Policy mining method and device based on knowledge graph, equipment and medium

Abstract

The invention relates to the technical field of artificial intelligence and the field of financial science and technology, and discloses a strategy mining method based on a knowledge graph, and the method comprises the steps: obtaining the data of a plurality of clients, including basic information, interaction habits, preference settings and social relationships; defining a customer portrait dimension for each customer; client data is imported into a graph database module, and a unified knowledge graph is constructed; mining an interaction relation mode and a behavior mode of each customer; distributing a label for each customer; and generating a corresponding customer strategy according to the label. By combining basic information, interaction habits, preference settings and social relation data of clients, multi-dimensional client portrait dimensions are constructed, interaction relation modes and behavior modes of the clients are further mined through a graph database module, behavior characteristics of the clients can be identified from complex data association, and the user experience is improved. And distributing a corresponding label to each customer, and finally generating an accurate strategy according to the labels of the customers.

Description

Technical Field

[0001] The present invention relates to the fields of artificial intelligence technology and financial technology, and in particular to a strategy mining method, device, equipment and storage medium based on a knowledge graph. Background Art

[0002] With the development of the financial industry, interest-bearing products in retail financial services have become one of the key tools for financial institutions to improve their profitability. Financial institutions use interest-bearing strategies to accurately identify and market potential loan interest-bearing customers. However, traditional financial systems rely on relational databases to manage and mine customer data. This approach has significant limitations in the processing and analysis of customer data and is difficult to meet the needs of multi-dimensional customer relationship analysis in modern financial services.

[0003] First, relational databases mainly process structured data and have limitations in dealing with the complexity of customer information. Customer relationships are not limited to simple attribute fields, but also include multi-dimensional relationships such as association information, interactive behaviors, and consumer preferences between customers. However, the table structure of relational databases cannot effectively represent these complex multi-dimensional relationships. This directly leads to the incomplete expression of customer information in the process of building customer portraits, especially when analyzing potential connections between customers, it is difficult to provide effective support.

[0004] Secondly, most existing relational database analysis methods rely on traditional rule-driven statistical models. These models are mainly based on existing field attributes and are difficult to capture hidden relationships and behavior patterns between customers. For example, information such as customers' social relationships, participation in common activities, and interaction frequency is difficult to associate and process in traditional systems, which leads to inaccurate classification of customer groups. This limitation directly affects the ability of financial institutions to identify potential loan interest-bearing customers and reduces the accuracy of customer labeling and marketing. Summary of the invention

[0005] The main purpose of the present invention is to provide a strategy mining method, device, equipment and storage medium based on knowledge graph, aiming to solve the technical problem that traditional relational databases cannot effectively process and analyze customers' multi-dimensional relationships and behavior patterns, resulting in inefficiency in formulating precise strategies and affecting the success rate of strategies.

[0006] To achieve the above object, the present invention provides a strategy mining method based on knowledge graph, comprising:

[0007] Acquire customer data of multiple customers, wherein the customer data includes basic information, interaction habits, preference settings, and social relationship data of the customers;

[0008] Based on each customer's customer data, define customer portrait dimensions for each customer, including basic information, interaction habits, preference settings, and social relationships;

[0009] Based on the customer portrait dimension, customer data corresponding to multiple customers are imported into the graph database module to build a unified knowledge graph;

[0010] Using the graph database module to mine the interaction relationship pattern and behavior pattern of each customer in the knowledge graph;

[0011] Assign corresponding labels to each customer based on their interaction relationship patterns and behavior patterns;

[0012] Generate a corresponding customer strategy for each customer based on the label of each customer.

[0013] Furthermore, to achieve the above-mentioned purpose, the present invention provides a strategy mining device based on knowledge graph, comprising:

[0014] A data acquisition module, which acquires customer data of multiple customers, wherein the customer data includes basic information, interaction habits, preference settings, and social relationship data of the customers;

[0015] The customer portrait generation module defines customer portrait dimensions for each customer, including basic information, interaction habits, preference settings, and social relationships, based on the customer data of each customer;

[0016] A knowledge graph construction module, based on the customer portrait dimension, imports customer data corresponding to multiple customers into the graph database module to construct a unified knowledge graph;

[0017] A data mining module, using the graph database module to mine the interaction relationship pattern and behavior pattern of each customer in the knowledge graph;

[0018] The label assignment module assigns corresponding labels to each customer based on their interaction relationship patterns and behavior patterns;

[0019] The strategy generation module generates a corresponding customer strategy for each customer based on the label of each customer.

[0020] Furthermore, to achieve the above-mentioned purpose, the present invention also provides a computer device, which includes a memory, a processor, and a knowledge graph-based strategy mining program stored in the memory and executable on the processor, wherein the knowledge graph-based strategy mining program, when executed by the processor, implements the steps of the knowledge graph-based strategy mining method as described above.

[0021] Furthermore, to achieve the above-mentioned purpose, the present invention also provides a computer-readable storage medium, on which a knowledge graph-based strategy mining program is stored, and when the knowledge graph-based strategy mining program is executed by a processor, the steps of the knowledge graph-based strategy mining method as described above are implemented.

[0022] Beneficial effects: The present invention relates to the fields of artificial intelligence technology and financial technology, and discloses a strategy mining method based on knowledge graphs, which obtains data of multiple customers, including basic information, interaction habits, preference settings, and social relationships; defines customer portrait dimensions for each customer; imports customer data into a graph database module to build a unified knowledge graph; mines the interaction relationship pattern and behavior pattern of each customer; assigns a label to each customer; and generates corresponding customer strategies based on the label. The present invention constructs a multi-dimensional customer portrait dimension by combining the customer's basic information, interaction habits, preference settings, and social relationship data, and further mines the customer's interaction relationship pattern and behavior pattern through a graph database module. It can identify the customer's behavioral characteristics from complex data associations, assign corresponding labels to each customer, and finally generate accurate strategies based on the customer's label. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

[0024] Figure 1 A schematic diagram of an application environment of a knowledge graph-based strategy mining method in an embodiment of the present invention;

[0025] Figure 2 It is a flow chart of an embodiment of a strategy mining method based on knowledge graph of the present invention;

[0026] Figure 3 It is a functional module diagram of a preferred embodiment of a strategy mining device based on knowledge graph of the present invention;

[0027] Figure 4 A schematic diagram of the structure of a computer device in one embodiment of the present invention;

[0028] Figure 5 FIG. 4 is another schematic diagram of the structure of a computer device in one embodiment of the present invention. DETAILED DESCRIPTION

[0029] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0030] The knowledge graph-based strategy mining method provided by the embodiment of the present invention can be applied to Figure 1In an application environment, the user terminal communicates with the server terminal through a network. The server terminal can obtain data of multiple customers through the user terminal, including basic information, interaction habits, preference settings and social relationships; define customer portrait dimensions for each customer; import customer data into the graph database module to build a unified knowledge graph; mine the interaction relationship pattern and behavior pattern of each customer; assign labels to each customer; and generate corresponding customer strategies based on the labels. The present invention constructs a multi-dimensional customer portrait dimension by combining the customer's basic information, interaction habits, preference settings and social relationship data, and further mines the customer's interaction relationship pattern and behavior pattern through the graph database module. It can identify the customer's behavioral characteristics from complex data associations, assign corresponding labels to each customer, and finally generate accurate strategies based on the customer's label. Among them, the user terminal can be but is not limited to various personal computers, laptops, smart phones, tablet computers and portable wearable devices. The server can be implemented with an independent server or a server cluster consisting of multiple servers. The present invention is described in detail below through specific embodiments.

[0031] See also Figure 2 , Figure 2 This is a flow chart of an embodiment of a strategy mining method based on knowledge graph provided by the present invention. It should be noted that although a logical sequence is shown in the flow chart, in some cases, the steps shown or described may be performed in a different order than that shown here.

[0032] like Figure 2 As shown, the strategy mining method based on knowledge graph proposed in the present invention includes the following steps:

[0033] S10, obtaining customer data of multiple customers, wherein the customer data includes basic information, interaction habits, preference settings, and social relationship data of the customers;

[0034] In this embodiment, the source of customer data may include multiple internal and external systems, which may be customer management systems, transaction record systems, social media data, etc. within financial institutions. Data can be collected through API, database connection, or file import. Customer data is obtained from different systems in real time through API, such as basic information of customer management system, interaction habits of transaction system, etc. Data integration tools, such as ETL (extract, transform, load) system, are used to extract data from multiple different data sources. Data is synchronized and batch processed to ensure the timeliness and integrity of data.

[0035] Basic information includes the customer's name, age, gender, occupation, income, etc. These data provide basic support for subsequent portrait construction and behavior analysis. Basic information can be extracted through the bank's internal system (such as the customer relationship management system) to ensure data accuracy. Use data cleaning tools to remove redundant or erroneous data and unify data formats.

[0036] Interaction habits include customer behavior data on different channels, such as visit frequency, purchase frequency, consultation records, etc. This data can reveal customer behavior patterns. Obtain customer interaction data on different channels from sales records or customer service systems, including offline and online behavior data. Use log analysis tools to process customer visit behaviors, click records, etc. to generate customer interaction frequencies and habit patterns.

[0037] Preference settings include customers' inclinations to choose financial products or services, such as preferences for loan products and savings products. These preference data can be directly extracted from customers' interactive behaviors. Analyze customers' product selection and service preference records, combine historical transaction data and consultation records, and generate customer preferences. Obtain clear preference information through questionnaires or direct interactions with customers, and store this information in the system for subsequent analysis.

