Intention matching method and device, computer device and storage medium

Abstract

The application relates to an intention matching method and device, computer equipment, a storage medium and a computer program product. The method comprises the following steps: obtaining a target business intention of a service provider and a target behavior intention of a service demander; determining a service intention graph based on a business resource combination, a business intention resource, a behavior resource combination and a behavior intention resource; calculating a first intention recognition cost of the target behavior intention and a second intention recognition cost of the target business intention according to the service intention graph; determining an intention matching degree between the target business intention and the target behavior intention based on the first intention recognition cost and the second intention recognition cost; and adjusting at least one of the business resource combination of the service provider and the behavior resource combination of the service demander until the intention matching is completed when the intention matching degree is no longer reduced, in the case that the intention matching degree is not equal to an intention matching degree threshold. The method can accurately match the intentions of the two parties.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to an intent matching method, apparatus, computer device, storage medium, and computer program product. Background Technology

[0002] With the development of artificial intelligence technology, various types of financial service terminals have emerged, provided by financial service providers. These providers use AI to identify the behavioral intentions of users (those seeking financial services) during interactions, obtain the information the user desires, and display it to the user through an interactive interface, facilitating related business operations.

[0003] Current methods for identifying user behavioral intent involve analyzing business needs through service feedback content or mining explicit user profiles to determine user preferences for using terminals to access related business services. This increases the content covered by user behavior analysis results, thereby identifying user behavioral intent.

[0004] However, current methods for identifying user behavioral intent rely solely on adding content to the user behavior analysis results to identify interaction intent. This fails to accurately identify user behavioral intent, leading to a mismatch between the intent of the service provider and the intent of the service requester during the interaction process. Summary of the Invention

[0005] Therefore, it is necessary to provide an intent matching method, apparatus, computer device, computer-readable storage medium, and computer program product that can accurately match the intents of both parties in an interaction, in order to address the aforementioned technical problems.

[0006] Firstly, this application provides an intent matching method. The method includes:

[0007] To obtain the target business intent of the service provider and the target behavioral intent of the service demander;

[0008] Based on the service provider's business resource combination, business intent resources, service demander's behavioral resource combination, and behavioral intent resources, a service intent graph is determined. The service intent graph includes a data graph, an information graph, a knowledge graph, and an intent graph.

[0009] Based on the service intent graph, calculate the first intent recognition cost of the target behavioral intent and the second intent recognition cost of the target business intent, respectively.

[0010] Based on the first intent recognition cost and the second intent recognition cost, determine the intent matching degree between the target business intent and the target behavioral intent;

[0011] If the intent matching degree is not equal to the intent matching degree threshold, adjust at least one of the service provider's business resource combination and the service demander's behavioral resource combination to change the service intent graph and reduce the intent matching degree. Intent matching is considered complete when the intent matching degree no longer decreases.

[0012] Secondly, this application also provides an intent matching device. The device includes:

[0013] The acquisition module is used to acquire the target business intent of the service provider and the target behavioral intent of the service demander.

[0014] The determination module is used to determine the service intent graph based on the service provider's business resource combination, business intent resources, service demander's behavioral resource combination, and behavioral intent resources. The service intent graph includes data graph, information graph, knowledge graph, and intent graph.

[0015] The calculation module is used to calculate the first intent recognition cost of the target behavioral intent and the second intent recognition cost of the target business intent based on the service intent graph.

[0016] The calculation module is also used to determine the intent matching degree between the target business intent and the target behavioral intent based on the first intent recognition cost and the second intent recognition cost;

[0017] The adjustment module is used to adjust at least one of the service provider's business resource combination and the service demander's behavioral resource combination when the intent matching degree is not equal to the intent matching degree threshold, so as to change the service intent graph and reduce the intent matching degree. The intent matching is determined to be complete when the intent matching degree no longer decreases.

[0018] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to perform the following steps:

[0019] To obtain the target business intent of the service provider and the target behavioral intent of the service demander;

[0020] Based on the service provider's business resource combination, business intent resources, service demander's behavioral resource combination, and behavioral intent resources, a service intent graph is determined. The service intent graph includes a data graph, an information graph, a knowledge graph, and an intent graph.

