then, for each target enterprise set, according to the association type corresponding to the target enterprise set, the associated attribute data of each target enterprise and the importance of the associated attribute data are determined from the multiple enterprise attribute data corresponding to each target enterprise in the target enterprise set, so that according to the associated attribute data of each target enterprise and the importance of the associated attribute data, the multiple target enterprises in the target enterprise set can be clustered to obtain a first set clustering result of the target enterprise set, and the first set clustering result of the target enterprise set is verified by a preset clustering threshold, and a second set clustering result of the target enterprise set is obtained according to the verification result; finally, according to the multiple second set clustering results of the multiple target enterprise sets, the association relationship between the multiple target enterprises in the multiple target enterprise sets is determined. From the above, it can be seen that when the embodiment of the present application is identifying enterprise association relationships, it can pre-determine the corresponding target enterprise sets to be clustered for different association types, so that targeted clustering processing can be performed according to the association type corresponding to each target enterprise set, that is, by determining the association attribute data corresponding to the association type and its importance, the target enterprise set is clustered, thereby achieving more fine-grained identification of enterprise association relationships. Moreover, after the clustering processing, the embodiment of the present application will further verify the first set clustering result according to the preset clustering threshold, so as to obtain the final second set clustering result according to the verification result, so as to determine the association relationship of multiple target enterprises according to the second set clustering result, effectively avoiding the phenomenon of incorrect identification of enterprise association relationships due to deviations in clustering results, and improving the identification effect of enterprise association relationships.
[0010] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS
[0011] The accompanying drawings are used to provide a further understanding of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. The above and other features and advantages will become more apparent to those skilled in the art by describing detailed example embodiments with reference to the accompanying drawings. In the accompanying drawings:
[0012] Figure 1 A flowchart of a method for identifying enterprise association relationships provided in an embodiment of the present application;
[0013] Figure 2A flowchart of a method for identifying enterprise association relationships provided in an embodiment of the present application;
[0014] Figure 3 A block diagram of a device for identifying enterprise association relationships provided in an embodiment of the present application;
[0015] Figure 4 A block diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0016] In order to enable those skilled in the art to better understand the technical solution of the present application, the exemplary embodiments of the present application are described below in conjunction with the accompanying drawings, including various details of the embodiments of the present application to facilitate understanding, which should be considered as merely exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present application. Similarly, for the sake of clarity and conciseness, the description of well-known functions and structures is omitted in the following description.
[0017] In the absence of conflict, the embodiments of the present application and the features therein may be combined with each other.
[0018] As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.
[0019] The terms used herein are only used to describe specific embodiments and are not intended to limit the present application. As used herein, the singular forms "a" and "the" are also intended to include plural forms, unless the context clearly indicates otherwise. It will also be understood that when the terms "including" and / or "made of" are used in this specification, the presence of the features, wholes, steps, operations, elements and / or components is specified, but the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or groups thereof is not excluded. "Connected" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect.
[0020] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art. It will also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant art and this application, and will not be interpreted as having an idealized or overly formal meaning unless explicitly defined herein.
[0021] In related technologies, only rough identification of corporate relationships can be performed. For example, related companies that have a relationship with a target company can be determined through a rule-based matching method. Since the relationships between different companies are relatively complex, the above identification method has certain limitations and is prone to omissions and errors in the identification of corporate relationships, resulting in poor identification of corporate relationships.
[0022] According to the method for identifying enterprise association relationships of the embodiment of the present application, when identifying enterprise association relationships, the corresponding target enterprise sets to be clustered can be pre-determined for different association types, so that targeted clustering processing can be performed according to the association type corresponding to each target enterprise set, that is, by determining the association attribute data corresponding to the association type and its importance, the target enterprise set is clustered, thereby achieving more fine-grained identification of enterprise association relationships. Moreover, after clustering processing, the method for identifying enterprise association relationships of the embodiment of the present application will further verify the first set clustering result obtained according to a preset clustering threshold, thereby obtaining the final second set clustering result based on the verification result, and determining the association relationships of multiple target enterprises based on the second set clustering result, effectively avoiding the phenomenon of incorrect identification of enterprise association relationships due to deviations in clustering results, and improving the identification effect of enterprise association relationships.
[0023] The method for identifying the enterprise association relationship according to the embodiment of the present application can be performed by an electronic device such as a terminal device or a server, and the terminal device can be a user equipment (UE), a mobile device, a user terminal, a terminal, a cellular phone, a cordless phone, a personal digital assistant (PDA), a handheld device, a computing device, a vehicle-mounted device, a wearable device, etc. The method can be implemented by a processor calling a computer-readable program instruction stored in a memory. The server can be an independent physical server, a server cluster composed of multiple servers, or a cloud server capable of cloud computing.
[0024] Figure 1 A flowchart of a method for identifying enterprise association relationships provided in an embodiment of the present application. Figure 1 , the method comprising:
[0025] Step S110: determining a plurality of target enterprise sets to be clustered, wherein each target enterprise set includes a plurality of target enterprises, and different target enterprise sets correspond to different association types.
[0026] The target enterprise set is a set that needs to be clustered. Each target enterprise set includes multiple target enterprises. The target enterprise is an enterprise object whose association relationship is to be determined. Specifically, the enterprise object that makes a resource application request can be used as the target enterprise, or the target enterprise can be directly obtained from a preset target enterprise library, which is not limited in the present embodiment of the application.
[0027] Each target enterprise set has a corresponding association type. The association type is used to indicate that there may be an association relationship corresponding to the association type between different target enterprises in the target enterprise set. For each association type, there is a corresponding association relationship.
[0028] It should be noted that the division of association types can be adaptively divided according to the association relationships between different enterprises in actual scenarios. For example, the association type can be divided into a total score association type corresponding to a total score association relationship, a franchise association type corresponding to a franchise association relationship, etc., and the embodiments of the present application do not limit this.
[0029] Among them, the total-to-specific relationship is used to represent that different enterprises belong to the same corporate legal person. For example, if enterprise A is a branch of enterprise B, then enterprise A and enterprise B have a total-to-specific relationship corresponding to the total-to-specific relationship type. For another example, if enterprise A and enterprise B are branches of enterprise C, then enterprise A, enterprise B, and enterprise C all have a total-to-specific relationship corresponding to the total-to-specific relationship type.
[0030] Among them, the franchise relationship is used to represent the franchise or investment relationship between different enterprises. For example, if enterprise A is a franchise institution of enterprise B, then enterprise A and enterprise B have a franchise relationship corresponding to the franchise relationship type. For another example, if enterprise A and enterprise B are franchise institutions of enterprise C, then enterprise A, enterprise B, and enterprise C all have a franchise relationship corresponding to the franchise relationship type.
[0031] The target enterprise set may be a pre-divided enterprise set. For example, the target enterprises have been pre-divided into association types in the target enterprise database, so that the target enterprise set to be clustered can be determined directly based on the target enterprise pre-division results in the target enterprise database. In the case where the target enterprise set is not pre-divided, multiple target enterprises can also be divided into corresponding target enterprise sets according to a preset set division rule, thereby determining multiple target enterprise sets to be clustered.
[0032] It should be noted that the set division rules can be adaptively set according to the actual division needs, and the embodiments of the present application do not limit this. Exemplarily, the target enterprise can be divided into the corresponding target enterprise set by judging whether the enterprise information corresponding to the target enterprise contains keywords corresponding to the association type. For example, by judging whether the enterprise name information of the target enterprise contains keywords such as "head office", "group", "branch", etc., it is determined whether the target enterprise is divided into the target enterprise set corresponding to the total-to-branch association type. For another example, by judging whether the brand name corresponding to the target enterprise matches the preset brand keyword, it is determined whether the target enterprise is divided into the target enterprise set corresponding to the franchise association type.
[0033] It should also be noted that multiple target enterprise sets may include the same target enterprise. For example, for a certain target enterprise, it may have a total-score association relationship with a target enterprise in a target enterprise set corresponding to a total-score association type, and it may also have a franchise association relationship with a target enterprise in a target enterprise set corresponding to a franchise association type, and this embodiment of the application does not limit this.