[0038] Social relationship data includes the customer's contact information, interaction frequency, and jointly participated activities in social networks, which can help financial institutions understand the customer's social circle and its influence. Obtain the customer's social relationship data from social media platforms or third-party social data providers, which may include contact information, social network connections, and jointly participated activities. Through social network analysis tools, extract the customer's social influence and relationship strength, and generate the customer's social map.

[0039] Example: In the financial sector, by obtaining basic information and social relationship data of customers, banks can analyze each customer's personal background and their influence in social networks. By further collecting interaction habits and preference setting data, banks can accurately determine customers' interests in different financial products and their interaction patterns. For example, a customer's interaction habits show that he frequently visits the loan product page. Combined with his social relationship data, it can be found that he has a strong interactive relationship with other customers who have successfully applied for loans. The bank can deduce from these data that the customer has a high loan intention. Subsequently, the bank can generate a personalized loan recommendation strategy for the customer, effectively improving customer conversion rate and increasing bank profits.

[0040] By integrating customer data, a more comprehensive and accurate customer portrait can be constructed. Multi-dimensional data collection makes subsequent behavior analysis and customer tag generation more accurate, effectively improving the depth of customer data.

[0041] S20, based on each customer’s customer data, defines customer portrait dimensions for each customer, including basic information, interaction habits, preference settings, and social relationships;

[0042] In this embodiment, the basic information dimension includes static information of the customer, such as name, age, gender, occupation, income level, etc. This information is usually extracted from the internal system of the bank and used as the basic dimension of the customer profile. The basic information data of the customer, such as the customer's identity information, contact information, account information, etc., is extracted from the customer relationship management system (CRM) or account opening system. Data cleaning tools are used to process possible redundant or incomplete data, such as uniformly formatting the fields such as the customer's occupation and income.

[0043] The interactive habits dimension records the customer's behavior data in different scenarios, including the customer's operation records, transaction behaviors, and frequency of access to financial products in banking services. This dimension reflects the interaction pattern between customers and banks, making it easier to analyze customer behavior. Extract the customer's transaction and interaction data on different channels from the transaction record system and sales record system. Use behavioral analysis tools to calculate the customer's access frequency, product consultation records, and interaction duration, and organize these data into a structured interaction pattern.

[0044] The preference dimension describes the customer's preference for financial products and services. Customers may prefer a certain loan product, savings method or financial management service. This information can help financial institutions understand the customer's preference tendencies and needs. Use the customer's transaction records and product consultation records to analyze the customer's preference trends. For example, by analyzing the customer's purchase history, you can infer his or her preference for a certain type of financial management product. Through questionnaires or user behavior tracking, further collect customers' subjective preferences, such as favorite product types or frequently used service channels.

[0045] The social relationship dimension reflects the customer's connections and interactions in their social network, usually including the customer's contact information, associates in the social circle, and whether there are common social contacts in financial business, etc. The customer's contact information in social media or communication applications is obtained through customer authorization, and the customer's social network data is associated with his or her financial behavior. Use social network analysis tools to mine the customer's social circle and identify his or her influence and social activity in the circle.

[0046] In one specific implementation, the generation of customer portrait dimensions may be:

[0047] Basic information dimension: Use ETL tools to batch process large-scale customer basic information, unify data formats, and extract multiple attributes to expand basic information dimensions (such as family members and property status).

[0048] Interaction habit dimension: Analyze the customer’s interaction with different financial services through the log analysis system to generate detailed interaction dimensions, including interaction duration, access channel (mobile application or web page), etc.

[0049] Preference setting dimension: Combine recommendation algorithms and historical preferences to dynamically derive customers’ product preferences. For example, machine learning models can be used to analyze customers’ performance in various financial products and generate real-time personalized preferences.

[0050] Social relationship dimension: Utilize third-party social data platforms and use graph analysis technology to discover the customer’s role in social networks (such as influence centers, active users, etc.), and thereby expand the social dimension.

[0051] Example (financial sector): A bank comprehensively analyzes the financial behavior of customer A by defining customer portrait dimensions including basic information, interaction habits, preference settings, and social relationships. The bank found that customer A has a high interaction frequency through transaction record analysis and that he prefers savings-type financial products. In addition, through social relationship dimension analysis, it was found that customer A has frequent interactions with other high-value customers and is a social node with strong influence. Based on this, the bank customized a personalized financial product recommendation plan for customer A, combined with promotional activities, and successfully improved customer participation and satisfaction.

[0052] By defining customer portrait dimensions including basic information, interaction habits, preference settings and social relationships, we can comprehensively and accurately analyze customer behavior patterns and potential needs. The data of each dimension helps to provide a basis for subsequent strategy generation and provide support for personalized financial service recommendations. Compared with the single-dimensional customer data analysis method, it realizes the construction of multi-dimensional and multi-level customer portraits, which can significantly improve the accuracy of strategies and the personalization of customer service.

[0053] S30, based on the customer portrait dimension, importing customer data corresponding to multiple customers into a graph database module to construct a unified knowledge graph;

[0054] In this embodiment, before importing customer data into the graph database, it is first necessary to map the portrait dimensions (basic information, interaction habits, preference settings, social relationships) of each customer. The attributes in each dimension need to establish a corresponding relationship with the nodes and edges in the graph database to ensure that the multi-dimensional characteristics of the customer portrait can be intuitively reflected in the graph database. Create node types corresponding to the customer portrait dimensions in the graph database, such as "customer node", "interaction node", "preference node", "social relationship node", etc. Define the attribute fields of the node according to the attributes of the customer portrait dimensions. For example, the "basic information" dimension can be mapped to the name, age, and gender attributes of the "customer node"; the "interaction habits" dimension can be mapped to the interaction frequency, transaction duration and other attributes of the "interaction node".

[0055] In a graph database, each customer's basic information needs to create an independent customer node, which contains all the basic information attributes of the customer. Each customer node is the core object of customer data in the graph, and other information (such as interactions, preferences, and social relationships) will be connected to the customer node through relationships. In a graph database (such as Neo4j, JanusGraph), perform the operation of creating a customer node. Each customer node contains the customer's basic attributes, such as name, gender, age, occupation, etc. Use the batch import tool to import the basic information of multiple customers from the database or CSV file into the graph database, and automatically generate multiple customer nodes.

[0056] The interactive relationship edge is used to connect the customer node with the corresponding interactive data node, reflecting the interaction between the customer and the bank's services, products or other entities. The interactive relationship edge not only contains the connection with the interactive object, but also contains specific information such as the frequency and duration of the interaction. In the graph database, the interactive relationship edge is generated based on the customer's interactive habit data (such as purchase frequency, visit duration, etc.). Each edge connects a customer node with an interactive object node (such as a product or service node), and adds interaction-related attributes (such as "interaction frequency", "transaction date", etc.) to the relationship edge.

[0057] Preference relationship edges are used to connect customer nodes with their preferred product or service nodes, describing the customer's interest in a certain type of product or service. These relationships help financial institutions understand the customer's preference for certain products. Analyze the customer's transaction history and product browsing history to extract the customer's product and service preference data. In the graph database, create a "preference relationship edge" between the customer node and the product node, and attach the customer's preference information (such as "strong preference", "general preference", etc.) to the relationship edge. Preference data can be derived from the customer's historical transaction behavior or obtained through questionnaires.

[0058] Social relationship edges are used to connect multiple customer nodes and describe the social interaction relationships between customers. Social relationship edges usually include attributes such as social strength and interaction frequency, reflecting the customer's connection and influence in their social network. By analyzing the customer's social data (such as joint participation in activities, recommendation relationships, etc.), social relationship edges between customer nodes are created in the graph database. The relationship edges contain attributes such as "number of interactions" and "relationship strength" to reflect the social interaction between customers. Social network analysis technology can also be used to automatically generate a social relationship network.

[0059] The unified knowledge graph organically combines all customer nodes, interaction nodes, preference nodes and social nodes to form a complete customer relationship network. The knowledge graph can not only show the individual information of customers, but also the complex relationships between customers and between customers and products. In the graph database, the basic information, interaction relationships, preference relationships and social relationships of customers are integrated to build a complete graph structure. Through the connection of relationship edges, the knowledge graph can reflect the complex associations between customers and their related entities (such as products, other customers, etc.). Through the visualization tools of the graph database (such as the Graph DataScience plug-in of Neo4j), the knowledge graph can be visualized to facilitate the analysis of the overall behavior patterns and relationship networks of customers.

[0060] Example: A bank integrates customers' basic information, interaction habits, preference settings, and social relationship data by building a unified knowledge graph. In the knowledge graph, the social interaction relationship between customer A and other high-value customers is identified, and customer A's frequent interactive behavior of purchasing wealth management products is also analyzed. Based on this data, the bank can more accurately understand customer A's investment preferences and social influence, and customize personalized wealth management product recommendations for him.

[0061] By importing the data of multiple customers into the graph database module based on the customer portrait dimension and building a unified knowledge graph, it is possible to effectively manage and analyze the multi-dimensional information of customers. The knowledge graph can not only show the individual attributes of customers, but also reflect the complex relationships between customers and between customers and products, helping financial institutions better understand customer behavior patterns.