[0021] Based on the service intent graph, calculate the first intent recognition cost of the target behavioral intent and the second intent recognition cost of the target business intent, respectively.

[0022] Based on the first intent recognition cost and the second intent recognition cost, determine the intent matching degree between the target business intent and the target behavioral intent;

[0023] If the intent matching degree is not equal to the intent matching degree threshold, adjust at least one of the service provider's business resource combination and the service demander's behavioral resource combination to change the service intent graph and reduce the intent matching degree. Intent matching is considered complete when the intent matching degree no longer decreases.

[0024] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, performs the following steps:

[0025] To obtain the target business intent of the service provider and the target behavioral intent of the service demander;

[0026] Based on the service provider's business resource combination, business intent resources, service demander's behavioral resource combination, and behavioral intent resources, a service intent graph is determined. The service intent graph includes a data graph, an information graph, a knowledge graph, and an intent graph.

[0027] Based on the service intent graph, calculate the first intent recognition cost of the target behavioral intent and the second intent recognition cost of the target business intent, respectively.

[0028] Based on the first intent recognition cost and the second intent recognition cost, determine the intent matching degree between the target business intent and the target behavioral intent;

[0029] If the intent matching degree is not equal to the intent matching degree threshold, adjust at least one of the service provider's business resource combination and the service demander's behavioral resource combination to change the service intent graph and reduce the intent matching degree. Intent matching is considered complete when the intent matching degree no longer decreases.

[0030] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, performs the following steps:

[0031] To obtain the target business intent of the service provider and the target behavioral intent of the service demander;

[0032] Based on the service provider's business resource combination, business intent resources, service demander's behavioral resource combination, and behavioral intent resources, a service intent graph is determined. The service intent graph includes a data graph, an information graph, a knowledge graph, and an intent graph.

[0033] Based on the service intent graph, calculate the first intent recognition cost of the target behavioral intent and the second intent recognition cost of the target business intent, respectively.

[0034] Based on the first intent recognition cost and the second intent recognition cost, determine the intent matching degree between the target business intent and the target behavioral intent;

[0035] If the intent matching degree is not equal to the intent matching degree threshold, adjust at least one of the service provider's business resource combination and the service demander's behavioral resource combination to change the service intent graph and reduce the intent matching degree. Intent matching is considered complete when the intent matching degree no longer decreases.

[0036] The aforementioned intent matching method, apparatus, computer equipment, storage medium, and computer program product acquire the target business intent of the service provider and the target behavioral intent of the service demander; determine a service intent graph based on the service provider's business resource combination, business intent resources, the service demander's behavioral resource combination, and behavioral intent resources. The service intent graph includes a data graph, an information graph, a knowledge graph, and an intent graph; calculate the first intent recognition cost of the target behavioral intent and the second intent recognition cost of the target business intent based on the service intent graph; determine the intent matching degree between the target business intent and the target behavioral intent based on the first and second intent recognition costs; and adjust at least one of the service provider's business resource combination and the service demander's behavioral resource combination when the intent matching degree is not equal to the intent matching degree threshold, thereby changing the service intent graph to reduce the intent matching degree. Intent matching is considered complete when the intent matching degree no longer decreases. This achieves the goal of accurately matching the interaction intents of both parties, thereby maximizing the elimination of asymmetries in data, information, knowledge, and intent between the two parties and improving the interactive experience. Attached Figure Description

[0037] Figure 1 This is a diagram illustrating the application environment of the intent matching method in one embodiment;

[0038] Figure 2 This is a flowchart illustrating an intent matching method in one embodiment;

[0039] Figure 3 This is a schematic diagram of the structure of a service intent graph in one embodiment;

[0040] Figure 4 This is a flowchart illustrating the adjustment steps in one embodiment;

[0041] Figure 5 This is a schematic diagram illustrating the steps for calculating the explicit intent recognition cost in one embodiment;

[0042] Figure 6 This is a schematic diagram illustrating the step of calculating the first implicit intent recognition cost in one embodiment;

[0043] Figure 7 This is a schematic diagram illustrating the step of calculating the second implicit intent recognition cost in one embodiment;