[0034] Step S120: For each target enterprise set, according to the association type corresponding to the target enterprise set, determine the associated attribute data of each target enterprise and the importance of the associated attribute data from the multiple enterprise attribute data corresponding to each target enterprise in the target enterprise set; according to the associated attribute data of each target enterprise and the importance of the associated attribute data, cluster the multiple target enterprises in the target enterprise set to obtain a first set clustering result of the target enterprise set; verify the first set clustering result of the target enterprise set according to a preset clustering threshold, and obtain a second set clustering result of the target enterprise set according to the verification result.
[0035] The enterprise attribute data is attribute data that characterizes the operating characteristics of the target enterprise. For example, the enterprise attribute data may include enterprise proprietary name data, enterprise industry classification data, enterprise geographic location data, enterprise financial data, enterprise supplier data, etc.
[0036] It should be noted that there is a corresponding relationship between multiple enterprise attribute data of different target enterprises. For example, the enterprise attribute data of target enterprise A includes enterprise attribute data a1, enterprise attribute data a2, and enterprise attribute data a3 corresponding to attribute type 1, attribute type 2, and attribute type 3, respectively. Then the enterprise attribute data of target enterprise B also includes enterprise attribute data b1, enterprise attribute data b2, and enterprise attribute data b3 corresponding to attribute type 1, attribute type 2, and attribute type 3, respectively, so that the enterprise attribute data of target enterprise A and target enterprise B correspond to each other.
[0037] Among them, the associated attribute data refers to the enterprise attribute data related to the associated type selected from the multiple enterprise attribute data corresponding to the target enterprise. The importance of the associated attribute data is used to characterize the degree of association between the associated attribute data and the associated type, which is expressed in numerical form. For example, the importance of the associated attribute data is 0.9, 0.8, etc. The higher the degree of association between the associated attribute data and the associated type, the higher the importance of the associated attribute data. For different association types, the corresponding associated attribute data may be different. For each association type, the importance of the multiple associated attribute data corresponding to it may also be different.
[0038] Among them, clustering multiple target enterprises in the target enterprise set according to the associated attribute data of each target enterprise and the importance of the associated attribute data means obtaining the to-be-clustered features of each target enterprise according to the associated attribute data and the importance of the associated attribute data. Thus, clustering is performed according to the to-be-clustered features of each target enterprise to obtain the first set clustering result of the target enterprise set.
[0039] According to the importance of the associated attribute data, the feature weight value corresponding to the feature of the associated attribute data can be determined. Specifically, the importance of the associated attribute data can be directly used as the feature weight value, or the corresponding feature weight value can be assigned to the features of different associated attribute data according to the ratio between the importances of different associated attribute data.
[0040] When there are multiple associated attribute data, the features of different associated attribute data can be fused according to the feature weight value of each associated attribute data to obtain the features to be clustered for each target enterprise. When there is only one associated attribute data, the features to be clustered for each target enterprise can be obtained by multiplying the features of the associated attribute data with the corresponding feature weight value.
[0041] In some possible implementations, due to the large number of associated attribute data of the target enterprise, the clustering features constructed according to the associated attribute data and the importance of the associated attribute data are not effective. Therefore, the target enterprise set can be clustered according to each associated attribute data group by dividing the multiple associated attribute data into multiple associated attribute data groups, and the first set clustering result of the target enterprise set is obtained according to the clustering result corresponding to each associated attribute data group. Correspondingly, clustering the multiple target enterprises in the target enterprise set according to the associated attribute data of each target enterprise and the importance of the associated attribute data may include: for each target enterprise, determining multiple groups of associated attribute data groups for clustering from the multiple associated attribute data of each target enterprise; determining the weight value of each associated attribute data in each group of associated attribute data groups according to the importance of the associated attribute data; clustering the multiple target enterprises in the target enterprise set according to the multiple associated attribute data groups of the multiple target enterprises and the weight value of each associated attribute data in the multiple associated attribute data groups to obtain multiple attribute clustering results of the target enterprise set; obtaining the first set clustering result according to the multiple attribute clustering results.
[0042] Among them, when determining the associated attribute data group, multiple associated attribute data can be divided into the same associated attribute data group by random combination, or it can be determined whether multiple associated attribute data are divided into the same associated attribute data group based on the importance of the associated attribute data. The embodiment of the present application does not limit this.
[0043] There is a corresponding relationship between the multiple associated attribute data groups of different target enterprises, that is, for any associated attribute data group of any target enterprise, other target enterprises have corresponding associated attribute data groups, and the associated attribute data in the associated attribute data groups with corresponding relationships have the same attribute type. For example, the associated attribute data group 1 for target enterprise A includes [associated attribute data corresponding to attribute type 1, associated attribute data corresponding to attribute type 2]. Then the associated attribute data group 1 for target enterprise B also includes [associated attribute data corresponding to attribute type 1, associated attribute data corresponding to attribute type 2].
[0044] The weight value of the associated attribute data refers to the feature weight value corresponding to the feature of the associated attribute data. The importance of the associated attribute data can be directly used as the weight value, or the corresponding weight value can be assigned to different associated attribute data according to the ratio between the importance of different associated attribute data in the associated attribute data group.
[0045] According to the weight value of each associated attribute data in the associated attribute data group, the feature of each associated attribute data can be fused to obtain the to-be-clustered feature corresponding to the associated attribute data group.
[0046] Among them, when clustering multiple target enterprises of the target enterprise set according to multiple associated attribute data groups, it means that clustering is performed on multiple target enterprises of the target enterprise set according to multiple associated attribute data groups with corresponding relationships among the multiple target enterprises. For any target enterprise, each of its associated attribute data groups has a corresponding associated attribute data group in other target enterprises. Therefore, for each of its associated attribute data groups, it corresponds to a clustering process, so that multiple attribute clustering results are finally obtained.
[0047] It should be noted that when the first set clustering result is obtained based on multiple attribute clustering results, the probability value of different target enterprise data belonging to the same cluster cluster can be determined through multiple attribute clustering results, thereby determining whether to divide multiple target enterprises into the same cluster cluster based on the probability value to obtain the first set clustering result.
[0048] In an embodiment of the present application, multiple associated attribute data are divided into multiple associated attribute data groups, and the target enterprise set is clustered according to the multiple associated attribute data groups to obtain multiple attribute clustering results. Finally, based on the multiple attribute clustering results, a first set clustering result of the target enterprise set is obtained, thereby realizing grouping and clustering of the target enterprise set based on multiple associated attribute data, and being able to effectively combine the data features of different associated attribute data, thereby improving the accuracy of the first set clustering result of the target enterprise set.
[0049] The first set of clustering results includes at least one clustered cluster, and each cluster includes at least one target enterprise.
[0050] The preset clustering threshold is used to verify the clustering effect of the first set clustering result. For example, the preset clustering threshold is used to verify the clustering tightness of the clustering cluster of the first set clustering result or to verify the number of target enterprises included in the clustering cluster of the first set clustering result, etc. Correspondingly, the preset clustering threshold may include a preset feature threshold for verifying the clustering tightness of the clustering cluster and a preset quantity threshold for verifying the number of target enterprises in the clustering cluster, etc., which is not limited in the embodiment of the present application.
[0051] When the first set clustering results of the target enterprise set are verified according to a preset feature threshold and the second set clustering results of the target enterprise set are obtained according to the verification results, the feature distance between each target enterprise in the cluster cluster and the cluster center can be calculated for each cluster cluster in the first set clustering results. Therefore, when the feature distance between the target enterprise and the cluster center exceeds the preset feature threshold, the target enterprise is deleted from the cluster cluster to obtain the second set clustering results of the target enterprise set.
[0052] The verification result is used to characterize whether there is a clustering deviation in the clustering clusters in the first set clustering result, so as to adjust the clustering clusters with clustering deviation in the first set clustering result according to the verification result to obtain the second set clustering result of the target enterprise set. The second set clustering result is the final clustering result of the target enterprise set.
