A knowledge graph-based method and system for identifying key personnel and departments

Abstract

The application discloses a key personnel and department identification method and system based on a knowledge graph, and belongs to the technical field of big data. Internal service visit data of customer service is acquired, and an organization customer service relationship topology graph is constructed. External information data is acquired and combined with the internal service visit data, and a full-amount organization customer service relationship knowledge graph is constructed. According to customer service content, personnel and departments in the full-amount organization customer service relationship knowledge graph are explored. Graph algorithms are used to calculate the importance of various nodes in the full-amount organization customer service relationship knowledge graph, and key personnel and departments are identified. Through construction of the full-amount organization customer service relationship knowledge graph, the internal service visit data and the external information data are integrated, and the completeness and accuracy of data quality are ensured. Through combination of graph algorithms and stratified sampling, not only computing resources are saved, but also the difficulty of searching is reduced.

Description

Technical Field

[0001] This invention relates to the field of big data technology, and in particular to a method and system for identifying key personnel and key departments based on knowledge graphs. Background Technology

[0002] The rise and widespread adoption of the Internet, the Internet of Things, and big data are profoundly changing the current financial ecosystem and landscape. With the continuous empowerment of fintech, financial institutions are undergoing digital transformation, applying emerging and cutting-edge technologies such as data governance, AI, knowledge graphs, and NLP to various areas of their financial operations.

[0003] Currently, there are existing technical cases related to enterprise customer relationship graphs, including enterprise relationship graphs. Enterprise relationship graphs integrate relationships such as shareholders, investors, senior executives, and guarantors, and uncover complex relationships through layer-by-layer penetration, and are applied to scenarios such as discovering the actual controller of an enterprise and the group to which an enterprise belongs.

[0004] However, existing technologies fail to integrate internal service visit information with external equity investments, suppliers, and other relationships to construct a complete institutional customer service relationship graph. Secondly, the completeness and accuracy of data quality present certain challenges. Furthermore, the computational resources and storage costs are substantial, with hundreds of millions of nodes and relationships, making the selection of key nodes for practical graph algorithm calculations challenging. Finally, obtaining the complete internal organizational structure of enterprise customers is difficult; acquiring the complete internal departmental structure of listed companies from prospectuses or quarterly / annual reports is challenging and costly, and contact information for enterprise customers is often limited to basic details such as names, making retrieval difficult.

[0005] Therefore, there is an urgent need for methods and systems for identifying key personnel and key departments based on knowledge graphs, in order to overcome the limitations of existing technologies. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for identifying key personnel and key departments based on knowledge graphs, so as to solve the problems mentioned in the background art.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0008] A knowledge graph-based method for identifying key personnel and departments includes the following steps:

[0009] Step S100: Collect internal service visit data for customer service and construct an organizational customer service relationship topology diagram;

[0010] Step S200: Obtain external information data and combine it with internal service visit data to construct a knowledge graph of customer service relationships across the entire organization;

[0011] Step S300: Based on customer service content and graph structure, explore the personnel and departments in the full-scale institutional customer service relationship knowledge graph;

[0012] Step S400: Use graph algorithms to calculate the importance of various nodes in the full-scale institutional customer service relationship knowledge graph, and identify key personnel and key departments based on the calculation results.

[0013] As a preferred embodiment, the constructed organization customer service relationship topology diagram includes:

[0014] Obtain internal service visit data for customer service, including visit records of account managers and customer service records; specifically, obtain visit records of account managers and customer service records from the organization's CRM system via API interface;

[0015] Based on organization codes, customer name variant identification rules, and unique identifiers for customer entities, customer entities are standardized and uniformly identified, and a mapping relationship is established between customer personnel and applicant company employees; based on internal customer service data and the aforementioned mapping relationship, an organizational customer service relationship topology is constructed.

[0016] It needs to be explained that, through the visit records and customer service records of account managers, customer entities are normalized to establish a unified customer identifier mapping table; regular expressions are used to match customer name variations, and a preset alias rule library is used to handle the mapping relationship between customer personnel and enterprises; the uniqueness of customer entities is verified based on the organization code certificate number; and a mapping relationship is established between customer personnel and applicant company employees.

[0017] As a preferred approach, the specific implementation process for constructing a knowledge graph of full-scale institutional customer service relationships includes:

[0018] External information data is obtained from the external data platform ECIF system through API interface. The external information data includes corporate equity investment relationships and director, supervisor and senior management appointment relationships in the entire market.