[0062] S40, using the graph database module to mine the interaction relationship pattern and behavior pattern of each customer in the knowledge graph;

[0063] In this embodiment, the interactive relationship pattern refers to the interactive behavior pattern between customers and other entities (such as products, services, and other customers). This pattern is reflected by the relationship edge between the customer node and the interactive object node, including the customer's purchase behavior with the product, the customer's social behavior with other customers, etc. In the graph database, the relationship between the customer node and its related interactive object nodes is analyzed using a query language (such as Neo4j's Cypher) to extract information such as the customer's interaction frequency, interaction duration, and transaction amount. By calculating the intensity of interaction between customers and different products or services, customers' preferences and usage habits for different financial services can be identified. For example, by querying the relationship edge between a customer and a certain financial product, the frequency of use of the customer's financial product and the transaction amount can be determined.

[0064] Behavioral patterns refer to the overall behavioral characteristics of customers, including multi-dimensional data such as their purchasing habits, preference choices, and frequency of service use. Mining behavioral patterns can help financial institutions understand customers' potential needs and future behavioral tendencies. In the graph database, the customer node and its associated product nodes, preference nodes, etc. are analyzed, and the customer's purchase records, preference settings, social relationships and other information are combined to generate the customer's overall behavioral pattern. For example, by querying the customer's financial product purchase records and combining the customer's product preferences, the customer's possible future investment tendencies can be identified. Through machine learning algorithms (such as cluster analysis and association rule analysis), the customer's behavioral data is modeled to discover potential behavioral patterns. For example, the customer's high-frequency trading behavior pattern can be identified through a clustering algorithm, and the customer's consumption habits can be further classified.

[0065] Social interaction relationships refer to the interaction patterns between customers and other customers, including the customer's social influence, interaction frequency, position in the social network, etc. Social interaction relationships reflect the customer's behavior and influence in their social network. Using the social network analysis tool of the graph database, the interaction intensity and frequency between customers and other customers are analyzed through social relationship edges to generate the customer's social relationship network. For example, by querying the interaction relationship edges between customers and other customers, the customer's activity and influence in the social network are calculated. Using graph analysis algorithms (such as PageRank and community detection algorithms), the core position of the customer in the social network is identified, and the interaction relationship between the customer and other customers is analyzed to derive its potential social influence.

[0066] Comprehensive analysis refers to the comprehensive identification of the overall behavioral characteristics of customers by combining their interactive relationship patterns, behavioral patterns, and social relationship patterns. This type of analysis can more accurately reflect the multi-dimensional behavior of customers and can be used to formulate personalized strategies. In the graph database, the customer's interactive relationship edges, behavioral pattern nodes, and social relationship edges are combined to use multi-dimensional data for customer behavior analysis. For example, by combining a customer's purchasing behavior, product preferences, and social influence, it is possible to identify a customer's potential interest in a certain type of financial product. Through machine learning models (such as decision trees and neural networks), multi-dimensional data is input into the model to generate personalized behavior analysis results for customers, which are used for subsequent strategy formulation.

[0067] The behavior pattern report is a summary of the comprehensive behavior analysis results of each customer. The report content includes the customer's interaction relationship, preference settings, social relationship, etc. This report is used for decision support and personalized service recommendation of financial institutions. The query and analysis tools of the graph database are used to automatically generate customer behavior pattern reports. The report content includes the customer's interaction frequency, product preferences, social activity, etc., and presents the customer's overall behavior pattern through visualization tools (such as graphic display). The automatically generated report can be used as a decision-making basis for financial institutions for personalized product recommendations and strategy adjustments.

[0068] Example: A bank mined the interaction relationship and behavior pattern of customer A through a graph database and found that customer A frequently participated in the purchase of wealth management products, and other customers in his social circle also had similar purchasing behaviors. By analyzing customer A's social relationships and purchasing habits, the bank identified that customer A had a strong interest in high-risk and high-return wealth management products. Based on these behavioral pattern analyses, the bank customized a personalized investment portfolio recommendation plan for customer A and made further customer recommendations through his social circle, significantly improving the bank's marketing conversion rate.

[0069] By using the graph database module to mine the customer's interactive relationship patterns and behavior patterns, we can fully understand the customer's behavioral characteristics and help financial institutions accurately identify the customer's potential needs. The powerful query and analysis capabilities of the graph database enable financial institutions to mine customer behavior patterns from multi-dimensional data and provide strong data support for the formulation of personalized strategies.

[0070] S50, assigning a corresponding label to each customer based on the interaction relationship pattern and behavior pattern of each customer;

[0071] In this embodiment, the labeling of the interactive relationship model is to classify customers into specific label categories based on the behavioral characteristics such as the intensity, frequency, and category of the interaction between customers and products, services, or other customers. These labels are used to describe the behavioral characteristics of customers in interaction, such as "high-frequency interactors", "low-frequency buyers", etc. Based on the interaction data between customers and products or other customers, information such as interaction frequency and interaction duration is obtained through graph database queries. Based on these data, different label categories are defined, and labels are assigned to customers based on their performance. For example, if a customer has a high frequency of interaction with a certain type of financial product, the label of "high-frequency product user" is assigned. Use machine learning algorithms (such as K-means clustering) to classify customers' interactive behaviors and automatically assign appropriate label categories to customers.

[0072] The labeling of behavioral patterns is based on the overall behavioral patterns of customers, dividing them into different behavioral groups or characteristic categories. These behavioral pattern labels can reflect customers' purchasing preferences, consumption habits, or investment intentions, etc. Analyze customers' purchasing behavior and preference data in financial products, obtain customers' historical transaction data, consultation records, etc. through graph databases, and combine these data to generate customer behavioral pattern labels. For example, if a customer's transaction data shows that they have a high frequency of purchases on financial products, the customer is assigned the label of "high-frequency financial management user". Through behavioral pattern recognition models (such as association rule algorithms), an automated label allocation plan is generated based on customers' transaction data and behavioral characteristics.

[0073] The labeling of social relationship patterns is to assign labels that reflect the social behavior characteristics of customers based on their social interaction behaviors with other customers, such as "high-influence users" and "core social nodes". Such labels can help financial institutions identify customers with greater influence in social networks. Use social network analysis algorithms (such as the PageRank algorithm) to analyze data such as the strength of customers' connections and the frequency of interactions in social networks. Based on these social characteristics, assign labels such as "active socializer" and "social center" to customers. Obtain the frequency of interaction between customers and other customers through the social relationship edges in the graph database, and automatically assign customer role labels in social networks. For example, when a customer interacts frequently with other high-value customers, assign the label of "high-value social node".

[0074] Multi-dimensional label assignment refers to the comprehensive analysis of the customer's interaction relationship pattern, behavior pattern and social relationship pattern to generate comprehensive labels. These labels combine the customer's performance in multiple dimensions and can more comprehensively describe the customer's characteristics. In the graph database, the customer's interaction relationship edges, behavior pattern nodes and social relationship edges are integrated to assign comprehensive labels to customers using multi-dimensional data. For example, the customer is assigned the label of "high-value financial management customer" based on the customer's purchase frequency, social influence and preference characteristics. Use a rule-based label assignment system to define the combination rules of labels of different dimensions, such as "high-frequency financial management user" + "high social influence" = "high potential customer", and automatically assign combined labels to customers based on these rules.

[0075] Tag library maintenance refers to timely updating of customer tag information when customer behavior and interaction patterns change, ensuring the timeliness and accuracy of customer tags. Set up a regular data update mechanism to monitor changes in customer behavior and social relationships. Automatically update customer tags when customer behavior patterns or social relationship patterns change significantly. For example, if a customer suddenly increases the frequency of purchasing a certain type of product, the tag is updated from "low-frequency user" to "high-frequency user". Through the real-time analysis function of the graph database, dynamically monitor changes in customer behavior patterns, update the customer's tag library in a timely manner, and ensure the effectiveness of the tags.

[0076] Example description: A bank assigned labels for interaction relationship patterns and behavior patterns to customer B through a graph database, and found that customer B was outstanding in the frequency of interaction with financial products, and frequently interacted with other high-value customers in his social circle. Based on these characteristics, the bank assigned the labels of "high-frequency financial management user" and "core social node" to customer B. Subsequently, the bank generated a personalized high-yield financial product recommendation strategy for customer B, and further expanded the potential customer base through his social circle, improving the bank's marketing conversion rate.

[0077] By assigning labels to customers based on their interactive relationship patterns and behavior patterns, we can accurately identify their behavioral characteristics and provide an important basis for subsequent strategy formulation. Labels of different dimensions help financial institutions refine customer classifications, conduct precision marketing and recommend personalized services.

[0078] S60, generating a corresponding customer policy for each customer based on the label of each customer;

[0079] In this embodiment, customer tag analysis is to comprehensively analyze the characteristics of customers based on the interaction tags, behavior pattern tags, preference tags, and social tags assigned to each customer. This step is the basis for generating personalized customer strategies, and the customer's tags reflect their behaviors, social relationships, and preferences. Through the graph database, query each customer's tag (such as high-frequency interactors, product preference users, social actives, etc.) to analyze the customer's performance in various dimensions. By combining tag information, determine the customer's specific characteristics in terms of product demand, social influence, interaction frequency, etc. Use rule-based or data-driven algorithms (such as association rules or recommendation system algorithms) to analyze the customer's tag combination. For example, when a customer has both a "high-frequency interactor" tag and a "financial product preference" tag, he or she can be identified as a potential customer with financial product needs.