[0053] In some possible implementations, when clustering the target enterprise set, clustering deviation may exist, for example, unrelated enterprises are grouped into the same cluster. Therefore, the first set clustering results of the target enterprise set can be verified by a preset quantity threshold, and the second set clustering results of the target enterprise set can be obtained based on the verification results. Correspondingly, the preset clustering threshold includes a preset quantity threshold, and the first set clustering result of the target enterprise set is verified according to the preset clustering threshold, and the second set clustering result of the target enterprise set is obtained according to the verification result, including: for each cluster cluster in the first set clustering result of the target enterprise set, the number of target enterprises contained in each cluster cluster is counted respectively; the number of target enterprises contained in each cluster cluster is verified according to the preset quantity threshold, and the quantity verification result of each cluster cluster is obtained; for each cluster cluster, if the quantity verification result of the cluster cluster fails, the multiple target enterprises in the cluster cluster are divided into multiple target enterprise groups to be retrieved; for each target enterprise group, according to the enterprise name information of multiple target enterprises in the target enterprise group, the associated information of whether the enterprise name information of multiple target enterprises is contained in the preset information source is retrieved; according to the search result corresponding to each target enterprise group, the cluster cluster is updated, and the updated cluster cluster is used as the cluster cluster belonging to the second set clustering result.
[0054] The preset quantity threshold is used to verify the number of target enterprises in the cluster. The preset quantity threshold can be adaptively set according to actual verification needs, and the embodiment of the present application does not limit this.
[0055] If the result of the cluster quantity verification passes, the cluster is regarded as a cluster belonging to the second set of clustering results.
[0056] The failure of the cluster quantity verification result means that the number of target enterprises included in the cluster exceeds the preset number threshold. The passing of the cluster quantity verification result means that the number of target enterprises included in the cluster does not exceed the preset number threshold.
[0057] Specifically, in actual applications, there are many companies that deliberately imitate the corporate information of a certain company to engage in malicious competition, so that although many target companies are not related, their corresponding corporate attribute data are similar. For example, "AA Real Estate Co., Ltd." and "AAA Real Estate Co., Ltd.", "AA Property Co., Ltd.", "AAB Real Estate Co., Ltd.", etc., although they are not actually related, but due to the similarity of corporate attribute data, these companies are easily classified into the same cluster. Therefore, in the clustering process, it is easy to cluster many target companies with similar corporate attribute data but not actually related into one category.
[0058] The present application verifies the number of target enterprises belonging to the same cluster through a preset quantity threshold. If the quantity verification result of the cluster fails, it indicates that the number of target enterprises belonging to the cluster is too large, that is, clustering deviation is likely to exist in the cluster. At this time, multiple target enterprises in the cluster are searched, and the correlation between the multiple target enterprises is further determined through the search results to improve the accuracy of the clustering results.
[0059] Each target enterprise group to be searched includes at least two target enterprises. When multiple target enterprises in the cluster are divided into multiple target enterprise groups to be searched, the target enterprise groups can be divided in the form of random combination, and the same target enterprise can be included in multiple target enterprise groups at the same time.
[0060] Among them, the preset information source can be an official website information source, a news report information source, or a corporate information database or other related information source, and the embodiment of the present application does not limit this.
[0061] The associated information refers to text data that contains the enterprise name information of multiple target enterprises at the same time.
[0062] It should be noted that updating the clustering cluster according to the search results corresponding to each target enterprise group may include: for each target enterprise group, when the search result is yes, extracting keywords in the association information that are used to characterize the association relationship between multiple target enterprises in the target enterprise group; determining whether the multiple target enterprises in the target enterprise group are marked enterprises belonging to the clustering cluster based on the matching results of the association type corresponding to the keyword and the association type corresponding to the target enterprise set; detecting whether the clustering cluster contains non-marked enterprises that do not belong to the clustering cluster, and if so, deleting the non-marked enterprises in the clustering cluster to obtain an updated clustering cluster.
[0063] Among them, keywords are used to characterize the association relationship between multiple target enterprises in the target enterprise group. For each association type, there are preset keywords, such as the keywords corresponding to the total-score association type can be "second-level company", "subsidiary", etc., and the keywords corresponding to the franchise association type can be "franchise", "chain", "investment", etc.
[0064] Therefore, according to the association type corresponding to the extracted keyword, it is possible to determine whether the multiple target enterprises in the target enterprise group match the association type corresponding to the target enterprise set. If they match, the multiple target enterprises in the target enterprise group are determined to be marked enterprises belonging to the cluster.
[0065] When the association type corresponding to the keyword does not match the association type corresponding to the target enterprise set, it indicates that the association relationship corresponding to the multiple target enterprises in the target enterprise group is inconsistent with the association relationship corresponding to the cluster. For example, the association relationship corresponding to the multiple target enterprises in the target enterprise group is a total-score association relationship corresponding to the total-score association type, while the association relationship corresponding to the multiple target enterprises in the cluster is a franchise association relationship corresponding to the franchise association type.
[0066] Therefore, the multiple target enterprises in the target enterprise group can be divided into clusters corresponding to the association type. Specifically, the first cluster to which the multiple target enterprises in the target enterprise group belong can be determined according to the association type corresponding to the keyword.
[0067] When determining the first cluster to which multiple target enterprises of the target enterprise group belong, it is possible to search whether any target enterprise of the target enterprise group is included in all clusters of the first set clustering results of the target enterprise set corresponding to the association type. If not, a new cluster is created for the multiple target enterprises of the target enterprise group in the first set clustering results of the target enterprise set corresponding to the association type as the first cluster. If included, the cluster containing any target enterprise of the target enterprise group is determined as the first cluster to which multiple target enterprises of the target enterprise group belong. If there are multiple first clusters, multiple first clusters can be merged. Finally, by determining whether the target enterprise group contains the remaining target enterprises that are not in the first cluster, the remaining target enterprises are added to the first cluster.
[0068] After the target enterprises are marked in the clusters, for each cluster, if it contains unmarked enterprises that do not belong to the cluster, the unmarked enterprises in the cluster are deleted so that the cluster only contains marked enterprises, thereby obtaining a more accurate clustering result.
[0069] When the result of the cluster quantity verification passes, it indicates that the cluster result of the cluster has a high credibility. Therefore, the cluster can be directly used as the cluster belonging to the second set of cluster results.
[0070] According to an embodiment of the present application, a first set clustering result of the target enterprise set is verified by a preset quantity threshold, and a second set clustering result of the target enterprise set is obtained based on the verification result, so that a more accurate clustering result can be obtained.
[0071] Step S130: determining association relationships between multiple target enterprises in the multiple target enterprise sets according to the multiple second set clustering results of the multiple target enterprise sets.
[0072] For each cluster included in the second set clustering result of each target enterprise set, it is determined that multiple target enterprises belonging to the same cluster have an association relationship, and the association relationship matches the association type corresponding to the target enterprise set. For example, if the association type corresponding to the target enterprise set is a total score association type, then multiple target enterprises in each cluster in the corresponding second set clustering result have a total score association relationship corresponding to the total score association type.
[0073] In some possible implementations, the second set clustering results of different target enterprise sets may include the same target enterprise. Therefore, in order to determine a more comprehensive association relationship between multiple target enterprises, an association graph can be constructed based on the multiple second set clustering results, so as to obtain the association relationship between the multiple target enterprises based on the association graph. Accordingly, determining the association relationship between multiple target enterprises in multiple target enterprise sets based on the multiple second set clustering results of multiple target enterprise sets may include: for each cluster cluster in the second set clustering results of each target enterprise set, according to the multiple target enterprises included in each cluster cluster, construct multiple target enterprise nodes; according to the association type corresponding to the target enterprise set, construct an association edge for any two target enterprise nodes, and obtain multiple association edges corresponding to the multiple target enterprise nodes, wherein the association edge is used to characterize that the two target enterprise nodes have an association relationship corresponding to the association type corresponding to the target enterprise set; according to the multiple target enterprise nodes and the multiple association edges corresponding to the multiple target enterprise nodes, obtain an enterprise association graph; according to the enterprise association graph, determine the association relationship between multiple target enterprises.
[0074] The target enterprise node is used to represent a target enterprise. The association edge between any two target enterprise nodes is used to represent the association relationship between the two associated target enterprises. The association relationship represented by the association edge depends on the association type corresponding to the target enterprise set to which the two associated target enterprises belong.