[0019] Based on the organization's customer service relationship topology, internal service visit data and external information data are integrated to define graph entity nodes, relationship edges, and attributes;

[0020] Through data synchronization, processing, and storage, a full-scale institutional customer service relationship graph is constructed, which includes 5 types of entity nodes and 10 types of relationship edges.

[0021] It's important to explain that a knowledge graph is a data structure that efficiently stores and retrieves knowledge by modeling entities and their relationships as nodes and edges in a graph. The core components of a knowledge graph include entities, relationships, and attributes. Entities represent concrete things, such as people, places, and objects, and constitute the basic units of the graph. Relationships, as connections between entities, depict the inherent connections between them. Each entity and relationship can possess specific attributes, which describe the characteristics of the entity in detail.

[0022] Based on the mapping relationships between entities and time validity labels in the institutional customer service relationship topology diagram, internal service visit data and external information data are effectively integrated. Based on the conclusions of the data survey, 5 types of nodes and 10 types of relationships are finally identified. The source data is processed synchronously through node extraction, entity alignment, and relationship extraction. Among them, the enterprise node includes nearly 90 million industrial and commercial enterprises in the entire market, including both applicant company clients and non-applicant company clients; the customer personnel node includes over 100 million shareholders and senior executives in the entire market, including employees of applicant company clients and employees of non-applicant company clients; the applicant company personnel node includes applicant company employees in the institutional CRM; the applicant company organizational structure node includes the applicant company's organizational structure within the institutional CRM system (including headquarters, business units, branches, and other departments at all levels); and the organizational structure node includes the customer's organizational structure (including headquarters, business units, branches, and other departments at all levels) extracted from customer service and visit records. Based on the applicant company's internal service visit data, it constructs subordinate, customer service, visit / reception, and customer personnel employment relationship edges; based on the applicant company's internal institutional CRM information, it constructs applicant company personnel employment and applicant company organizational structure subordinate relationship edges; and based on external information data, it constructs enterprise subordinate branches, enterprise shareholding, directors, supervisors, senior executives, and customer personnel shareholding relationship edges. By extracting internal service visit information, we can effectively combine internal and external data to construct a comprehensive institutional customer service relationship knowledge graph that integrates internal business information and contains relevant data from all industrial and commercial enterprises in the market. This effectively connects enterprise customer contacts with the applicant company's account manager and uploads the knowledge graph to the cloud database for visualization through the Hilbert graph platform.

[0023] As a preferred embodiment, the entity nodes and relation edges include:

[0024] The entity nodes include enterprise nodes, customer personnel nodes, applicant company personnel nodes, applicant company organizational structure nodes, and organizational structure nodes.

[0025] Among them, the enterprise node includes all industrial and commercial enterprises in the market, including both applicant company clients and non-applicant company clients;

[0026] The customer personnel node includes employees of the applicant company's customers and all market-shareholders and senior executives who are not employees of the applicant company's customers;

[0027] The applicant company personnel node includes the applicant company employees within the organization's CRM system;

[0028] The applicant company's organizational structure node includes the applicant company's organizational structure within the organization's CRM system. The applicant company's organizational structure includes various levels of departments such as headquarters, business units, and branches.

[0029] The organizational structure nodes include the customer's organizational structure extracted from customer service and visit records. The customer's organizational structure includes departments at all levels, such as headquarters, business units, and branches.

[0030] Based on the applicant company's internal service visit data, construct a relationship edge between subordinates, customer service, visitors / receptionists, and customer personnel employment.

[0031] Based on the applicant company's internal CRM information, construct the applicant company's personnel employment and the subordinate relationships under the applicant company's organizational structure;

[0032] Based on external information data, construct a boundary for the shareholding relationships of the company's subordinate branches, corporate shareholding, directors, supervisors, senior executives, and customer personnel.

[0033] As a preferred embodiment, the specific implementation process for identifying key personnel and key departments includes:

[0034] Import the entity nodes and relationship edge data of the entire institutional customer service relationship graph from the big data platform to the AI ​​platform;

[0035] The degree centrality algorithm with computational complexity is used to identify nodes with low complexity as super nodes. These super nodes are then marked and sent to business personnel and removed from the graph's dataset.

[0036] By using stratified sampling, sampling is performed according to the distribution characteristics of the number of various types of nodes and relation edges;

[0037] The PageRank algorithm is used to obtain the out-degree, in-degree, and PageRank values ​​of customer personnel and organizational structure nodes, and these values ​​are then imported into the Hilbert graph platform along with the entity node and relation edge data stored in the AI ​​platform's HDFS.