[0080] According to the analysis results of customer tags, customers are classified into different customer groups. These groups can be classified according to different dimensions of tags, such as high-frequency financial management users, socially active people, product preference groups, etc. Use clustering algorithms or predefined classification rules in graph databases to divide customers into different groups according to their tags. For example, all customers with "high-frequency interaction" and "high social influence" tags can be classified into the "high-potential social customers" group. Customer classification is automatically generated through tag combination rules. For example, if the customer tag shows that the customer has a high preference for loan products and frequent interactions, it can be classified into the "potential loan customers" group.

[0081] Strategy template matching refers to selecting appropriate strategy templates from a pre-set strategy template library based on the customer's label and classification group. These strategy templates can be recommended products, service plans or marketing strategies, designed to meet the personalized needs of customers. A set of strategy templates is set in the system, such as a wealth management product recommendation strategy for "high-frequency wealth management customers" or a loan product recommendation strategy for "potential loan customers". Based on the customer's label and classification results, the rule matching algorithm is used to select the most appropriate strategy from the template library. For example, customers with the labels of "high-frequency interactor" and "wealth management product preference" will be matched with the "wealth management product portfolio recommendation strategy".

[0082] Personalized strategy generation is based on the customer's specific label characteristics and strategy templates to generate personalized product or service strategies for each customer. Personalized strategies are tailored to the customer's needs, interests, and behavioral habits. Personalized strategies are generated by combining the customer's label, classification group, and matching strategy template. For example, if the customer's label is displayed as a "high-frequency interactive financial management user", the system will generate a set of personalized financial product recommendation strategies for the customer, including portfolio investment advice, risk control, etc. Use recommendation system technology (such as recommendation algorithms based on collaborative filtering) to further optimize the strategy to ensure that the strategy is in line with the customer's behavior patterns and preferences. For example, the recommendation system recommends the most suitable products or services for the current customer by analyzing the historical behavior of similar customers.

[0083] Dynamic strategy adjustment means that when a customer's label, behavior or demand changes, the system can adjust the customer's strategy in real time to ensure the timeliness and accuracy of the strategy. The system updates the customer's label based on the customer's real-time data and behavior, and automatically adjusts the customer's personalized strategy. For example, if a customer's label changes from "financial management preference" to "loan demand", the system will automatically switch the strategy template from financial product recommendation to loan product recommendation. Use real-time data stream analysis tools to monitor changes in customer behavior and update customer strategies in real time. For example, if a customer frequently visits a certain type of loan product page in a short period of time, the system will promptly generate the corresponding loan product recommendation strategy.

[0084] Example: A bank is generating a personalized financial strategy for customer C. Customer C is a high-net-worth customer. After label analysis, customer C is labeled as a "high-frequency financial management user" and a "social core user."

[0085] Customer C's interactive behavior shows that he pays more attention to financial products and has invested a large amount of money in financial products, earning him the label of "high-frequency financial management user". At the same time, Customer C has frequent social interactions with other high-net-worth customers, so he is labeled as a "social core user", indicating that he has a certain influence in the social circle.

[0086] For the label of "high-frequency financial management user", the system matched the "high-yield financial management product recommendation strategy" from the strategy template library. This strategy focuses on recommending high-yield and risk-controlled financial management products to customers. For the label of "social core user", the system matched the "social activity incentive strategy", which aims to attract other potential customers in the circle of customer C to participate in the bank's high-end activities and product purchases through the social influence of customer C.

[0087] The recommended financial products include a group of high-yield bond fund portfolios with an annual yield of 8%-10%, low risk, and a detailed risk assessment report. In view of customer C's high-frequency trading habits, the system also adds personalized customized investment portfolio suggestions to help customers optimize their capital allocation to maximize returns. In addition, customer C is also invited to join the bank's high-end financial customer club to enjoy personalized investment advisory services and track the performance of his investment portfolio at any time.

[0088] The system generates a social promotion activity, inviting client C to recommend other high-net-worth clients in his social circle to participate in the bank's VIP exclusive wealth management activities. As an incentive, client C can receive corresponding commission rewards or enjoy higher wealth management yields for each new client he successfully recommends to participate in wealth management product investment. The bank provides exclusive online and offline activities for client C and the clients he recommends, including wealth management lectures, wealth management strategy sharing sessions, etc., through which client C's social network can strengthen the promotion effect of bank products.

[0089] In subsequent observations, the system found that customer C had recently frequently browsed the bank's loan product page and had inquired about loan interest rates and application procedures many times. Based on this behavior, the system dynamically adjusted its strategy and added a "personalized loan product recommendation strategy": it recommended a low-interest credit loan product designed for high-net-worth customers, with a high credit limit and fast approval; the system also provided customer C with a loan simulator that can automatically generate a loan plan based on the customer's asset situation, allowing customer C to evaluate the impact of different loan plans on his cash flow at any time.

[0090] By generating personalized customer strategies based on customer tags, it is possible to effectively meet customers' personalized needs, improve customer experience and the service efficiency of financial institutions. Combining customer behavior patterns, social relationships, product preferences and other tags, the system can accurately identify customer needs and generate strategy recommendations that meet the customer's actual situation.

[0091] The present invention relates to the fields of artificial intelligence technology and financial technology, and discloses a strategy mining method based on a knowledge graph, which obtains data of multiple customers, including basic information, interaction habits, preference settings, and social relationships; defines customer portrait dimensions for each customer; imports customer data into a graph database module to build a unified knowledge graph; mines the interaction relationship pattern and behavior pattern of each customer; assigns a label to each customer; and generates corresponding customer strategies based on the label. The present invention constructs a multi-dimensional customer portrait dimension by combining the basic information, interaction habits, preference settings, and social relationship data of the customer, and further mines the interaction relationship pattern and behavior pattern of the customer through the graph database module, which can identify the customer's behavioral characteristics from complex data associations, and assign corresponding labels to each customer, and finally generates accurate strategies based on the customer's label.

[0092] In one embodiment, the above S10 includes:

[0093] S101, extracting basic information data of the customer from the internal database, wherein the basic information data includes the customer's age, gender, occupation and income level;

[0094] S102, obtaining the customer's interaction habit data from the sales record, wherein the interaction habit data includes interaction frequency, interaction category, and interaction channel;

[0095] S103, collecting the customer's preference setting data from the interactive relationship management module, wherein the preference setting data includes product preference, service preference and interactive preference;

[0096] S104, obtaining the social relationship data of the client from the social media platform, wherein the social relationship data includes contact information, social network connection information, and group activity participation records;

[0097] S105, performing data cleaning and standardization operations on the basic information data, interaction habit data, preference setting data, and social relationship data;

[0098] S106, integrating the processed basic information data, interaction habit data, preference setting data and social relationship data into a unified data repository.

[0099] In this embodiment, the basic information data includes the core static attributes of the customer, such as age, gender, occupation and income level. These data are usually stored in the customer relationship management (CRM) system or user account information library of the bank. The basic information of the customer is extracted from the bank's CRM system using ETL tools. These data may have been stored in a structured manner and can be obtained through SQL query, API or file import. After the data is extracted, it will be initially processed to eliminate redundant, duplicate or invalid customer records to ensure the integrity and accuracy of the data.

[0100] Interaction habit data reflects the customer's interaction behavior with the bank or products, including purchase frequency, purchase category and interaction channel. Interaction habits can indicate the frequency, method and preference of customers for certain products. Extract customer interaction data from the bank's sales record system or transaction log, including the type, frequency and amount of products purchased. Interaction channels can be recorded through the customer's operating system data (such as mobile APP, online banking, over-the-counter transactions, etc.). Use big data analysis tools (such as Hadoop, Spark, etc.) to process large-scale transaction records and extract key interaction data.

[0101] Preference data includes customer preferences for different products and services, such as customer preference for high-yield investment products, frequent use of certain types of services, etc. Preference data helps analyze customer personalized needs. Through the interactive relationship management module or marketing automation platform, collect the preference information expressed by customers when interacting with the bank. This module usually contains the customer's historical use of products, customer service selection preferences, and feedback information. The system can infer the customer's product preferences by directly querying the user's account settings or user behavior tracks (such as web browsing records), and generate preference data for each customer.

[0102] Social relationship data reflects the connections and influence of customers in their social circles, and usually includes contact information, social network connection information, and records of participation in collective activities. Social relationship data helps to identify customers' social behavior patterns. Utilize the social media platform API authorized by the customer to obtain the customer's contact information and their social network connection information, and identify other users or related activities that are closely related to the customer. This data can be extracted through API interfaces (such as Facebook GraphAPI or LinkedInAPI). Collective activity participation records can refer to the behavior of customers in collective activities held by social platforms or banks, such as participation records in online seminars, offline activities, etc., which can be used as a basis for evaluating the customer's social activity.

[0103] Data cleaning refers to processing the extracted raw data to remove redundant, inconsistent or erroneous data to ensure the accuracy and completeness of the data. The cleaned data lays the foundation for subsequent analysis and use. Use data cleaning tools (such as OpenRefine, Python's Pandas library) to clean customer data, remove duplicate records, correct inconsistent data formats (such as date format, currency unit, etc.), and fill in possible missing values. Data cleaning can be divided into a combination of automated processing and manual verification, especially for key attributes (such as customer age, income level) with high accuracy requirements, and outliers need to be processed or verified.