[0075] The enterprise association graph is composed of multiple target enterprise nodes and multiple associated edges corresponding to the multiple target enterprise nodes. Therefore, for any target enterprise, by determining the target enterprise node corresponding to the target enterprise and searching for the target enterprise node with an associated edge with the target enterprise node in the enterprise association graph, the target enterprise with an associated relationship with the any target enterprise can be obtained.
[0076] By constructing an enterprise association graph based on the clustering results of multiple second sets of multiple target enterprise sets, the association relationship between multiple target enterprises can be fully represented in the enterprise association graph. Moreover, for any target enterprise, by querying the association edge corresponding to the target enterprise node of the target enterprise in the enterprise association graph, all associated enterprises with different association relationships with the target enterprise can be quickly determined based on the association relationship represented by the association edge and another target enterprise node connected to the association edge, thereby achieving comprehensive and rapid identification of the association relationship between multiple target enterprises.
[0077] According to the embodiments of the present application, when identifying enterprise association relationships, the corresponding target enterprise sets to be clustered can be pre-determined for different association types, so that targeted clustering processing can be performed according to the association type corresponding to each target enterprise set, that is, by determining the association attribute data corresponding to the association type and its importance, the target enterprise set is clustered, thereby achieving more fine-grained identification of enterprise association relationships. In addition, after the clustering processing is performed, the embodiments of the present application will further verify the first set clustering results obtained according to the preset clustering threshold, so as to obtain the final second set clustering results based on the verification results, and determine the association relationships of multiple target enterprises based on the second set clustering results, effectively avoiding the phenomenon of incorrect identification of enterprise association relationships due to deviations in the clustering results, and improving the identification effect of enterprise association relationships.
[0078] In addition, those skilled in the art may also make various modifications and variations to the method in this application:
[0079] In some possible implementations, different association types correspond to different association attribute data and the importance of the association attribute data. Therefore, in order to improve the extraction effect of the association attribute data from the enterprise attribute data, the preset association coefficient corresponding table can be queried according to the attribute label corresponding to the enterprise attribute data, so as to determine the association attribute data of the target enterprise and the importance of the association attribute data. Accordingly, according to the association type corresponding to the target enterprise set, the association attribute data of each target enterprise and the importance of the association attribute data are determined from the multiple enterprise attribute data corresponding to each target enterprise in the target enterprise set, including: determining the attribute label corresponding to each enterprise attribute data respectively, and according to the association type and the attribute label, querying the preset association coefficient corresponding table to obtain the association coefficient between the attribute label corresponding to each enterprise attribute data and the association type; for each enterprise attribute data, according to the comparison result between the association coefficient between the attribute label corresponding to the enterprise attribute data and the association type and the preset association threshold, determining whether the enterprise attribute data belongs to the association attribute data, and obtaining the association attribute data of each target enterprise; for each association attribute data, according to the association coefficient between the attribute label corresponding to the association attribute data and the association type, determining the importance of the association attribute data.
[0080] The attribute label corresponding to the enterprise attribute data is used to characterize the attribute type corresponding to the enterprise attribute data. For example, the attribute label is a proprietary name, an industry attribute, etc.
[0081] The preset correlation coefficient corresponding table stores the correlation coefficients between the attribute labels and the correlation types corresponding to the attribute data of different enterprises. The correlation coefficient is used to characterize the correlation between the attribute labels and the correlation types. The correlation coefficient can be expressed in numerical form, for example, the correlation coefficient is 0.2, 0.5, etc.
[0082] The preset association threshold is used to measure the value of the association coefficient between the attribute label and the association type. According to the comparison result between the association coefficient and the preset association threshold, when the association coefficient exceeds the preset association threshold, the enterprise attribute data is determined to be associated attribute data.
[0083] The importance of the associated attribute data is determined based on the correlation coefficient between the attribute label corresponding to the associated attribute data and the associated type. The higher the correlation coefficient, the higher the importance of the associated attribute data. Specifically, the correlation coefficient can be directly used as the importance of the associated attribute data, or the correlation coefficient can be converted into the importance of the associated attribute data according to the correlation coefficient and a preset numerical conversion rule. This embodiment of the present application does not limit this.
[0084] In the embodiment of the present application, by querying a preset correlation coefficient correspondence table, the correlation attribute data of the target enterprise and the importance of the correlation attribute data are determined, thereby improving the extraction effect of the correlation attribute data from the enterprise attribute data.
[0085] In some possible implementations, in order to more accurately obtain the correlation coefficients corresponding to different attribute labels and association types, feature extraction can be performed on multiple associated enterprise attribute data corresponding to the preset attribute label, so as to calculate the correlation coefficients corresponding to the preset attribute label and the association type according to the extracted multiple attribute feature values. Accordingly, the preset correlation coefficient corresponding table is constructed in the following manner: for each association type, multiple preset attribute labels corresponding to the association type are determined, and multiple associated enterprise attribute data corresponding to each preset attribute label are obtained respectively; for each preset attribute label, feature extraction is performed on multiple associated enterprise attribute data corresponding to the preset attribute label respectively, and multiple attribute feature values corresponding to the preset attribute label are obtained; the average feature value between multiple attribute feature values is calculated, and the label type correlation coefficient between the preset attribute label and the association type is determined according to the feature distance between the multiple attribute feature values corresponding to the preset attribute label and the average feature value; according to the multiple label type correlation coefficients between multiple preset attribute labels and multiple association types, a preset correlation coefficient corresponding table is constructed.
[0086] The preset attribute tag is the attribute tag corresponding to the enterprise attribute data, and the associated enterprise attribute data is the enterprise attribute data of the preset associated enterprise corresponding to the preset attribute tag.
[0087] For each association type, there are multiple preset associated enterprises corresponding to each other, and there are association relationships corresponding to the association type between the multiple preset associated enterprises.
[0088] The attribute characteristic value is a characteristic value obtained by extracting the characteristics of the associated enterprise attribute data of the preset associated enterprise. Each preset attribute label corresponds to a plurality of attribute characteristic values.
[0089] By calculating the average value of multiple attribute feature values, the average feature value between the multiple attribute feature values can be obtained. By calculating the feature distance between each attribute feature value and the average feature value, the closeness of the distance between each attribute feature value and the average feature value can be determined. The higher the closeness of the distance, the stronger the correlation between the attribute feature values, that is, the larger the value of the correlation coefficient between the preset attribute label and the association type.
[0090] The tag type association coefficient is the association coefficient between the preset attribute tag and the association type. By associating and storing the preset attribute tag, the association type, and the tag type association coefficient between the preset attribute tag and the association type, a preset association coefficient corresponding table can be obtained.
[0091] In an embodiment of the present application, feature extraction is performed through multiple associated enterprise attribute data corresponding to preset attribute labels, so as to determine the correlation coefficient between the preset attribute label and the associated type based on the feature distance between multiple attribute feature values and the average feature value, thereby improving the accuracy of the correlation coefficient.
[0092] In some possible implementations, since the target enterprise may forge or conceal enterprise attribute information, in order to achieve the mutual correspondence of multiple enterprise attribute data of different target enterprises and improve the accuracy of enterprise attribute data, the corresponding enterprise attribute data can be directly extracted from the enterprise name information of the target enterprise. Accordingly, the multiple enterprise attribute data corresponding to each target enterprise in the target enterprise set are obtained in the following manner: determining the enterprise name information corresponding to each target enterprise in the target enterprise set; extracting the enterprise attribute data corresponding to the preset attribute type from the enterprise name information according to the preset sequence labeling model.
[0093] The preset attribute type is used to characterize the attribute characteristics corresponding to the enterprise attribute data. For example, the preset attribute type may include a proper name type, an industry attribute type, a geographic location type, an organizational form type, and the like.
[0094] Among them, the proprietary name type is used to identify the proprietary name of the target enterprise, the industry attribute type is used to identify the operating characteristics or industry of the target enterprise, the geographical location type is used to identify the geographical location of the target enterprise, and the organizational form type is used to identify the operating model of the target enterprise.