[0044] Figure 8 This is a schematic diagram illustrating the step of calculating the cost of the third implicit intent recognition in one embodiment;

[0045] Figure 9 This is a schematic diagram illustrating the calculation of the fourth implicit intent recognition cost step in one embodiment;

[0046] Figure 10 This is a flowchart illustrating the intent matching method in another embodiment;

[0047] Figure 11 This is a structural block diagram of an intention matching device in one embodiment;

[0048] Figure 12 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0050] The intent matching method provided in this application embodiment can be applied to, for example, Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or placed on a cloud or other network server. Terminal 102 can execute the intent matching method provided in this application embodiment independently, or terminal 102 and server 104 can collaboratively execute the intent matching method provided in this application embodiment.

[0051] When terminal 102 executes the intent matching method alone, terminal 102 obtains the target business intent of the service provider and the target behavioral intent of the service demander; based on the service provider's business resource combination, business intent resources, service demander's behavioral resource combination, and behavioral intent resources, it determines a service intent graph, which includes a data graph, an information graph, a knowledge graph, and an intent graph; according to the service intent graph, it calculates the first intent recognition cost of the target behavioral intent and the second intent recognition cost of the target business intent; based on the first intent recognition cost and the second intent recognition cost, it determines the intent matching degree between the target business intent and the target behavioral intent; if the intent matching degree is not equal to the intent matching degree threshold, it adjusts at least one of the service provider's business resource combination and the service demander's behavioral resource combination to change the service intent graph, thereby reducing the intent matching degree, until the intent matching degree no longer decreases, at which point the intent matching is determined to be complete.

[0052] When terminal 102 and server 104 collaboratively execute the intent matching method, terminal 102 obtains the target business intent of the service provider and the target behavioral intent of the service demander, and sends the target business intent and target behavioral intent to server 104. Server 104 determines a service intent graph based on the service provider's business resource combination, business intent resources, the service demander's behavioral resource combination, and behavioral intent resources. The service intent graph includes a data graph, an information graph, a knowledge graph, and an intent graph. Based on the service intent graph, it calculates the first intent recognition cost of the target behavioral intent and the second intent recognition cost of the target business intent. Based on the first and second intent recognition costs, it determines the intent matching degree between the target business intent and the target behavioral intent. If the intent matching degree is not equal to the intent matching degree threshold, it adjusts at least one of the service provider's business resource combination and the service demander's behavioral resource combination to change the service intent graph, thereby reducing the intent matching degree. Intent matching is considered complete when the intent matching degree no longer decreases.

[0053] The terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, and smart in-vehicle systems. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted devices. The server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers.

[0054] It should be understood that the terms "first," "second," "third," "fourth," and similar terms used in the embodiments of this application do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Unless the context clearly indicates otherwise, the singular forms of "a," "one," or "the," etc., do not indicate a quantity limitation, but rather indicate the presence of at least one.

[0055] In one embodiment, such as Figure 2 As shown, an intent matching method is provided, which can be executed by the terminal or server alone, or by the terminal and server collaboratively. This method is applied to... Figure 1 Taking the terminal in the example, the explanation includes the following steps:

[0056] Step 202: Obtain the target business intent of the service provider and the target behavioral intent of the service demander.

[0057] In this context, the service provider and the service requester are the two parties interacting through a terminal to provide business services. The service provider is the party offering the service (e.g., an organization), and the service requester is the party receiving the service (e.g., a user). The target business intent is the service provider's intention to display a particular service, such as displaying resource borrowing information. The target behavioral intent is the service requester's intention to perform a particular service, such as viewing resource borrowing information.

[0058] For example, the terminal obtains the target business intent of the service provider and the target behavioral intent of the service demander.

[0059] Step 204: Based on the service provider's business resource combination, business intent resources, service demander's behavioral resource combination, and behavioral intent resources, determine the service intent graph, which includes data graph, information graph, knowledge graph, and intent graph.

[0060] In this context, business resource combinations are combinations of business data, business information, and business knowledge, while business intent resources represent the collection of business intents of the service provider. Business data includes at least two types of data, business information includes at least two types of information, and business knowledge includes at least two types of knowledge. For example, if business data includes three types of data, business information includes three types of information, and business knowledge includes three types of knowledge, then there are 3*3*3=27 possible combinations of business resources.