[0104] Data standardization refers to the conversion of customer data from different sources and formats into a unified format and unit for subsequent use in data analysis or mining. For example, unify income information in different currencies into one monetary unit. Use standardization tools to unify customer data. For example, unify customer income levels into the same monetary unit, and unify data in different time formats into a standard timestamp format. The data standardization process can be combined with automated processing tools and rules to ensure that all customer data is processed and stored in a consistent standard for subsequent analysis.

[0105] The integrated data is stored in a unified data repository, usually a distributed database or data warehouse, for subsequent analysis and mining. This process ensures that data from different sources can be seamlessly integrated and efficiently accessed. Use a distributed database (such as Hadoop, Cassandra) or a data warehouse (such as Amazon Redshift, Google BigQuery) to store the cleaned and standardized customer data. The data repository supports efficient query and data access functions, allowing subsequent real-time or batch customer data analysis. All cleaned and standardized data can be imported into the data repository through the ETL (Extract, Transform, Load) process.

[0106] This embodiment can provide customers with a complete and accurate portrait by comprehensively collecting and cleaning customers' basic information, interaction habits, preference settings, and social relationship data, and performing standardized processing. The unified processing of different data sources makes subsequent strategy mining more accurate and can provide strong data support for personalized financial services and product recommendations.

[0107] In one embodiment, the above S20 includes:

[0108] S201, analyzing the customer data to determine key dimensions of the customer portrait, where the key dimensions include basic information, interaction habits, preference settings, and social relationships;

[0109] S202, defining attributes including age, gender, occupation and income level for the basic information dimension;

[0110] S203, defining attributes including interaction frequency, interaction category and interaction channel for the interaction habit dimension;

[0111] S204, defining attributes including product preference, service preference and interaction preference for the preference setting dimension;

[0112] S205 , defining attributes including contact information, social network connections, and joint activity participation records for the social relationship dimension.

[0113] In this embodiment, the key dimensions of the customer portrait refer to high-level classifications used to describe various types of customer information. These dimensions reflect the customer's basic attributes, behavioral characteristics, preferences, and social network relationships. By analyzing customer data, determine which dimensions can better reflect the customer's overall behavior. Analyze each customer's basic information, interactive behavior, product preferences, and social network data, and determine the key dimensions that can effectively describe customer characteristics through data mining and pattern recognition. Typical key dimensions include: basic information (age, gender, etc.), interactive habits (frequency of interaction with products or services), preference settings (preferences for products and services), and social relationships (interactions with other customers and social network relationships). The system determines the high-value dimensions of customers by analyzing customer historical data, using cluster analysis, classification algorithms, etc., to ensure the effectiveness of the portrait dimensions.

[0114] The basic information dimension is used to describe the static demographic characteristics of customers. These characteristics are the basis for other behaviors and preferences. By recording information such as the customer's age, gender, occupation, and income level, it can provide an important reference for subsequent behavior and preference judgments. In the basic information part of the customer data, the customer's age, gender, occupation, and income level are extracted and classified, and these attributes are used as the core attributes of the "basic information dimension". The basic information dimension is represented as a static attribute of the customer node in the system, and can be used as important metadata in the knowledge graph to connect other dynamic attributes (such as interaction habits or preference settings). The system will regularly update or verify this basic information to ensure its accuracy and timeliness, which is especially critical for high-value customers.

[0115] The interaction habit dimension reflects the customer's interaction behavior with bank products and services, including the customer's interaction frequency, the categories of products or services with which they interact, and the channels through which they interact. This dimension can reflect the customer's usage pattern and the closeness of their relationship with the financial institution. The customer's interaction frequency (such as daily logins or transaction times), interaction category (such as financial products, loan services), and interaction channel (such as online banking, counter services, APP, etc.) are extracted from the customer's transaction records and service usage logs. The system will generate a detailed interaction habit dimension for each customer based on these interaction habits, covering the comprehensive performance of various interaction channels to form a multi-dimensional behavior description. These data will be continuously collected and updated to ensure that the dynamics of interaction habits reflect the customer's latest usage.

[0116] The preference dimension is used to record customer preferences, including customer preferences for products, services, and interaction methods. By analyzing customer usage behavior and feedback data, it is possible to infer or confirm customer preferences, which helps to make personalized service recommendations. Based on the customer's product usage history, feedback data, and interaction behavior, the system identifies the customer's product preferences (such as preference for financial products or savings products), service preferences (such as preference for self-service or manual service), and interaction preferences (such as preference for online or offline transactions). The preference dimension stores this data as customer preference tags for subsequent personalization strategies. In the knowledge graph, these preference settings will form a preference connection between the customer node and the product or service node. Regularly analyze changes in customer preferences to ensure that the preference dimension can dynamically reflect the customer's latest needs.

[0117] The social relationship dimension describes the performance of customers in social networks, reflecting the connection, interaction and influence of customers with other people. By analyzing the social circle of customers, other customers with similar needs or potential connections can be identified to improve the accuracy of social marketing. The system obtains their contact information, social network connection information and jointly participated activities (such as financial lectures, financial salons, etc.) through the social media API authorized by the customer. This information will form the basis of the social relationship dimension. Contact information reflects the customer's direct social circle, social network connection information reflects the intensity of interaction between customers and other customers, and joint activity participation records show the customer's activeness in banking activities. In the knowledge graph, these social relationships form relationship edges between customer nodes, reflecting the customer's social network structure and its influence in the social circle.

[0118] This embodiment can comprehensively and dynamically describe the multi-dimensional characteristics of customers by defining customer portrait dimensions including basic information, interaction habits, preference settings and social relationships for each customer. Based on these dimensions, financial institutions can more accurately analyze customer behavior, make personalized product recommendations and precision marketing, and improve customer satisfaction and business conversion rate. Compared with traditional single-dimensional analysis, the customer portrait dimension provides a more comprehensive perspective and can better tap into customers' potential needs.

[0119] In one embodiment, the above S30 includes:

[0120] S301, connecting to the interface of the graph database module;

[0121] S302, classifying the basic information, interaction habits, preference settings, and social relationship data of the multiple customers according to the customer portrait dimension;

[0122] S303, based on the classification result, creating customer nodes in the graph database, each customer node including basic information, interaction habits, preference settings and social relationship attributes of the corresponding customer;

[0123] S304, creating product nodes in the graph database, each product node including the category, price and brand attributes of the corresponding product;

[0124] S305, establishing interaction relationship edges between customer nodes and product nodes in the graph database, each interaction relationship edge including corresponding interaction date and interaction quantity attributes;

[0125] S306, establishing social relationship edges between customer nodes in the graph database, each social relationship edge including relationship strength, interaction frequency and social interaction type attributes between customers.

[0126] In this embodiment, customer data is imported into the graph database module through the system interface. The graph database is a database system specifically used to store and manage complex relationships, which can intuitively represent entities (such as customers and products) and their relationships. The system is connected to the graph database module through an API, driver, or database connector (such as Neo4jJDBC driver or CypherAPI). Graph databases such as Neo4j, JanusGraph, etc. can receive external input data through a programming interface. Data is uploaded to the graph database from the external system through an interface. The role of the interface is to ensure that the data can be correctly received by the graph database module and subsequently processed.

[0127] Customer data is classified and processed according to the dimensions of customer portraits (basic information, interaction habits, preferences, social relationships) to ensure that each type of data can accurately reflect the different characteristics of the customer. Use data processing tools or scripts to extract customer data and classify it according to the dimensions of customer portraits. For example, basic information (age, gender), interaction habits (purchase frequency), preferences (product or service preferences) and social relationships (contact information). The corresponding customer data is structured into different categories for subsequent storage and query in the graph database. This process can be completed using ETL (Extract, Transform, Load) tools.

[0128] In a graph database, customers are stored in the form of nodes. Each node represents a customer and carries all the important information about the customer, such as basic information, interaction habits, preferences, and social relationships. Customer nodes are created using the query language of the graph database (such as Cypher in Neo4j). Each node contains attribute fields that contain the customer's basic information (such as age, occupation), interaction habits (such as purchase frequency), preferences (such as financial product preferences), and social relationships (such as the number of contacts). After each customer node is created, it will be connected to other related nodes, such as product nodes and other customer nodes.

[0129] Product nodes represent specific products or services that customers interact with. Each product node contains basic attribute information of the product, such as category, price, brand, etc. Product nodes are stored as entities in the graph database and connected to related customer nodes. Product nodes are created using the query language of the graph database. Each node represents a specific product (such as financial products, investment products) and contains attribute fields, such as product category (savings, financial management), price (amount), and brand (financial institution). Each product node is connected to the relevant customer node to reflect the relationship between the customer and the product.

[0130] The interactive edge is the connection between the customer node and the product node, reflecting the specific interaction between the customer and the product. The interactive edge contains the interaction frequency (such as the number of purchases) and interaction time (such as the transaction date) between the customer and the product. Using the relationship building function of the graph database, create interactive edges between the customer node and the product node through the Cypher query language. Each relationship edge represents an interaction, such as a record of a customer purchasing a product, and is accompanied by the interaction time (such as the transaction date) and the number of interactions (such as the number of purchases). These interactive edges will be stored in the graph database, which supports querying the customer's purchasing behavior pattern. For example, you can query the historical interaction records between a customer and a certain type of product.

[0131] Social relationship edges connect customer nodes and represent the social interactions between customers, including the frequency of interaction, relationship strength, and interaction type (such as friends, colleagues, business partners, etc.). Use a graph database to establish social relationship edges between customer nodes and record the social interactions between customers. Each social relationship edge is accompanied by relationship strength (such as the number of interactions), interaction frequency (such as the number of interactions per week), and social type (such as friends, colleagues, etc.). By analyzing these social relationship edges, the system can identify the customer's activity or influence in the social network, supporting subsequent customer behavior analysis and strategy formulation.