[0038] Based on the PageRank value distribution, multiple intervals were divided. On the full institutional customer service relationship graph, customer personnel nodes and organizational structure nodes were classified and marked with importance levels according to different colors. The importance levels include critical, important, attention, and ordinary.

[0039] It should be explained that graph algorithms include degree centrality algorithms and PageRank algorithms. When identifying key personnel and departments, based on the already constructed institutional customer service relationship graph, degree centrality and PageRank algorithms are used to calculate the importance of two types of nodes. Since the customer personnel nodes in the graph include all individual shareholders, directors, supervisors, and senior executives in the entire market, the number of nodes and relationship edges in the graph is in the hundreds of millions. The computational complexity of algorithms such as PageRank is very high, which puts enormous pressure on computing resources and memory. The solution adopted in this invention is to optimize business processes and filter out unnecessary data to reduce the data volume: the node and relationship edge data for constructing the graph are imported from a big data platform to an AI platform. Data preprocessing filters and removes redundant attributes from the node and edge sets to improve performance. In data preprocessing, a low-computational-complexity degree centrality algorithm is first used to identify super nodes. These nodes are removed from the dataset and provided to business personnel for separate analysis. After filtering out super nodes, the data volume is still huge. A stratified sampling method is used to sample according to the distribution characteristics of the number of nodes and relationship edges of each type. The PageRank algorithm is used to evaluate the importance of customer personnel and organizational structure nodes. PageRank is a directed graph random walk model where a random worker moves randomly along the graph's edges, visiting one node after another. Under certain conditions, this random walk eventually converges to a stationary distribution. The PageRank algorithm iteratively calculates the PageRank value for each node. Initially, all nodes have equal probability PageRank values. Through iterative calculations, each node's PageRank value is updated to a certain proportion of the sum of the PageRank values ​​of all nodes linked to it, plus a decay factor. This decay factor prevents infinite loops. After multiple iterations, each node's PageRank value tends to stabilize, and this stationary probability value is the node's PageRank value, which can be used to assess the node's importance.

[0040] As a preferred embodiment, the PageRank algorithm is used to obtain the out-degree, in-degree, and PageRank values ​​of customer personnel and organizational structure nodes. Its specific implementation process includes:

[0041] Initialize the PageRank values ​​of all nodes to equal probability values, and iteratively update the PageRank value of each node using the following formula:

[0042]

[0043] in, This represents the calculated PageRank value of node X. Let represent the damping coefficient, and i represent the node pointing to X. Let I represent the out-degree of node i, and let I represent the total number of nodes pointing to X, i.e., the in-degree.

[0044] The PageRank value of each node is calculated iteratively using the above formula until it converges to a stable value. The stationary probability value of each node is the final PageRank value, which is used to represent the importance of the node in the graph.

[0045] A knowledge graph-based identification system for key personnel and key departments, comprising a data acquisition module, a knowledge graph construction module, and an identification module;

[0046] The data collection module is used to collect internal service visit data and external information data for customer service.

[0047] The graph construction module combines external information data with internal service visit data to construct a comprehensive knowledge graph of customer service relationships for the entire organization.

[0048] The identification module uses graph algorithms to calculate the importance of various nodes in the full-scale institutional customer service relationship knowledge graph, and identifies key personnel and key departments based on the calculation results.

[0049] As a preferred embodiment, the acquisition module includes:

[0050] Obtain internal service visit data for customer service, including visit records of account managers and customer service records; specifically, obtain visit records of account managers and customer service records from the organization's CRM system via API interface;

[0051] External information data is obtained from the external data platform ECIF system through API interface. The external information data includes corporate equity investment relationships and director, supervisor and senior management appointment relationships in the entire market.

[0052] As a preferred embodiment, the map construction module includes:

[0053] Based on internal customer service data, customer entities are normalized and a unified customer identifier mapping table is established; regular expressions are used to match customer name variations, and a preset alias rule base is used to handle the mapping relationship between customer personnel and enterprises; the uniqueness of customer entities is verified based on the organization code certificate number; a mapping relationship between customer personnel and applicant company employees is established, and an organizational customer service relationship topology diagram is constructed.