[0095] The preset sequence labeling model is a trained model for extracting enterprise attribute data. By inputting the enterprise name information corresponding to each enterprise into the preset sequence labeling model, enterprise attribute data corresponding to the preset attribute type can be extracted.
[0096] For example, the corporate name information of a target enterprise is "Beijing AA Culture Media Co., Ltd." and is input into a preset sequence annotation model, so that the corporate attribute data corresponding to the proper name type is AA, the corporate attribute data corresponding to the industry attribute type is culture media, the corporate attribute data corresponding to the geographical location type is Beijing, and the corporate attribute data corresponding to the organizational form type is Co., Ltd.
[0097] In an embodiment of the present application, enterprise attribute data is extracted from enterprise name information based on a preset sequence annotation model, thereby improving the accuracy and credibility of the enterprise attribute data.
[0098] In some possible implementations, in order to improve the effect of the preset sequence labeling model on extracting enterprise attribute data of different enterprise name information, the model structure of the preset sequence labeling model can include a pre-trained entity recognition model, a feature extraction layer connected to the output end of the entity recognition model, and a conditional random field layer. Accordingly, the preset sequence labeling model is trained in the following manner: the enterprise name training information is input into the pre-trained entity recognition model to obtain the first feature value corresponding to the segmentation in the enterprise name training information and the attribute type prediction label corresponding to the segmentation; the first feature value corresponding to the segmentation is input into the feature extraction layer to obtain the second feature value corresponding to the segmentation; the second feature value corresponding to the segmentation and the attribute type prediction label corresponding to the segmentation are input into the conditional random field layer to obtain the predicted segmentation corresponding to the attribute type training label; according to the comparison result between the predicted segmentation corresponding to the attribute type training label and the label segmentation corresponding to the attribute type training label, the training parameters of the feature extraction layer and the conditional random field layer are updated to obtain the preset sequence labeling model.
[0099] The enterprise name training information is an enterprise name training sample, which includes enterprise attribute data corresponding to a preset attribute type.
[0100] The pre-trained entity recognition model refers to a language model with entity recognition capability obtained through pre-training. By inputting the enterprise name training information into the pre-trained entity recognition model, the attribute type prediction label corresponding to each word segment in the enterprise name training information and the first feature value corresponding to each word segment can be obtained.
[0101] The word segmentation refers to the words contained in the enterprise name training information, which can be a single word or a combination of multiple words.
[0102] Among them, the first eigenvalue corresponding to the word segmentation is the eigenvalue corresponding to the word segmentation obtained by the preset entity recognition model for feature extraction of the word segmentation.
[0103] Among them, the predicted label of the attribute type corresponding to the word segmentation is the attribute type corresponding to the word segmentation predicted by the preset entity recognition model. For example, the predicted label of the attribute type is "North": <Geographical location, b>, "Beijing": <Geographical location, i>, "City": <Geographical location, e>. Among them, b, i, and e are used to describe the relative position relationship of the word segmentation.
[0104] Among them, the feature extraction layer is used to extract features from the first eigenvalue corresponding to the word segmentation output by the entity recognition model, so as to extract a second eigenvalue with more characteristic representativeness. The network structure of the feature extraction layer can be adaptively selected according to actual application needs. For example, a convolutional layer is selected as the feature extraction layer.
[0105] Among them, the training label of the attribute type is used to represent the attribute type corresponding to the enterprise attribute data in the enterprise name training information. For example, if the attribute type is the proper name type, industry attribute type, geographical location type, or organizational form type, then the training label of the attribute type can be proper name, industry attribute, geographical location, or organizational form.
[0106] Among them, the labeled word segmentation corresponding to the training label of the attribute type is the enterprise attribute data in the enterprise name training information that has a true corresponding relationship with the training label of the attribute type.
[0107] Among them, the predicted word segmentation corresponding to the training label of the attribute type is the predicted enterprise attribute data in the enterprise name training information that has a predicted corresponding relationship with the training label of the attribute type and is obtained by prediction through the conditional random field layer.
[0108] By jointly inputting the second eigenvalue and the predicted label of the attribute type corresponding to the word segmentation into the conditional random field layer, the conditional random field layer can combine the second eigenvalue and the predicted label of the attribute type to obtain the predicted word segmentation corresponding to the training label of the attribute type. Specifically, the conditional random field layer can determine the training label of the attribute type to which the second eigenvalue of the current word segmentation finally belongs based on the second eigenvalue corresponding to the word segmentation and the predicted label of the attribute type related to the second eigenvalue. For example, the second eigenvalue of the current word segmentation and the predicted labels of the attribute types corresponding to the second eigenvalues of the word segmentations before and after it, so as to obtain the predicted word segmentation corresponding to the training label of the attribute type.
[0109] It should be noted that the predicted word segmentation corresponding to the attribute type training label is the word segmentation with complete word meanings. When obtaining the predicted word segmentation corresponding to the attribute type training label, the word segmentation with adjacent positional relationships predicted to belong to the same attribute type training label can be merged. For example, the word segmentation "North", "Beijing", and "City" belonging to the same attribute type training label are merged to obtain the predicted word segmentation "Beijing City".
[0110] After the conditional random field layer outputs the predicted word segmentation corresponding to the attribute type training label, by comparing the labeled word segmentation corresponding to the attribute type training label with the predicted word segmentation corresponding to the attribute type training label, the training parameters of the feature extraction layer and the conditional random field layer are updated according to the comparison result, so as to improve the extraction accuracy of the model for the enterprise attribute data corresponding to different attribute types.
[0111] Among them, the comparison result can be the intersection-over-union ratio between the predicted word segmentation and the labeled word segmentation, or the accuracy rate of the predicted word segmentation compared to the labeled word segmentation. The embodiments of the present application do not limit this.
[0112] In the embodiments of the present application, by setting the model structure of the preset sequence labeling model to include a pre-trained entity recognition model, a feature extraction layer connected to the output end of the entity recognition model, and a conditional random field layer, the feature extraction layer can further extract the word segmentation features of the entity recognition model to obtain more representative word segmentation features. Furthermore, the conditional random field layer can combine the word segmentation features output by the feature extraction layer and the output of the entity recognition model to obtain the predicted attribute type label corresponding to the word segmentation, improving the accuracy of the predicted word segmentation corresponding to the finally obtained attribute type training label.
[0113] In some possible implementation manners, since there are association relationships between different enterprises, according to the association relationships between different enterprises, the resource application information corresponding to different enterprises can be associated, and thus, according to the association processing result, the potential risk of the enterprise can be determined, realizing effective risk control of the enterprise resource application. Correspondingly, after determining the association relationships between the multiple target enterprises in the multiple target enterprise sets, it further includes: determining the associated enterprises having a preset association relationship with a preset enterprise according to the association relationships between the multiple target enterprises; associating the resource application information corresponding to the associated enterprises with the resource application information corresponding to the preset enterprise to determine the total resource application amount of the preset enterprise and the associated enterprises; determining the risk score of the preset enterprise according to the total resource application amount, and generating a risk prompt message when the risk score exceeds the preset risk threshold, where the risk prompt message is used to indicate that there is an association risk in the resource applications of the preset enterprise and the associated enterprises.
[0114] The preset enterprise is a requesting enterprise that applies for resources. The preset association relationship is an association relationship corresponding to a preset association type, for example, a headquarter-branch association relationship, a franchise association relationship, etc., which is not limited in the present embodiment of the application.
[0115] After determining the association relationship among multiple target enterprises, a preset enterprise is used as a target enterprise for association relationship query, so that all target enterprises that have a preset association relationship with the preset enterprise are regarded as associated enterprises.
[0116] Among them, associating the resource application information corresponding to the associated enterprise with the resource application information corresponding to the preset enterprise means associating the resource application amount corresponding to the associated enterprise with the resource application amount corresponding to the preset enterprise, thereby obtaining the total resource application amount of the preset enterprise and the associated enterprise.
[0117] Accordingly, the risk score of the preset enterprise can be determined based on the total amount of resource applications, for example, by comparing the total amount of resource applications with the resource application threshold corresponding to the preset enterprise, and determining the risk score of the preset enterprise based on the comparison result. The risk score of the preset enterprise is used to characterize the risk level of the resource application of the preset enterprise.