[0061] A behavioral resource portfolio is a combination of behavioral data, behavioral information, and behavioral knowledge. Behavioral intent resources represent a collection of behavioral intentions from service requesters. Specifically, behavioral data includes at least two types of data, behavioral information includes at least two types of information, and behavioral knowledge includes at least two types of knowledge. For example, if behavioral data includes three types of data, behavioral information includes three types of information, and behavioral knowledge includes three types of knowledge, then there are 3*3*3=27 possible combinations of behavioral resources.

[0062] A service intent graph is obtained by integrating data graphs, information graphs, knowledge graphs, and intent graphs. For example, if there are 3*3*3=27 combinations of business resources and 3*3*3=27 combinations of behavioral resources, then the service intent graph will contain at least 27*27=729 different service intent graphs.

[0063] refer to Figure 3 The data graph comprises a business data graph for service providers and a behavioral data graph for service demanders. Specifically, the business data graph for service providers is obtained by organizing and storing business data using a graph structure of data nodes and relationships; similarly, the behavioral data graph for service demanders is obtained by organizing and storing behavioral data using the same graph structure. The nodes in the data graph are raw data, categorized through observation of structured and unstructured resources. They are individual pieces of information with limited context (e.g., annualized interest rate as business data without context in the business data graph).

[0064] Business data refers to the context-free data within the upper-level framework that the interactive system needs to display, representing knowledge and information. Examples include resource borrowing values, interest rate values, identifier strings, resource borrowing periods, resource borrowing dates, and revenue values ​​for non-physical resource products. Behavioral data consists of data generated during interactions, such as clicks, swipes, dwell time, and user profile models.

[0065] An information graph comprises a business information graph for service providers and a behavioral information graph for service demanders. Specifically, a business information graph for service providers is obtained by organizing and storing business information using a graph structure of data nodes and interaction relationships; similarly, a behavioral information graph for service demanders is obtained by organizing and storing behavioral information using the same graph structure. Each node in an information graph is a piece of data, i.e., information. Information is related to interaction intent; currently, relational databases can store content that can be considered information. Information conveys context through data combinations, allowing for coherent observation.

[0066] Business information is a combination of business data with inherent uncertainties, such as the revenue trajectory of non-physical resource products over a specific period. Behavioral information includes access patterns and user profile information.

[0067] A knowledge graph comprises a business knowledge graph for service providers and a behavioral knowledge graph for service demanders. Specifically, the business knowledge graph for service providers is obtained by organizing and storing business knowledge through a graph structure of content nodes and relationships; similarly, the behavioral knowledge graph for service demanders is obtained by organizing and storing business knowledge through the same graph structure. The nodes in a knowledge graph represent knowledge, which embodies the ability to understand, interpret, and determine concepts, actions, and intentions. It involves categorical reasoning based on statistical experience, gaining general understanding and knowledge from accumulated information, collecting rules from data, and aggregating information.

[0068] Business knowledge refers to content with patterns and concepts, such as question-and-answer statistics and sales rankings. Behavioral knowledge is obtained during user interactions. For example, if a user recently applied for resource borrowing from a company or organization, then the user's clicks on behavioral knowledge are more likely related to resource borrowing services.

[0069] An intent graph comprises the business intent graph of service providers and the behavioral intent graph of service demanders. Specifically, the business intent graph of service providers is obtained by organizing and storing business intents through intent nodes and their associated graph structures; the behavioral intent graph of service demanders is obtained by organizing and storing behavioral intents through intent nodes and their associated graph structures. The nodes in the intent graph are intents, which are planning instances chosen by an agent to achieve a specific goal. The role of intents is to guide rational decision-making and plan future behavior.

[0070] Business intent refers to the visual instances of data, information, and knowledge presented to different service providers. For example, for users borrowing resources to purchase real estate, organizations would prefer to recommend relevant support resource borrowing products. Behavioral intent refers to instances planned by users to achieve business objectives, such as viewing resource data.