[0054] Based on the institutional customer service relationship topology diagram, internal service visit data and external information data are integrated to define graph entity nodes, relationship edges, and attributes; the entity nodes include enterprise nodes, customer personnel nodes, applicant company personnel nodes, applicant company organizational structure nodes, and organizational structure nodes; wherein:

[0055] The enterprise node includes all industrial and commercial enterprises in the market, including both applicant company clients and non-applicant company clients;

[0056] The customer personnel node includes employees of the applicant company's customers and all market-shareholders and senior executives who are not employees of the applicant company's customers;

[0057] The applicant company personnel node includes the applicant company employees within the organization's CRM system;

[0058] The applicant company's organizational structure node includes the applicant company's organizational structure within the organization's CRM system. The applicant company's organizational structure includes various levels of departments such as headquarters, business units, and branches.

[0059] The organizational structure nodes include the customer's organizational structure extracted from customer service and visit records. The customer's organizational structure includes departments at all levels, such as headquarters, business units, and branches.

[0060] Based on the applicant company's internal service visit data, construct a relationship edge between subordinates, customer service, visitors / receptionists, and customer personnel employment.

[0061] Based on the applicant company's internal CRM information, construct the applicant company's personnel employment and the subordinate relationships under the applicant company's organizational structure;

[0062] Construct a boundary based on external information data to establish the shareholding relationships among the company's subordinate branches, corporate shareholding, directors, supervisors, senior executives, and customer personnel.

[0063] Through data synchronization, processing, and storage, a full-scale institutional customer service relationship graph is constructed, which includes 5 types of entity nodes and 10 types of relationship edges.

[0064] As a preferred embodiment, the identification module includes:

[0065] Import the entity nodes and relationship edge data of the entire institutional customer service relationship graph from the big data platform to the AI ​​platform;

[0066] The degree centrality algorithm with computational complexity is used to identify nodes with low complexity as super nodes. These super nodes are then marked and sent to business personnel and removed from the graph's dataset.

[0067] By using stratified sampling, sampling is performed according to the distribution characteristics of the number of various types of nodes and relation edges;

[0068] Initialize the PageRank values ​​of all nodes to equal probability values, and iteratively update the PageRank value of each node using the following formula:

[0069]

[0070] in, This represents the calculated PageRank value of node X. Let represent the damping coefficient, and i represent the node pointing to X. Let I represent the out-degree of node i, and let I represent the total number of nodes pointing to X, i.e., the in-degree.

[0071] The above formula is used to iterate and calculate until the PageRank value of each node converges to a stable value. The stationary probability value of each node is the final PageRank value, which is used to represent the importance of the node in the graph.

[0072] Obtain the out-degree, in-degree, and PageRank values ​​of customer personnel and organizational structure nodes, and import them along with the entity node and relation edge data stored in the AI ​​platform HDFS into the Hilbert graph platform used to query and display important values ​​of customer personnel and organizational structure.

[0073] Based on the PageRank value distribution, multiple intervals were divided. On the full institutional customer service relationship graph, customer personnel nodes and organizational structure nodes were classified and marked with importance levels according to different colors. The importance levels include critical, important, attention, and ordinary.

[0074] Compared with existing technologies, the beneficial effects achieved by this invention are as follows: The method and system for identifying key personnel and key departments based on knowledge graphs provided by this invention include: acquiring internal service visit data of customer service to construct an organizational customer service relationship topology graph; acquiring external information data and combining it with internal service visit data to construct a full-scale organizational customer service relationship knowledge graph; exploring personnel and departments in the full-scale organizational customer service relationship knowledge graph according to customer service content; using graph algorithms to calculate the importance of various nodes in the full-scale organizational customer service relationship knowledge graph to identify key personnel and key departments; ensuring the integrity and accuracy of data quality through the construction of the full-scale organizational customer service relationship knowledge graph; and saving computational resources and reducing search difficulty through the combination of graph algorithms and stratified sampling. Attached Figure Description

[0075] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0076] Figure 1 This is a schematic diagram of the method flow in an embodiment of the present invention;

[0077] Figure 2 This is a schematic diagram of the institutional customer service relationship graph structure in an embodiment of the present invention;

[0078] Figure 3 This is a data processing flowchart for key node identification in an embodiment of the present invention. Detailed Implementation

[0079] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention are within the scope of protection of the present invention.

[0080] like Figure 1 , Figure 2 , Figure 3 As shown, this is an embodiment of the present invention, which provides a method for identifying key personnel and key departments based on knowledge graphs. This method includes the following steps:

[0081] Step S100: Collect internal service visit data for customer service and construct an organizational customer service relationship topology diagram;

[0082] Step S200: Obtain external information data and combine it with internal service visit data to construct a knowledge graph of customer service relationships across the entire organization;

[0083] Step S300: Based on customer service content and graph structure, explore the personnel and departments in the full-scale institutional customer service relationship knowledge graph;

[0084] Step S400: Use graph algorithms to calculate the importance of various nodes in the full-scale institutional customer service relationship knowledge graph, and identify key personnel and key departments based on the calculation results.