[0118] Among them, the preset risk threshold is used to evaluate the risk score, so that when the risk score exceeds the preset risk threshold, risk warning information is generated to indicate that there are associated risks in the resource applications of the preset enterprise and its associated enterprises.
[0119] In the embodiment of the present application, by using the association relationship between multiple target enterprises, the resource application information corresponding to the target enterprises with the association relationship can be associated and processed, thereby achieving effective risk control of the enterprise resource application.
[0120] For ease of understanding, the technical implementation details in the above embodiment are described in detail below using a specific example:
[0121] In the related technology, matching rules can only be set based on manual experience to determine the associated enterprises that have an associated relationship with a target enterprise. Since the relationship between different enterprises is relatively complex, the above identification method has certain limitations.
[0122] 1) Reliance on human experience: The recognition of corporate relationships relies on human experience and judgment, which results in insufficient integrity and accuracy in the recognition of relationships. In addition, the recognition of relationships is easily affected by personal subjective factors, resulting in insufficient objectivity in the recognition results.
[0123] 2) Incomplete identification: Due to the large number of enterprises, manual processing cannot achieve comprehensive identification of enterprise relationships. In addition, the existing identification methods of enterprise relationships mainly focus on large companies, but for medium-sized emerging companies, there is a lack of effective relationship identification mechanisms, which can easily lead to omissions and errors in the identification of enterprise relationships, resulting in poor identification of enterprise relationships.
[0124] In order to solve the above problems, a method for identifying enterprise association relationships is proposed in this example to achieve comprehensive and accurate identification of enterprise association relationships.
[0125] Figure 2 FIG. 1 is a flow chart showing a method for identifying enterprise association relationships in this example. Figure 2 As shown, the example includes the following steps:
[0126] Step S201: determining a plurality of target enterprise sets to be clustered, wherein each target enterprise set includes a plurality of target enterprises, and different target enterprise sets correspond to different association types.
[0127] The target enterprises may be all enterprises that apply for resources. According to the preset set division rule, multiple target enterprises may be divided into corresponding target enterprise sets, thereby determining multiple target enterprise sets to be clustered.
[0128] Step S202: for each target enterprise set, according to the association type corresponding to the target enterprise set, determine the associated attribute data of each target enterprise and the importance of the associated attribute data from the multiple enterprise attribute data corresponding to each target enterprise in the target enterprise set; according to the associated attribute data of each target enterprise and the importance of the associated attribute data, cluster the multiple target enterprises in the target enterprise set to obtain a first set clustering result of the target enterprise set; verify the first set clustering result of the target enterprise set according to a preset clustering threshold, and obtain a second set clustering result of the target enterprise set according to the verification result.
[0129] Among them, the enterprise attribute data can be extracted from the enterprise name information of the target enterprise according to the preset sequence labeling model. For example, the sequence labeling model is used to decompose the unit name into enterprise attribute data corresponding to attribute types such as "property name", "geographic location", "organizational form", "industry type", and "branch".
[0130] Among them, the associated attribute data corresponding to different association types may be different, and the importance of multiple associated attribute data corresponding to each association type may also be different. For example, for the total score association type, the corresponding associated attribute data may be the enterprise attribute data corresponding to the attribute types of "proper name", "organizational form", "industry type", and "branch", and the importance of the enterprise attribute data is 0.8, 0.6, 0.6, and 0.5 respectively. For the franchise association type, the corresponding associated attribute data may be the enterprise attribute data corresponding to the attribute types of "proper name", "geographic location", "organizational form", and "industry type", and the importance of the enterprise attribute data is 0.85, 0.3, 0.8, and 0.7 respectively. Therefore, according to the associated attribute data and the importance of the associated attribute data, multiple target enterprises in the target enterprise set are clustered to obtain the first set clustering result of the target enterprise set.
[0131] Among them, after obtaining the first set clustering results, the first set clustering results can be verified according to a preset number threshold, so that when the number of target enterprises in the cluster exceeds the preset number threshold, the company name information of different target enterprises in the cluster is searched in the official website information source, news report information source and other relevant information sources, and according to the search results, it is determined whether the different target enterprises have an actual association relationship, so as to determine whether to update the cluster, and finally obtain the second set clustering results.
[0132] Among them, when it is determined that different target enterprises have no actual relationship, the different target enterprises can be added to the exclusion list to avoid repeated identification of the relationship between different target enterprises.
[0133] Step S203: determining association relationships between multiple target enterprises in the multiple target enterprise sets according to the multiple second set clustering results of the multiple target enterprise sets.
[0134] Among them, an association type watch list can be established according to the association relationship between multiple target enterprises, so that the company names of the target enterprises with the association relationship are recorded in the watch list, thereby realizing the storage of the association relationship between multiple target enterprises.
[0135] Therefore, when searching for the associated enterprises corresponding to a certain enterprise, the enterprises that have an associated relationship with the enterprise can be searched in the associated type watch list according to the associated type corresponding to the enterprise. If the enterprise is not in the associated type watch list, it can be further determined whether the enterprise is on the exclusion list. If the enterprise is not on the exclusion list, manual investigation and other methods can be adopted to determine whether there are enterprises associated with it in the associated type watch list. Among them, the associated type watch list can be adaptively updated according to the actual identification situation. For example, for a certain enterprise, it is not in the associated type watch list, but through manual investigation, it is found that it has an associated relationship with the enterprise in the associated type watch list, so the enterprise name of the enterprise is added to the associated type watch list.
[0136] Step S204: Determine, based on the association relationships among the plurality of target enterprises, associated enterprises that have a preset association relationship with the preset enterprise.
[0137] Step S205: Associating the resource application information corresponding to the associated enterprise with the resource application information corresponding to the preset enterprise to determine the total amount of resource applications of the preset enterprise and the associated enterprise.
[0138] Specifically, the resource application amount corresponding to the associated enterprises and the resource application amount corresponding to the preset enterprise may be accumulated to obtain the total resource application amount of the preset enterprise and the associated enterprises.
[0139] Step S206: Determine the risk score of the preset enterprise based on the total amount of resource applications. If the risk score exceeds the preset risk threshold, generate risk warning information. The risk warning information is used to indicate that there are associated risks in the resource applications of the preset enterprise and its associated enterprises.
[0140] Among them, when the risk score exceeds the preset risk threshold, the risk management strategy can also be referred to to process the resource application of the preset enterprise. For example, through the manual investigation mechanism, the correlation between the preset enterprise and its affiliated enterprises in terms of business entities and models can be determined, so as to determine whether there is any associated risk in the resource application of the preset enterprise and its affiliated enterprises. For example, the resource application of the preset enterprise can be directly rejected, or the threshold of the resources that can be applied for by the preset enterprise can be lowered.
[0141] Among them, when the risk score does not exceed the preset risk threshold, it can be determined that the probability of the preset enterprise being at risk is low. At this time, the resource application for the preset enterprise can be processed according to the preset process.
[0142] It can be understood that the above-mentioned various method embodiments mentioned in this application can be combined with each other to form a combined embodiment without violating the principle logic. Due to space limitations, this application will not go into details. Those skilled in the art can understand that in the above-mentioned method of the specific implementation method, the specific execution order of each step should be determined by its function and possible internal logic.
[0143] In addition, the present application also provides an enterprise relationship identification device, an electronic device, and a computer-readable storage medium, all of which can be used to implement any enterprise relationship identification method provided in the present application. The corresponding technical solutions and descriptions are referred to in the corresponding records of the method part and will not be repeated here.
[0144] Figure 3 A block diagram of a device for identifying enterprise association relationships provided in an embodiment of the present application.