[0132] This embodiment can comprehensively manage the relationship between customers and products, and between customers by importing the customer's basic information, interaction habits, preference settings, and social relationship data into the graph database module and building a unified knowledge graph. The construction of the knowledge graph can not only improve the visual management of customer relationships, but also provide accurate product recommendations and strategy formulation through the analysis of customer-product interaction data, thereby enhancing customer experience.

[0133] In one embodiment, the above S40 includes:

[0134] S401, generating a social interaction relationship for each customer by querying the social relationship edges between each customer and other customers in the knowledge graph, analyzing the interaction frequency, relationship strength and interaction type in the social relationship edges;

[0135] S402, by querying the interaction relationship edge between each customer and the product in the knowledge graph, analyzing the interaction date, interaction quantity and interaction category contained in the interaction relationship edge, and generating the interaction relationship between each customer and the product;

[0136] S403, by querying the preference setting attributes of each customer node in the knowledge graph, analyzing the customer's product preference, service preference and interaction preference data, and determining the customer's preference tendency and preference characteristics;

[0137] S404, combining each customer's social interaction relationship, the customer's interaction relationship with the product, and the customer's preference tendency and preference characteristics, to generate each customer's overall interaction relationship and overall behavior pattern.

[0138] In this embodiment, the social interaction relationship is obtained through the social relationship edges between customers, reflecting the interaction behavior between customers and other customers. The main analysis dimensions of the social relationship edges include interaction frequency (such as the number of interactions), relationship strength (such as a strong relationship with a specific customer) and interaction type (such as friends, colleagues or business partners). Using the query language of the graph database (such as Cypher in Neo4j), the attributes of interaction frequency, relationship strength and interaction type are extracted from the social relationship edges between customer nodes. The system generates social interaction relationships for each customer by analyzing these attributes. For example, customer A may have high-frequency interactions with customer B and have a strong social relationship, and can be marked as a "core social partner."

[0139] The interactive relationship between customers and products is obtained through the analysis of the interactive relationship edges between customer nodes and product nodes, focusing on the interactive behavior between customers and a certain type of product or service. The attributes in the interactive relationship edge include the interaction date (such as purchase time), the number of interactions (such as the amount or number of purchases), and the interaction category (such as product type). The system extracts the interaction data between customers and products by querying the interactive relationship edges in the graph database. This includes when and with which products the customer interacted, the specific number of interactions, and the interaction category (such as financial products, loan products, etc.). By analyzing these interaction records, the interactive relationship between customers and products is generated. For example, if customer C often purchases high-yield financial products, the system will generate a label of "high-frequency interactor of financial products." This analysis helps identify customer preferences and behaviors in product use, and provides data support for product recommendations and strategy formulation.

[0140] Each customer's preference attributes reflect their preferences for different products and services. By querying these attributes, the system can determine the customer's product preferences (such as preference for financial products), service preferences (such as preference for online services), and interaction preferences (such as preference for using apps). Use the graph database to query the preference attributes in the customer node and extract the customer's product, service, and interaction preference data. For example, customer D may show a preference for high-risk investment products, or prefer self-service rather than offline services. By analyzing this data, the system can determine the customer's preference tendencies, such as which type of product or service the customer prefers, and make personalized recommendations. These preference characteristics help financial institutions grasp customer needs more accurately and make targeted product recommendations.

[0141] By integrating the customer's social interaction relationships, product interaction relationships and preference characteristics, the system generates the customer's overall interaction relationships and behavior patterns. This comprehensive analysis can more comprehensively reflect the customer's overall behavior and help financial institutions better understand customer needs and behavior trends. The system integrates each customer's social interaction relationships (such as the strength of relationships with other customers, interaction frequency), product interaction relationships (such as purchasing behavior, usage frequency) and preference characteristics (such as product and service preferences) to form an overall behavior pattern.

[0142] For example, customer E may show high frequency of social interaction, high frequency of financial product use, and preference for self-service. The system will generate a behavior pattern of "active social user, high frequency financial purchaser, preference for self-service" for him.

[0143] These behavioral patterns provide the basis for subsequent steps such as customer tag assignment and strategy recommendation, helping to improve the accuracy of personalized services.

[0144] This embodiment uses the graph database to mine the interactive relationship patterns and behavior patterns of customers, and can deeply analyze the social activities, product usage behaviors and preference characteristics of customers. Combining these analyses, the system can generate a comprehensive behavior pattern for each customer, helping financial institutions better understand customer needs, behavior trends and potential business opportunities.

[0145] In one embodiment, the above S50 includes:

[0146] S501, analyzing the social interaction intensity, social influence and activity data in the customer's interactive relationship model;

[0147] S502, assigning corresponding social tags to the customer according to the analysis results, wherein the social tags reflect the customer's activity level and influence in the social network;

[0148] S503, analyzing the purchase frequency, purchase category and purchase channel data in the customer's interactive relationship model;

[0149] S504, assigning a corresponding product interaction tag to the customer according to the analysis result, wherein the product interaction tag reflects the customer's consumption habits and purchasing behavior;

[0150] S505, analyzing product preference, service preference and interaction preference data in the customer's behavior pattern;

[0151] S506, assigning a corresponding personalized tag to the customer according to the analysis result, wherein the personalized tag reflects the personalized needs and preference characteristics of the customer;

[0152] S507, integrating the customer's social tags, product interaction tags and personalized tags to generate a comprehensive tag that reflects the customer's comprehensive characteristics.

[0153] In this embodiment, the social relationship data between each customer and other customers in the graph database is used to analyze their social interaction intensity (such as interaction frequency and frequency), social influence (such as whether they are core contacts) and activity in social networks (such as whether they are the main interactors in the social circle). The system queries the customer's social relationship edges through the graph database and extracts relevant attributes, such as interaction frequency (such as the number of interactions per week) and social network coreness (such as the relationship with other high-influence customers), so as to evaluate the customer's social influence and activity. By comparing the customer's social activities with the behavior patterns of other customers, their position in the social network is determined, and a quantitative social activity and influence score is generated.

[0154] Based on the results of the customer's social interaction analysis, the system assigns social tags to the customer. These tags reflect the customer's status and influence in the social network, such as "socially active user", "social core member" or "high-influence customer". The system uses preset classification rules to assign corresponding social tags based on the customer's social interaction intensity, influence and activity. For example, customer A has frequent social activities and interacts with multiple high-value customers, so he is labeled as a "high-influence customer". These social tags are stored in the customer's graph database node as part of the customer portrait, and are subsequently used for strategy recommendations and personalized services.

[0155] Through the interactive relationship edges between customers and products, the system analyzes customers' purchasing behavior data, including purchase frequency (such as the number of purchases per month), category (such as the type of product purchased), and channel (such as online or offline purchases). These data reflect customers' consumption habits. The system extracts customers' purchase frequency (such as the number of times they purchase financial products), purchase category (such as investment products, savings products, etc.), and purchase channel (such as online banking, mobile APP, etc.) through the interactive relationship edges between customers and products in the graph database. Through these data, the system can identify customers' consumption patterns and preferences, and provide support for subsequent product recommendations and personalized strategies.

[0156] Based on the results of the customer's purchase behavior analysis, the system assigns product interaction tags to the customer. These tags reflect the customer's consumption habits, such as "high-frequency financial management user", "savings product preference user", "multi-channel user", etc. The system assigns product interaction tags based on the customer's purchase frequency, category, and channel. For example, customer B often purchases financial products and operates through the APP, and the system assigns him the tags of "high-frequency financial management user" and "online channel preference user". These product interaction tags reflect their consumption behavior patterns in the customer portrait and can be used for personalized product recommendations.

[0157] By analyzing the preference data in customer behavior patterns, the system can identify customers' specific preferences for products and services, such as preference for high-risk investment products, preference for online services, or preference for specific forms of interaction. The system analyzes customers' historical behavior data on products, services, and interaction methods by querying the preference setting attributes in the customer node. For example, customer C shows a preference for high-yield financial products and frequently uses mobile APP operations. These data help the system determine the customer's preference characteristics and provide a data basis for personalized recommendations and strategy formulation.

[0158] The system generates personalized labels based on the customer's product, service and interaction preference data to reflect the customer's personalized needs and preference characteristics. These labels can be used for refined marketing and personalized service recommendations. By analyzing the customer's preference data, the system assigns personalized labels to customers. For example, customer D prefers high-risk financial products and likes to operate through mobile apps. The system assigns him the labels "high-risk investor" and "online operation preference user". These labels support the generation of personalized customer strategies to ensure that the recommended products and services meet the customer's specific needs and preferences.

[0159] The system generates comprehensive labels that reflect the customer's all-round characteristics by comprehensively analyzing the customer's social tags, product interaction tags, and personalized tags. These comprehensive labels can provide customers with an overall portrait and help financial institutions understand customer behavior and needs more comprehensively. The system generates comprehensive feature labels by combining customers' social tags (such as "high social influence"), product interaction tags (such as "high-frequency financial management users"), and personalized tags (such as "high-risk investment preferences"). For example, the comprehensive label of customer E is "high-frequency interaction, high social activity, high-risk investor." Comprehensive labels reflect the overall characteristics of customers and are used for subsequent customer classification, strategy formulation, and product recommendations.