[0085] Specifically, the construction of the organization's customer service relationship topology includes:

[0086] Obtain internal service visit data for customer service, including visit records of account managers and customer service records; specifically, obtain visit records of account managers and customer service records from the organization's CRM system via API interface;

[0087] Based on organization codes, customer name variant identification rules, and unique identifiers for customer entities, customer entities are standardized and uniformly identified, and a mapping relationship is established between customer personnel and applicant company employees; based on internal customer service data and the aforementioned mapping relationship, an organizational customer service relationship topology is constructed.

[0088] Specifically, the implementation process for constructing a knowledge graph of customer service relationships across all institutions includes:

[0089] External information data is obtained from the external data platform ECIF system through API interface. The external information data includes corporate equity investment relationships and director, supervisor and senior management appointment relationships in the entire market.

[0090] Based on the organization's customer service relationship topology, internal service visit data and external information data are integrated to define graph entity nodes, relationship edges, and attributes;

[0091] Through data synchronization, processing, and storage, a full-scale institutional customer service relationship graph is constructed, which includes 5 types of entity nodes and 10 types of relationship edges.

[0092] Specifically, the entity nodes and relation edges include:

[0093] The entity nodes include enterprise nodes, customer personnel nodes, applicant company personnel nodes, applicant company organizational structure nodes, and organizational structure nodes; wherein:

[0094] The enterprise node includes all industrial and commercial enterprises in the market, including both applicant company clients and non-applicant company clients;

[0095] The customer personnel node includes employees of the applicant company's customers and all market-shareholders and senior executives who are not employees of the applicant company's customers;

[0096] The applicant company personnel node includes the applicant company employees within the organization's CRM system;

[0097] The applicant company's organizational structure node includes the applicant company's organizational structure within the organization's CRM system. The applicant company's organizational structure includes various levels of departments such as headquarters, business units, and branches.

[0098] The organizational structure nodes include the customer's organizational structure extracted from customer service and visit records. The customer's organizational structure includes departments at all levels, such as headquarters, business units, and branches.

[0099] Based on the applicant company's internal service visit data, construct a relationship edge between subordinates, customer service, visitors / receptionists, and customer personnel employment.

[0100] Based on the applicant company's internal CRM information, construct the applicant company's personnel employment and the subordinate relationships under the applicant company's organizational structure;

[0101] Based on external information data, construct a boundary for the shareholding relationships of the company's subordinate branches, corporate shareholding, directors, supervisors, senior executives, and customer personnel.

[0102] Specifically, the implementation process for identifying key personnel and key departments includes:

[0103] Import the entity nodes and relationship edge data of the entire institutional customer service relationship graph from the big data platform to the AI ​​platform;

[0104] The degree centrality algorithm with computational complexity is used to identify nodes with low complexity as super nodes. These super nodes are then marked and sent to business personnel and removed from the graph's dataset.

[0105] By using stratified sampling, sampling is performed according to the distribution characteristics of the number of nodes and relation edges of various types;

[0106] The PageRank algorithm is used to obtain the out-degree, in-degree, and PageRank values ​​of customer personnel and organizational structure nodes, and these values ​​are then imported into the Hilbert graph platform along with the entity node and relation edge data stored in the AI ​​platform's HDFS.

[0107] Based on the PageRank value distribution, multiple intervals were divided. On the full institutional customer service relationship graph, customer personnel nodes and organizational structure nodes were classified and marked with importance levels according to different colors to obtain key personnel and key departments. The importance levels include critical, important, attention, and ordinary.

[0108] Specifically, the PageRank algorithm is used to obtain the out-degree, in-degree, and PageRank values ​​of customer personnel nodes and organizational structure nodes. Its specific implementation process includes:

[0109] Initialize the PageRank values ​​of all nodes to equal probability values, and iteratively update the PageRank value of each node using the following formula:

[0110]

[0111] in, This represents the calculated PageRank value of node X. Let represent the damping coefficient, and i represent the node pointing to X. Let I represent the out-degree of node i, and let I represent the total number of nodes pointing to X, i.e., the in-degree.

[0112] The PageRank value of each node is calculated iteratively using the above formula until it converges to a stable value. The stationary probability value of each node is the final PageRank value, which is used to represent the importance of the node in the graph.