[0145] Reference Figure 3 The embodiment of the present application provides a device for identifying enterprise association relationships, and the device for identifying enterprise association relationships includes:
[0146] A set determination module 31 is used to determine a plurality of target enterprise sets to be clustered, wherein each target enterprise set includes a plurality of target enterprises, and different target enterprise sets correspond to different association types;
[0147] The cluster processing module 32 is used to determine, for each target enterprise set, the associated attribute data of each target enterprise and the importance of the associated attribute data from the multiple enterprise attribute data corresponding to each target enterprise in the target enterprise set according to the associated type corresponding to the target enterprise set; perform clustering processing on multiple target enterprises in the target enterprise set according to the associated attribute data of each target enterprise and the importance of the associated attribute data to obtain a first set clustering result of the target enterprise set; verify the first set clustering result of the target enterprise set according to a preset clustering threshold, and obtain a second set clustering result of the target enterprise set according to the verification result;
[0148] The relationship identification module 33 is used to determine the association relationship between multiple target enterprises in the multiple target enterprise sets according to the multiple second set clustering results of the multiple target enterprise sets.
[0149] In some possible implementations, the determining, based on the association type corresponding to the target enterprise set, the associated attribute data of each target enterprise and the importance of the associated attribute data from multiple enterprise attribute data corresponding to each target enterprise in the target enterprise set includes: determining the attribute label corresponding to each enterprise attribute data respectively, and querying a preset association coefficient correspondence table based on the association type and the attribute label to obtain the association coefficient between the attribute label corresponding to each enterprise attribute data and the association type; for each enterprise attribute data, determining whether the enterprise attribute data belongs to associated attribute data based on a comparison result between the association coefficient between the attribute label corresponding to the enterprise attribute data and the association type and a preset association threshold, and obtaining the associated attribute data of each target enterprise; for each associated attribute data, determining the importance of the associated attribute data based on the association coefficient between the attribute label corresponding to the associated attribute data and the association type.
[0150] In some possible implementations, the preset correlation coefficient correspondence table is constructed in the following manner: for each correlation type, multiple preset attribute tags corresponding to the correlation type are determined, and multiple associated enterprise attribute data corresponding to each preset attribute tag are respectively obtained; for each preset attribute tag, feature extraction is performed on the multiple associated enterprise attribute data corresponding to the preset attribute tag to obtain multiple attribute feature values corresponding to the preset attribute tag; the average feature value between the multiple attribute feature values is calculated, and the label type correlation coefficient between the preset attribute tag and the correlation type is determined based on the feature distance between the multiple attribute feature values corresponding to the preset attribute tag and the average feature value; based on the multiple label type correlation coefficients between multiple preset attribute tags and multiple association types, a preset correlation coefficient correspondence table is constructed.
[0151] In some possible implementations, multiple enterprise attribute data corresponding to each target enterprise in the target enterprise set are obtained in the following manner: determining the enterprise name information corresponding to each target enterprise in the target enterprise set; and extracting enterprise attribute data corresponding to a preset attribute type from the enterprise name information according to a preset sequence labeling model.
[0152] In some possible implementations, the preset sequence labeling model includes a pre-trained entity recognition model, a feature extraction layer connected to the output end of the entity recognition model, and a conditional random field layer. The preset sequence labeling model is trained in the following manner: the enterprise name training information is input into the pre-trained entity recognition model to obtain the first feature value corresponding to the segmentation in the enterprise name training information and the attribute type prediction label corresponding to the segmentation; the first feature value corresponding to the segmentation is input into the feature extraction layer to obtain the second feature value corresponding to the segmentation; the second feature value corresponding to the segmentation and the attribute type prediction label corresponding to the segmentation are input into the conditional random field layer to obtain the predicted segmentation corresponding to the attribute type training label; according to the comparison result between the predicted segmentation corresponding to the attribute type training label and the label segmentation corresponding to the attribute type training label, the training parameters of the feature extraction layer and the conditional random field layer are updated to obtain the preset sequence labeling model.
[0153] In some possible implementations, the preset clustering threshold includes a preset quantity threshold, and the first set clustering result of the target enterprise set is verified according to the preset clustering threshold, and the second set clustering result of the target enterprise set is obtained according to the verification result, including: for each cluster cluster in the first set clustering result of the target enterprise set, the number of target enterprises contained in each cluster cluster is counted respectively; the number of target enterprises contained in each cluster cluster is verified according to the preset quantity threshold, and the quantity verification result of each cluster cluster is obtained; for each cluster cluster, if the quantity verification result of the cluster cluster fails, the multiple target enterprises in the cluster cluster are divided into multiple target enterprise groups to be retrieved, and for each target enterprise group, according to the enterprise name information of the multiple target enterprises in the target enterprise group, the associated information of whether the enterprise name information of the multiple target enterprises is contained in the preset information source is retrieved; according to the search result corresponding to each target enterprise group, the cluster cluster is updated, and the updated cluster cluster is used as the cluster cluster belonging to the second set clustering result.
[0154] In some possible implementations, after the relationship identification module determines the relationship between multiple target enterprises in the set of multiple target enterprises, the device is also used to: determine the associated enterprises that have a preset relationship with the preset enterprise based on the relationship between the multiple target enterprises; associate the resource application information corresponding to the associated enterprises with the resource application information corresponding to the preset enterprise to determine the total amount of resource applications of the preset enterprise and the associated enterprises; determine the risk score of the preset enterprise based on the total amount of resource applications, and generate risk warning information when the risk score exceeds a preset risk threshold, wherein the risk warning information is used to characterize the existence of associated risks in the resource applications of the preset enterprise and the associated enterprises.
[0155] Each module in the above-mentioned enterprise association recognition device can be implemented in whole or in part by software, hardware, or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each of the above modules.
[0156] Figure 4 A block diagram of an electronic device provided in an embodiment of the present application.
[0157] Reference Figure 4 An embodiment of the present application provides an electronic device, which includes: at least one processor 401; at least one memory 402, and one or more I / O interfaces 403, connected between the processor 401 and the memory 402; wherein the memory 402 stores one or more computer programs that can be executed by at least one processor 401, and the one or more computer programs are executed by at least one processor 401 so that the at least one processor 401 can execute the above-mentioned method for identifying corporate association relationships.
[0158] Each module in the above electronic device can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.
[0159] The embodiment of the present application also provides a computer-readable storage medium on which a computer program is stored, wherein the computer program implements the above-mentioned enterprise association recognition method when executed by a processor / processing core. The computer-readable storage medium can be a volatile or non-volatile computer-readable storage medium.
[0160] An embodiment of the present application also provides a computer program product, including a computer-readable code, or a non-volatile computer-readable storage medium carrying the computer-readable code. When the computer-readable code runs in a processor of an electronic device, the processor in the electronic device executes the above-mentioned method for identifying enterprise association relationships.
[0161] It will be appreciated by those skilled in the art that all or some of the steps, systems, and functional modules / units in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. In a hardware implementation, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed by several physical components in cooperation. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or may be implemented as hardware, or may be implemented as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable storage medium, which may include a computer storage medium (or a non-transitory medium) and a communication medium (or a temporary medium).
[0162] As is known to those of ordinary skill in the art, the term computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable program instructions, data structures, program modules or other data). Computer storage media include, but are not limited to, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), static random access memory (SRAM), flash memory or other memory technology, portable compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. In addition, it is known to those of ordinary skill in the art that communication media typically contain computer-readable program instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transmission mechanism, and may include any information delivery medium.
[0163] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, optical fiber transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.
[0164] The computer program instructions for performing the operation of the present application can be assembler instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages, such as Smalltalk, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. Computer-readable program instructions can be executed completely on a user's computer, partially on a user's computer, executed as an independent software package, partially on a user's computer, partially on a remote computer, or completely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, using an Internet service provider to connect through the Internet). In some embodiments, by using the state information of a computer-readable program instruction to personalize an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA) or a programmable logic array (PLA), the electronic circuit can execute a computer-readable program instruction, thereby realizing various aspects of the present application.
[0165] The computer program product described herein may be implemented in hardware, software, or a combination thereof. In one optional embodiment, the computer program product is embodied as a computer storage medium, and in another optional embodiment, the computer program product is embodied as a software product, such as a software development kit (SDK), etc.
[0166] Various aspects of the present application are described herein with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer-readable program instructions.