[0160] This embodiment can better reflect the customer's behavioral characteristics and demand preferences by assigning corresponding social tags, product interaction tags and personalized tags based on the customer's interactive relationship pattern and behavior pattern. This way of assigning tags not only improves the accuracy of customer portraits, but also provides data support for personalized product recommendations and strategy formulation of financial institutions. The tag assignment system can quickly adapt to changes in customer behavior and provide real-time and accurate customer portraits.

[0161] In one embodiment, the above S60 includes:

[0162] S601, classifying each customer into corresponding customer groups according to the label of each customer;

[0163] S602, selecting a corresponding policy template from a predefined policy template library according to corresponding characteristics of each customer group;

[0164] S603, configuring parameters of the selected strategy template based on the behavior data and preferences of the customer group, and generating a group strategy corresponding to the customer group;

[0165] S604, adjusting the corresponding group strategy according to the target needs of the customer and generating the corresponding customer strategy.

[0166] In this embodiment, customers with similar tags are grouped into the same customer group based on each customer's personalized tags (such as social tags, product interaction tags, personalized tags, etc.). These customer groups can reflect certain common characteristics of customers, such as consumption habits, social behaviors, or product preferences. The system classifies customers' tags and divides them into different groups using tag similarity analysis. For example, customers with tags of "high-frequency financial management users" and "high social activity" may be classified as "financial management social active groups." The system uses clustering algorithms (such as K-means, DBSCAN, etc.) to analyze customer tags and automatically divide customers into the most suitable customer groups.

[0167] Each customer group has specific behavior patterns and preference characteristics. The system selects the most appropriate policy template from the policy template library based on these group characteristics. Policy templates are predefined service plans or product recommendation models that are suitable for different customer groups. The system matches the corresponding policy template for each customer group in the policy template library. The policy template library stores various policy models for marketing, product recommendations, or service promotion, which have been optimized for different customer group characteristics. For example, for "high-frequency financial management users", the system will select "high-yield recommendation strategy for financial products"; for "highly socially active customers", the system will select "social network promotion strategy".

[0168] After the strategy template is selected, the system will customize the strategy template according to the actual behavior data of the customer group (such as purchase frequency, preference characteristics) and personalized needs. These parameter adjustments make the strategy more adaptable to the specific needs of the customer group. The system parameterizes the strategy template according to the behavioral data of the customer group (such as the frequency of use of financial products and purchase channels). For example, the system will configure the recommended product yield range and risk level based on the group's financial preferences. Parameter configuration is automatically adjusted through preset rules or machine learning models to ensure that each customer group obtains the most accurate strategy. For example, a customer group that prefers online services will be configured with a service plan with a higher frequency of online interaction.

[0169] The system fine-tunes the group strategy based on the specific target needs of the client, ensuring that the generated strategy fully matches the needs of each client. Ultimately, the system generates personalized customer strategies for each client that adapt to the client's individual needs and goals. The system personalizes the group strategy by analyzing the client's target needs (such as the expected return the client hopes to achieve through investment, expectations for services, etc.). For example, the system will add additional portfolio recommendations to Client E's strategy to meet Client E's specific goals. This process is automatically completed through data analysis, rule engines, or machine learning models to ensure that the client strategy is not only suitable for the group to which it belongs, but also meets its individual needs.

[0170] Example description: A bank used this technical solution to generate a personalized financial strategy for customer G. First, the system classified customer G and other customers with similar labels into the "high net worth investment group". Then, the system selected the "high-yield, low-risk investment strategy" template from the strategy template library, configured parameters based on the investment preferences and behavioral data of the customer group, and recommended a number of high-yield financial products. Finally, the system fine-tuned the strategy based on customer G's target needs (such as short-term profit goals) and generated the final personalized investment strategy for customer G. This strategy not only meets the personalized needs of customer G, but also helps customer G obtain the best returns in market fluctuations through dynamic adjustments.

[0171] This embodiment can realize the automated process from customer tags to strategy generation by generating corresponding customer strategies according to customer tags. Through customer group division, strategy template selection and parameterized configuration, the system can provide each customer with accurate personalized strategies to meet their specific needs.

[0172] In one embodiment, a knowledge graph-based strategy mining device is provided, and the knowledge graph-based strategy mining device corresponds one-to-one to the knowledge graph-based strategy mining method in the above embodiment. Figure 3 , Figure 3 This is a functional module diagram of a preferred embodiment of a strategy mining device based on knowledge graph of the present invention. Data acquisition module 10, customer portrait generation module 20, knowledge graph construction module 30, data mining module 40, label assignment module 50 and strategy generation module 60. Each functional module is described in detail as follows:

[0173] A data acquisition module 10 is used to acquire customer data of multiple customers, wherein the customer data includes basic information, interaction habits, preference settings and social relationship data of the customers;

[0174] A customer profile generation module 20, for defining customer profile dimensions including basic information, interaction habits, preference settings and social relationships for each customer based on the customer data of each customer;

[0175] A knowledge graph construction module 30, for importing customer data corresponding to a plurality of customers into a graph database module based on the customer portrait dimension to construct a unified knowledge graph;

[0176] A data mining module 40 is used to mine the interaction relationship pattern and behavior pattern of each customer in the knowledge graph using the graph database module;

[0177] The label assignment module 50 is used to assign a corresponding label to each customer based on the interaction relationship pattern and behavior pattern of each customer;

[0178] The policy generating module 60 is used to generate a corresponding customer policy for each customer based on the label of each customer.

[0179] In one embodiment, the data acquisition module 10 is specifically used for:

[0180] Extracting basic information data of the customer from the internal database, wherein the basic information data includes the customer's age, gender, occupation and income level;

[0181] Acquire the customer's interaction habit data from the sales records, wherein the interaction habit data includes interaction frequency, interaction category, and interaction channel;

[0182] Collecting customer preference data from the interactive relationship management module, wherein the preference data includes product preferences, service preferences, and interaction preferences;

[0183] Acquire the client's social relationship data from the social media platform, wherein the social relationship data includes contact information, social network connection information, and group activity participation records;

[0184] Performing data cleaning and standardization operations on the basic information data, interaction habit data, preference setting data, and social relationship data;

[0185] Integrate the processed basic information data, interaction habit data, preference data and social relationship data into a unified data repository.

[0186] In one embodiment, the customer portrait generation module 20 is specifically used to:

[0187] Analyze the customer data to determine key dimensions of customer portraits, including basic information, interaction habits, preference settings, and social relationships;

[0188] Define attributes including age, gender, occupation, and income level for the basic information dimension;

[0189] Define attributes including interaction frequency, interaction category and interaction channel for the interaction habit dimension;

[0190] Define attributes including product preference, service preference and interaction preference for the preference dimension;

[0191] Define attributes for the social relationship dimension that include contact information, social network connections, and joint activity participation records.

[0192] In one embodiment, the knowledge graph construction module 30 is specifically used to:

[0193] An interface connected to the graph database module;

[0194] Classify the basic information, interaction habits, preference settings and social relationship data of the multiple customers according to the customer portrait dimension;

[0195] Based on the classification results, customer nodes are created in the graph database, each of which contains basic information, interaction habits, preference settings, and social relationship attributes of the corresponding customer;

[0196] Creating product nodes in the graph database, each product node including the category, price and brand attributes of the corresponding product;

[0197] Establishing interaction relationship edges between customer nodes and product nodes in the graph database, each interaction relationship edge including corresponding interaction date and interaction quantity attributes;

[0198] Social relationship edges are established between customer nodes in the graph database, and each social relationship edge contains relationship strength, interaction frequency, and social interaction type attributes between customers.

[0199] In one embodiment, the data mining module 40 is specifically used to:

[0200] By querying the social relationship edges between each customer and other customers in the knowledge graph, analyzing the interaction frequency, relationship strength and interaction type in the social relationship edges, the social interaction relationship of each customer is generated;

[0201] By querying the interaction relationship edge between each customer and the product in the knowledge graph, analyzing the interaction date, interaction quantity and interaction category contained in the interaction relationship edge, the interaction relationship between each customer and the product is generated;

[0202] By querying the preference setting attributes of each customer node in the knowledge graph, analyzing the customer's product preference, service preference and interaction preference data, and determining the customer's preference tendency and preference characteristics;

[0203] By combining each customer's social interaction relationship, the interaction relationship between the customer and the product, and the customer's preference tendencies and preference characteristics, the overall interaction relationship and overall behavior pattern of each customer are generated.

[0204] In one embodiment, the label allocation module 50 is specifically configured to:

[0205] Analyze the social interaction intensity, social influence and activity data in the customer's interaction relationship model;

[0206] According to the analysis results, corresponding social tags are assigned to the customers, wherein the social tags reflect the customer's level of activity and influence in the social network;

[0207] Analyze the purchase frequency, purchase category and purchase channel data in the customer's interactive relationship pattern;

[0208] According to the analysis results, a corresponding product interaction tag is assigned to the customer, wherein the product interaction tag reflects the customer's consumption habits and purchasing behavior;

[0209] Analyze customer behavior patterns in terms of product preferences, service preferences, and interaction preferences;

[0210] According to the analysis results, a corresponding personalized label is assigned to the customer, wherein the personalized label reflects the customer's personalized needs and preference characteristics;

[0211] Integrate customers’ social tags, product interaction tags and personalized tags to generate comprehensive tags that reflect the customers’ comprehensive characteristics.