[0113] A knowledge graph-based identification system for key personnel and key departments, comprising a data acquisition module, a knowledge graph construction module, and an identification module;

[0114] The data collection module is used to collect internal service visit data and external information data for customer service.

[0115] The graph construction module combines external information data with internal service visit data to construct a comprehensive knowledge graph of customer service relationships for the entire organization.

[0116] The identification module uses graph algorithms to calculate the importance of various nodes in the full-scale institutional customer service relationship knowledge graph, and identifies key personnel and key departments based on the calculation results.

[0117] Specifically, the acquisition module includes:

[0118] Obtain internal service visit data for customer service, including visit records of account managers and customer service records; specifically, obtain visit records of account managers and customer service records from the organization's CRM system via API interface;

[0119] External information data is obtained from the external data platform ECIF system through API interface. The external information data includes corporate equity investment relationships and director, supervisor and senior management appointment relationships in the entire market.

[0120] Specifically, the map construction module includes:

[0121] Based on internal customer service data, customer entities are normalized and a unified customer identifier mapping table is established; regular expressions are used to match customer name variations, and a preset alias rule base is used to handle the mapping relationship between customer personnel and enterprises; the uniqueness of customer entities is verified based on the organization code certificate number; a mapping relationship between customer personnel and applicant company employees is established, and an organizational customer service relationship topology diagram is constructed.

[0122] Based on the institutional customer service relationship topology diagram, internal service visit data and external information data are integrated to define graph entity nodes, relationship edges, and attributes; the entity nodes include enterprise nodes, customer personnel nodes, applicant company personnel nodes, applicant company organizational structure nodes, and organizational structure nodes; wherein:

[0123] The enterprise node includes all industrial and commercial enterprises in the market, including both applicant company clients and non-applicant company clients;

[0124] The customer personnel node includes employees of the applicant company's customers and all market-shareholders and senior executives who are not employees of the applicant company's customers;

[0125] The applicant company personnel node includes the applicant company employees within the organization's CRM system;

[0126] The applicant company's organizational structure node includes the applicant company's organizational structure within the organization's CRM system. The applicant company's organizational structure includes various levels of departments such as headquarters, business units, and branches.

[0127] The organizational structure nodes include the customer's organizational structure extracted from customer service and visit records. The customer's organizational structure includes departments at all levels, such as headquarters, business units, and branches.

[0128] Based on the applicant company's internal service visit data, construct a relationship edge between subordinates, customer service, visitors / receptionists, and customer personnel employment.

[0129] Based on the applicant company's internal CRM information, construct the applicant company's personnel employment and the subordinate relationships under the applicant company's organizational structure;

[0130] Construct a boundary based on external information data to establish the shareholding relationships among the company's subordinate branches, corporate shareholding, directors, supervisors, senior executives, and customer personnel.

[0131] Through data synchronization, processing, and storage, a full-scale institutional customer service relationship graph is constructed, which includes 5 types of entity nodes and 10 types of relationship edges.

[0132] Specifically, the identification module includes:

[0133] Import the entity nodes and relationship edge data of the entire institutional customer service relationship graph from the big data platform to the AI ​​platform;

[0134] The degree centrality algorithm with computational complexity is used to identify nodes with low complexity as super nodes. These super nodes are then marked and sent to business personnel and removed from the graph's dataset.

[0135] By using stratified sampling, sampling is performed according to the distribution characteristics of the number of various types of nodes and relation edges;

[0136] Initialize the PageRank values ​​of all nodes to equal probability values, and iteratively update the PageRank value of each node using the following formula:

[0137]

[0138] in, This represents the calculated PageRank value of node X. Let represent the damping coefficient, and i represent the node pointing to X. Let I represent the out-degree of node i, and let I represent the total number of nodes pointing to X, i.e., the in-degree.

[0139] The above formula is used to iterate and calculate until the PageRank value of each node converges to a stable value. The stationary probability value of each node is the final PageRank value, which is used to represent the importance of the node in the graph.

[0140] Obtain the out-degree, in-degree, and PageRank values ​​of customer personnel and organizational structure nodes, and import them along with the entity node and relation edge data stored in the AI ​​platform HDFS into the Hilbert graph platform used to query and display important values ​​of customer personnel and organizational structure.

[0141] Based on the PageRank value distribution, multiple intervals were divided. On the full institutional customer service relationship graph, customer personnel nodes and organizational structure nodes were classified and marked with importance levels according to different colors. The importance levels include critical, important, attention, and ordinary.