[0167] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, thereby producing a machine, so that when these instructions are executed by the processor of the computer or other programmable data processing device, a device that implements the functions / actions specified in one or more boxes in the flowchart and / or block diagram is generated. These computer-readable program instructions can also be stored in a computer-readable storage medium, and these instructions cause the computer, programmable data processing device, and / or other equipment to work in a specific manner, so that the computer-readable medium storing the instructions includes a manufactured product, which includes instructions for implementing various aspects of the functions / actions specified in one or more boxes in the flowchart and / or block diagram.
[0168] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operating steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to implement the functions / actions specified in one or more boxes in the flowchart and / or block diagram.
[0169] The flow chart and block diagram in the accompanying drawings show the possible architecture, function and operation of the system, method and computer program product according to multiple embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a part of a module, program segment or instruction, and a part of the module, program segment or instruction includes one or more executable instructions for realizing the logical function of the specification. In some alternative implementations, the function marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two continuous square boxes can actually be executed substantially in parallel, and they can also be executed in the opposite order sometimes, depending on the function involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be realized by a dedicated hardware-based system that performs the function or action of the specification, or can be realized by a combination of special-purpose hardware and computer instructions.
[0170] Example embodiments have been disclosed herein, and although specific terms are employed, they are used and should be interpreted only in a general illustrative sense and not for limiting purposes. In some instances, it will be apparent to those skilled in the art that, unless otherwise expressly noted, features, characteristics, and / or elements described in conjunction with a particular embodiment may be used alone or in combination with features, characteristics, and / or elements described in conjunction with other embodiments. Therefore, those skilled in the art will appreciate that various changes in form and detail may be made without departing from the scope of the present application as set forth in the appended claims.
Claims
1. A method for identifying corporate association relationships, characterized in that: The method comprises: Determine multiple target enterprise sets to be clustered, wherein each target enterprise set includes multiple target enterprises, and different target enterprise sets correspond to different association types; For each target enterprise set, according to the association type corresponding to the target enterprise set, from multiple enterprise attribute data corresponding to each target enterprise in the target enterprise set, determine the associated attribute data of each target enterprise and the importance of the associated attribute data; according to the associated attribute data of each target enterprise and the importance of the associated attribute data, perform clustering processing on multiple target enterprises in the target enterprise set to obtain a first set clustering result of the target enterprise set; verify the first set clustering result of the target enterprise set according to a preset clustering threshold, and obtain a second set clustering result of the target enterprise set according to the verification result; According to the clustering results of the second sets of the target enterprise sets, the association relationship between the target enterprises in the target enterprise sets is determined.
2. The method according to claim 1, characterized in that: The determining, according to the association type corresponding to the target enterprise set, from a plurality of enterprise attribute data corresponding to each target enterprise in the target enterprise set, the associated attribute data of each target enterprise and the importance of the associated attribute data comprises: Determine the attribute label corresponding to each enterprise attribute data respectively, and query the preset association coefficient corresponding table according to the association type and the attribute label to obtain the association coefficient between the attribute label corresponding to each enterprise attribute data and the association type; For each enterprise attribute data, determining whether the enterprise attribute data belongs to associated attribute data according to a comparison result between an associated coefficient between an attribute label corresponding to the enterprise attribute data and the associated type and a preset associated threshold, and obtaining associated attribute data for each target enterprise; For each associated attribute data, the importance of the associated attribute data is determined according to the association coefficient between the attribute label corresponding to the associated attribute data and the association type.
3. The method according to claim 2, characterized in that The preset correlation coefficient correspondence table is constructed in the following manner: For each association type, determining a plurality of preset attribute tags corresponding to the association type, and respectively obtaining a plurality of associated enterprise attribute data corresponding to each preset attribute tag; For each preset attribute tag, feature extraction is performed on a plurality of associated enterprise attribute data corresponding to the preset attribute tag to obtain a plurality of attribute feature values corresponding to the preset attribute tag; Calculating an average characteristic value between the multiple attribute characteristic values, and determining a label type association coefficient between the preset attribute label and the association type according to a characteristic distance between the multiple attribute characteristic values corresponding to the preset attribute label and the average characteristic value; According to a plurality of label type correlation coefficients between a plurality of preset attribute labels and a plurality of correlation types, a preset correlation coefficient correspondence table is constructed.
4. The method according to claim 1, characterized in that: The plurality of enterprise attribute data corresponding to each target enterprise in the target enterprise set is obtained in the following manner: Determine the enterprise name information corresponding to each target enterprise in the target enterprise set; According to a preset sequence labeling model, enterprise attribute data corresponding to a preset attribute type is extracted from the enterprise name information.
5. The method according to claim 4, characterized in that The preset sequence labeling model includes a pre-trained entity recognition model, a feature extraction layer connected to the output end of the entity recognition model, and a conditional random field layer. The preset sequence labeling model is trained in the following manner: Input the enterprise name training information into the pre-trained entity recognition model to obtain the first feature value corresponding to the segmentation in the enterprise name training information and the attribute type prediction label corresponding to the segmentation; Inputting the first feature value corresponding to the word segmentation into the feature extraction layer to obtain the second feature value corresponding to the word segmentation; Inputting the second feature value corresponding to the word segmentation and the attribute type prediction label corresponding to the word segmentation into the conditional random field layer to obtain the predicted word segmentation corresponding to the attribute type training label; According to the comparison result between the predicted word segmentation corresponding to the attribute type training label and the label word segmentation corresponding to the attribute type training label, the training parameters of the feature extraction layer and the conditional random field layer are updated to obtain a preset sequence labeling model.
6. The method according to claim 1, characterized in that The preset clustering threshold includes a preset quantity threshold, and the first set clustering result of the target enterprise set is verified according to the preset clustering threshold, and the second set clustering result of the target enterprise set is obtained according to the verification result, including: For each cluster in the first set clustering result of the target enterprise set, respectively counting the number of target enterprises included in each cluster; Verifying the number of target enterprises included in each cluster according to the preset number threshold, and obtaining a number verification result for each cluster; For each cluster, if the number verification result of the cluster fails, the multiple target enterprises in the cluster are divided into multiple target enterprise groups to be retrieved; for each target enterprise group, based on the enterprise name information of the multiple target enterprises in the target enterprise group, whether the preset information source contains the associated information of the enterprise name information of the multiple target enterprises; based on the search results corresponding to each target enterprise group, the cluster is updated, and the updated cluster is used as the cluster belonging to the second set clustering result.
7. The method according to claim 1, characterized in that After determining the association relationship between the multiple target enterprises in the multiple target enterprise sets, the method further includes: Determining, according to the association relationships among the plurality of target enterprises, an associated enterprise having a preset association relationship with the preset enterprise; Associating the resource application information corresponding to the associated enterprise with the resource application information corresponding to the preset enterprise to determine the total amount of resource applications of the preset enterprise and the associated enterprise; The risk score of the preset enterprise is determined based on the total amount of resource applications. When the risk score exceeds a preset risk threshold, risk warning information is generated. The risk warning information is used to characterize the existence of associated risks in the resource applications of the preset enterprise and the associated enterprises.
8. A device for identifying corporate association relationships, characterized in that: include: A set determination module is used to determine multiple target enterprise sets to be clustered, wherein each target enterprise set includes multiple target enterprises, and different target enterprise sets correspond to different association types; A cluster processing module is used to determine, for each target enterprise set, from multiple enterprise attribute data corresponding to each target enterprise in the target enterprise set according to the association type corresponding to the target enterprise set, the associated attribute data of each target enterprise and the importance of the associated attribute data; perform cluster processing on multiple target enterprises in the target enterprise set according to the associated attribute data of each target enterprise and the importance of the associated attribute data to obtain a first set clustering result of the target enterprise set; verify the first set clustering result of the target enterprise set according to a preset clustering threshold, and obtain a second set clustering result of the target enterprise set according to the verification result; The relationship identification module is used to determine the association relationship between multiple target enterprises in the multiple target enterprise sets according to the multiple second set clustering results of the multiple target enterprise sets.
9. An electronic device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores one or more computer programs that can be executed by the at least one processor, and the one or more computer programs are executed by the at least one processor to enable the at least one processor to execute the method for identifying enterprise association relationships as described in any one of claims 1-7.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the computer program implements the method for identifying enterprise association relationships as described in any one of claims 1 to 7.