[0212] In one embodiment, the strategy generation module 60 is specifically configured to:

[0213] Divide customers into corresponding customer groups based on their labels;

[0214] According to the corresponding characteristics of each customer group, select the corresponding policy template from the pre-defined policy template library;

[0215] Based on the behavioral data and preferences of the customer group, configure the parameters of the selected strategy template to generate a group strategy for the corresponding customer group;

[0216] Adjust the corresponding group strategy and generate the corresponding customer strategy according to the customer's target needs.

[0217] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile and / or volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external user terminal through a network connection. When the computer program is executed by the processor, it realizes the functions or steps of a server-side of a strategy mining method based on a knowledge graph.

[0218] In one embodiment, a computer device is provided. The computer device may be a user terminal, and its internal structure diagram may be as follows: Figure 5As shown. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external server via a network connection. When the computer program is executed by the processor, it implements the functions or steps on the user side of a strategy mining method based on a knowledge graph

[0219] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program:

[0220] Acquire customer data of multiple customers, wherein the customer data includes basic information, interaction habits, preference settings, and social relationship data of the customers;

[0221] Based on each customer's customer data, define customer portrait dimensions for each customer, including basic information, interaction habits, preference settings, and social relationships;

[0222] Based on the customer portrait dimension, customer data corresponding to multiple customers are imported into the graph database module to build a unified knowledge graph;

[0223] Using the graph database module to mine the interaction relationship pattern and behavior pattern of each customer in the knowledge graph;

[0224] Assign corresponding labels to each customer based on their interaction relationship patterns and behavior patterns;

[0225] Generate a corresponding customer strategy for each customer based on the label of each customer.

[0226] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0227] Acquire customer data of multiple customers, wherein the customer data includes basic information, interaction habits, preference settings, and social relationship data of the customers;

[0228] Based on each customer's customer data, define customer portrait dimensions for each customer, including basic information, interaction habits, preference settings, and social relationships;

[0229] Based on the customer portrait dimension, customer data corresponding to multiple customers are imported into the graph database module to build a unified knowledge graph;

[0230] Using the graph database module to mine the interaction relationship pattern and behavior pattern of each customer in the knowledge graph;

[0231] Assign corresponding labels to each customer based on their interaction relationship patterns and behavior patterns;

[0232] Generate a corresponding customer strategy for each customer based on the label of each customer.

[0233] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can refer to the relevant descriptions on the server side and the user side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0234] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0235] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0236] It should be noted that if software tools or components other than those of the Company appear in the embodiments of the present application, they are only used for illustration and do not represent actual use. The above-described embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the above-mentioned embodiments, a person of ordinary skill in the art should understand that the technical solutions described in the above-mentioned embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents; and these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A strategy mining method based on knowledge graph, characterized in that: The following steps are involved: Acquire customer data of multiple customers, wherein the customer data includes basic information, interaction habits, preference settings, and social relationship data of the customers; Based on each customer's customer data, define customer portrait dimensions for each customer, including basic information, interaction habits, preference settings, and social relationships; Based on the customer portrait dimension, customer data corresponding to multiple customers are imported into the graph database module to build a unified knowledge graph; Using the graph database module to mine the interaction relationship pattern and behavior pattern of each customer in the knowledge graph; Assign corresponding labels to each customer based on their interaction relationship patterns and behavior patterns; Generate a corresponding customer strategy for each customer based on the label of each customer.

2. The strategy mining method based on knowledge graph according to claim 1, characterized in that: Acquire customer data of multiple customers, the customer data including basic information, interaction habits, preference settings and social relationship data of the customers, including: Extracting basic information data of the customer from the internal database, wherein the basic information data includes the customer's age, gender, occupation and income level; Acquire the customer's interaction habit data from the sales records, wherein the interaction habit data includes interaction frequency, interaction category, and interaction channel; Collecting customer preference data from the interactive relationship management module, wherein the preference data includes product preferences, service preferences, and interaction preferences; Acquire the client's social relationship data from the social media platform, wherein the social relationship data includes contact information, social network connection information, and group activity participation records; Performing data cleaning and standardization operations on the basic information data, interaction habit data, preference setting data, and social relationship data; Integrate the processed basic information data, interaction habit data, preference data and social relationship data into a unified data repository.

3. The strategy mining method based on knowledge graph according to claim 1, characterized in that: Based on each customer's customer data, define customer portrait dimensions for each customer, including basic information, interaction habits, preferences, and social relationships, including: Analyze the customer data to determine key dimensions of customer portraits, including basic information, interaction habits, preference settings, and social relationships; Define attributes including age, gender, occupation, and income level for the basic information dimension; Define attributes including interaction frequency, interaction category and interaction channel for the interaction habit dimension; Define attributes including product preference, service preference and interaction preference for the preference dimension; Define attributes for the social relationship dimension that include contact information, social network connections, and joint activity participation records.

4. The strategy mining method based on knowledge graph according to claim 1, characterized in that: Based on the customer portrait dimension, the customer data corresponding to multiple customers are imported into the graph database module to build a unified knowledge graph, including: An interface connected to the graph database module; Classify the basic information, interaction habits, preference settings and social relationship data of the multiple customers according to the customer portrait dimension; Based on the classification results, customer nodes are created in the graph database, each of which contains basic information, interaction habits, preference settings, and social relationship attributes of the corresponding customer; Creating product nodes in the graph database, each product node including the category, price and brand attributes of the corresponding product; Establishing interaction relationship edges between customer nodes and product nodes in the graph database, each interaction relationship edge including corresponding interaction date and interaction quantity attributes; Social relationship edges are established between customer nodes in the graph database, and each social relationship edge contains relationship strength, interaction frequency, and social interaction type attributes between customers.

5. The knowledge graph-based strategy mining method as claimed in claim 1, characterized in that: The graph database module is used to mine the interaction relationship pattern and behavior pattern of each customer in the knowledge graph, including: By querying the social relationship edges between each customer and other customers in the knowledge graph, analyzing the interaction frequency, relationship strength and interaction type in the social relationship edges, the social interaction relationship of each customer is generated; By querying the interaction relationship edge between each customer and the product in the knowledge graph, analyzing the interaction date, interaction quantity and interaction category contained in the interaction relationship edge, the interaction relationship between each customer and the product is generated; By querying the preference setting attributes of each customer node in the knowledge graph, analyzing the customer's product preference, service preference and interaction preference data, and determining the customer's preference tendency and preference characteristics; By combining each customer's social interaction relationship, the interaction relationship between the customer and the product, and the customer's preference tendencies and preference characteristics, the overall interaction relationship and overall behavior pattern of each customer are generated.

6. The strategy mining method based on knowledge graph according to claim 1, characterized in that: Based on each customer's interaction relationship pattern and behavior pattern, each customer is assigned a corresponding label, including: Analyze the social interaction intensity, social influence and activity data in the customer's interaction relationship model; According to the analysis results, corresponding social tags are assigned to the customers, wherein the social tags reflect the customer's level of activity and influence in the social network; Analyze the purchase frequency, purchase category and purchase channel data in the customer's interactive relationship pattern; According to the analysis results, a corresponding product interaction tag is assigned to the customer, wherein the product interaction tag reflects the customer's consumption habits and purchasing behavior; Analyze customer behavior patterns in terms of product preferences, service preferences, and interaction preferences; According to the analysis results, a corresponding personalized label is assigned to the customer, wherein the personalized label reflects the customer's personalized needs and preference characteristics; Integrate customers’ social tags, product interaction tags and personalized tags to generate comprehensive tags that reflect the customers’ comprehensive characteristics.

7. The strategy mining method based on knowledge graph according to claim 1, characterized in that: Generate a corresponding customer strategy for each customer based on each customer's tag, including: Divide customers into corresponding customer groups based on their labels; According to the corresponding characteristics of each customer group, select the corresponding policy template from the pre-defined policy template library; Based on the behavioral data and preferences of the customer group, configure the parameters of the selected strategy template to generate a group strategy for the corresponding customer group; Adjust the corresponding group strategy and generate the corresponding customer strategy according to the customer's target needs.

8. A strategy mining device based on knowledge graph, characterized in that: The strategy mining device based on knowledge graph includes: A data acquisition module, used to acquire customer data of multiple customers, wherein the customer data includes basic information, interaction habits, preference settings and social relationship data of the customers; A customer portrait generation module is used to define customer portrait dimensions including basic information, interaction habits, preference settings and social relationships for each customer based on the customer data of each customer; A knowledge graph construction module, used to import customer data corresponding to multiple customers into a graph database module based on the customer portrait dimension to construct a unified knowledge graph; A data mining module, used to mine the interactive relationship pattern and behavior pattern of each customer in the knowledge graph using the graph database module; A label assignment module is used to assign corresponding labels to each customer based on the interaction relationship pattern and behavior pattern of each customer; The strategy generation module is used to generate a corresponding customer strategy for each customer based on the label of each customer.

9. A computer device, characterized in that: The computer device includes a memory, a processor, and a knowledge graph-based policy mining program stored in the memory and executable on the processor. When the knowledge graph-based policy mining program is executed by the processor, the steps of the knowledge graph-based policy mining method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The storage medium stores a knowledge graph-based strategy mining program, and when the knowledge graph-based strategy mining program is executed by the processor, the steps of the knowledge graph-based strategy mining method as described in any one of claims 1-7 are implemented.

Images