[0142] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any other combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product, which includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions according to this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another.

[0143] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. This storage medium can be a read-only memory, a disk, or an optical disk, etc.

[0144] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various variations or substitutions within the technical scope disclosed in this application, and these should all be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for identifying key personnel and key departments based on a knowledge graph, characterized in that, The method comprises the following steps: Step S100: Collect internal service visit data of customer service, and construct an organization customer service relationship topology graph; Obtain internal service visit data of customer service, which includes customer manager visit records and customer service records; wherein, the customer manager visit records and customer service records are obtained from an organization CRM system through an API interface; Based on the organization code, the customer name variant identification rule and the customer entity unique identifier, the customer entity is standardized and uniformly identified, and a mapping relationship between the customer personnel and the applicant company employees is established; based on the internal data of customer service and the mapping relationship, an organization customer service relationship topology graph is constructed; Step S200: Obtain external information data and combine the internal service visit data to construct a full-amount organization customer service relationship knowledge graph; Obtain external information data from an external data platform ECIF system through an API interface, wherein the external information data includes enterprise equity investment relationships and director and supervisor service relationships in the whole market; Integrate the internal service visit data and the external information data according to the organization customer service relationship topology graph, and define graph entity nodes, relationship edges and attributes; Through data synchronization, processing and storage, a full-amount organization customer service relationship graph is constructed, which includes five types of entity nodes and ten types of relationship edges; Step S300: Explore the personnel and departments in the full-amount organization customer service relationship knowledge graph based on customer service content and graph structure; Step S400: Calculate the importance of each type of node in the full-amount organization customer service relationship knowledge graph by using a graph algorithm, and identify key personnel and key departments based on the calculation results; Import the entity nodes and relationship edge data of the full-amount organization customer service relationship graph from a big data platform to an AI platform; Find nodes with low complexity as super nodes by using a degree centrality algorithm with complexity, mark the super nodes and send them to business personnel, and remove them from the data set of the graph; Sample in a hierarchical sampling manner according to the number distribution characteristics of each type of node and relationship edge; Obtain the out-degree, in-degree and PageRank value of the customer personnel and organization architecture nodes by using a PageRank algorithm, and import them to a Hilbert graph platform together with the entity nodes and relationship edge data stored in the AI platform HDFS; Based on the PageRank value distribution, divide multiple intervals, classify and label the customer personnel nodes and organization architecture nodes according to different colors on the full-amount organization customer service relationship graph, and obtain key personnel and key departments, wherein the importance levels include crucial, important, attention and ordinary. 2.The knowledge graph based critical person and department identification method according to claim 1, characterized in that, The entity nodes and relationship edges include: The entity nodes include enterprise nodes, customer personnel nodes, applicant company personnel nodes, applicant company organization architecture nodes and organization architecture nodes; wherein: The enterprise nodes include all market business enterprises of the applicant company customers and non-applicant company customers; The customer personnel nodes include all market shareholders and enterprise executives of the employees of the applicant company customers and non-applicant company customer employees; The applicant company personnel node comprises applicant company employees in the institutional CRM system; The applicant company organizational structure node comprises an applicant company organizational structure in the institutional CRM system, and the applicant company organizational structure comprises departments at all levels of headquarters, business units, and branches; The organizational structure node comprises a client-side organizational structure extracted from customer service and visit records, and the client-side organizational structure comprises departments at all levels of headquarters, business units, and branches; A subordinate-client service-visit / reception-client personnel employment relationship edge is constructed based on the internal service visit data of the applicant company; An applicant company personnel employment-applicant company organizational structure subordinate relationship edge is constructed based on the internal information of the institutional CRM of the applicant company; An enterprise subordinate branch-enterprise shareholding-Director and Supervisor-client personnel shareholding relationship edge is constructed based on external information data. 3.The knowledge graph based critical person and department identification method of claim 1, wherein, The PageRank algorithm is used to obtain the out-degree, in-degree, and PageRank value of the client personnel node and the organizational structure node, and the specific implementation process comprises: The PageRank values of all nodes are initialized to equal probability values, and the PageRank value of each node is iteratively updated, and the update formula is: wherein, denotes the PageRank computed value of node X, denotes a damping factor, i denotes a node pointing to X, denotes the out-degree of node i, I denotes the total number of nodes pointing to X, i.e. the in-degree; Iterative calculation is performed through the above formula until the PageRank value of each node converges to a stable value, and the stationary probability value of each node is the final PageRank value, which is used to represent the importance of the node in the graph.

4. A system for identifying key personnel and key departments based on a knowledge graph, using the method for identifying key personnel and key departments based on a knowledge graph according to any one of claims 1-3, characterized in that, The system comprises a collection module, a graph construction module, and an identification module; The collection module is configured to collect internal service visit data and external information data of customer service; The graph construction module is configured to construct a full-amount institutional customer service relationship knowledge graph by combining the external information data and the internal service visit data; The identification module is configured to calculate the importance of various nodes in the full-amount institutional customer service relationship knowledge graph by using a graph algorithm, and identify key personnel and key departments based on the calculation result.

5. The knowledge graph based critical person and critical department identification system of claim 4, wherein, The collection module comprises: The internal service visit data of customer service is obtained, and the internal data comprises visit records and customer service records of a client manager; wherein the visit records and the customer service records of the client manager are obtained from an institutional CRM system through an API interface; The external information data is obtained from an external data platform ECIF system through an API interface, and the external information data comprises enterprise equity investment relationships and Director and Supervisor employment relationships in the whole market.

6. The knowledge graph based critical person and critical department identification system of claim 5, wherein, The graph construction module comprises: Based on the internal data of customer service, the client entity is normalized, a unified client identification mapping table is established, regular expression matching of client name variants is used, and a preset alias rule library is used to process the mapping relationship between client personnel and enterprises; The uniqueness of the client entity is verified based on the organizational structure code certificate number, the mapping relationship between client personnel and applicant company employees is established, and an institutional customer service relationship topology graph is constructed; According to the institutional client service relationship topology, internal service visit data is integrated with external information data, and a graph entity node, a relationship edge and an attribute are defined; the entity node includes an enterprise node, a client personnel node, an applicant company personnel node, an applicant company organization structure node and an organization structure node; wherein: The enterprise node contains all market business enterprises of the applicant company clients and the non-applicant company clients; The client personnel node contains all market shareholders and enterprise executives of the employees of the applicant company clients and the employees of the non-applicant company clients; The applicant company personnel node includes the employees of the applicant company in the institutional CRM system; The applicant company organization structure node contains the organization structure of the applicant company in the institutional CRM system, and the organization structure of the applicant company includes departments at all levels of the headquarters, the business department and the branch company; The organization structure node includes the organization structure of the client extracted from the client service and the visit record, and the organization structure of the client includes departments at all levels of the headquarters, the business department and the branch company; A subordinate, client service, visit / reception and client personnel employment relationship edge is constructed based on the internal service visit data of the applicant company; An applicant company personnel employment and applicant company organization structure subordinate relationship edge is constructed based on the internal information of the institutional CRM of the applicant company; An enterprise subordinate branch, enterprise shareholding, director and supervisor and client personnel shareholding relationship edge is constructed based on the external information data; Through data synchronization, processing and storage, a full amount of institutional client service relationship graph is constructed, and the graph includes five types of entity nodes and ten types of relationship edges.

7. The knowledge graph based critical person and critical department identification system of claim 6, wherein, The identification module includes: Entity node and relationship edge data of the full amount of institutional client service relationship graph are imported from a big data platform to an AI platform; A node with low complexity is found as a super node through a degree centrality algorithm with a complexity, and the super node is marked and sent to a business personnel and removed from the data set of the graph; Through hierarchical sampling, each type of node and relationship edge is sampled according to the number distribution characteristics; PageRank values of all nodes are initialized as equal probability values, and the PageRank values of each node are iteratively updated, and the update formula is: wherein, denotes the PageRank computed value of node X, denotes a damping factor, i denotes a node pointing to X, denotes the out-degree of node i, I denotes the total number of nodes pointing to X, i.e. the in-degree; Through the above formula, the PageRank values of each node are iteratively calculated until the PageRank values of each node converge to a stable value, and the stable probability value of each node is the final PageRank value, which is used to represent the importance of the node in the graph; Out-degree, in-degree and PageRank values of the client personnel and organization structure nodes are obtained, and the values are imported to a Hilbert graph platform for querying and displaying the important values of the client personnel and organization structure nodes together with the entity node and relationship edge data stored in the HDFS of the AI platform; Based on the PageRank value distribution, a plurality of intervals are divided, the client personnel nodes and the organization structure nodes are classified and labeled according to different colors and importance levels on the full amount of institutional client service relationship graph, and the importance levels include crucial, important, attention and ordinary.

Images