[0071] For example, the terminal determines a service intent graph based on the service provider's business resource combination, business intent resources, service demander's behavioral resource combination, and behavioral intent resources. The service intent graph includes a data graph, an information graph, a knowledge graph, and an intent graph.

[0072] Step 206: Based on the service intent graph, calculate the first intent recognition cost of the target behavioral intent and the second intent recognition cost of the target business intent, respectively.

[0073] The first intent recognition cost is the intent recognition cost of the service demander's target behavioral intent in the service intent graph, and the second intent recognition cost is the intent recognition cost of the service provider's target business intent in the service intent graph.

[0074] For example, the terminal calculates the first intent recognition cost of the target behavioral intent and the second intent recognition cost of the target business intent based on the service intent graph.

[0075] Step 208: Based on the first intent recognition cost and the second intent recognition cost, determine the intent matching degree between the target business intent and the target behavioral intent.

[0076] Among them, the intent matching degree characterizes the degree of matching between the service provider's target business intent and the service demander's target behavioral intent. The smaller the intent matching degree, the more conducive it is to eliminating the asymmetry between the service provider and the service demander in terms of data, information, knowledge, intent, etc., and the easier it is for the service provider's target business intent and the service demander's target behavioral intent to match successfully.

[0077] For example, the terminal determines the intent matching degree between the target business intent and the target behavioral intent based on the first intent recognition cost and the second intent recognition cost.

[0078] Step 210: If the intent matching degree is not equal to the intent matching degree threshold, adjust at least one of the service provider's business resource combination and the service demander's behavioral resource combination to change the service intent graph and reduce the intent matching degree. When the intent matching degree no longer decreases, the intent matching is determined to be complete.

[0079] The intent matching threshold is preset according to intent matching requirements, and this application embodiment does not limit it.

[0080] For example, when the intent matching degree is not equal to the intent matching degree threshold, the terminal adjusts at least one of the service provider's business resource combination and the service demander's behavioral resource combination to obtain an adjusted service intent graph obtained by integrating at least one of the adjusted business resource combination and the service demander's behavioral resource combination. If the current intent matching degree determined based on the adjusted service intent graph is less than the previous intent matching degree (in the case of the first adjustment, the previous intent matching degree is the intent matching degree determined based on the service intent graph; in the case of the Nth adjustment, N is an integer greater than or equal to 2, and the previous intent matching degree is the intent matching degree determined based on the service intent graph adjusted in the (N-1)th time), the adjustment continues until all service intent graphs are exhausted and the minimum intent matching degree is obtained. At this point, the intent matching is determined to be complete.

[0081] In the aforementioned intent matching method, the target business intent of the service provider and the target behavioral intent of the service demander are obtained. Based on the service provider's business resource combination, business intent resources, the service demander's behavioral resource combination, and behavioral intent resources, a service intent graph is determined. The service intent graph includes a data graph, an information graph, a knowledge graph, and an intent graph. According to the service intent graph, the first intent recognition cost of the target behavioral intent and the second intent recognition cost of the target business intent are calculated respectively. Based on the first intent recognition cost and the second intent recognition cost, the intent matching degree between the target business intent and the target behavioral intent is determined. If the intent matching degree is not equal to the intent matching degree threshold, at least one of the service provider's business resource combination and the service demander's behavioral resource combination is adjusted to change the service intent graph and reduce the intent matching degree. Intent matching is considered complete when the intent matching degree no longer decreases. This method can accurately match the interaction intents of both parties, thereby maximizing the elimination of asymmetry in data, information, knowledge, intent, and other content between the two parties and improving the interactive experience.

[0082] In one embodiment, such as Figure 4 As shown, when the intent matching degree is not equal to the intent matching degree threshold, at least one of the service provider's business resource combination and the service demander's behavioral resource combination is adjusted to change the service intent graph, thereby reducing the intent matching degree. Intent matching is determined to be complete when the intent matching degree no longer decreases. This includes:

[0083] Step 402: When the intent matching degree is greater than the intent matching degree threshold and the cost of recognizing the first intent is greater than the cost of recognizing the second intent, adjust the business resource combination of the service provider to change the service intent graph to reduce the intent matching degree until the intent matching degree no longer decreases, and determine that the intent matching is complete.

[0084] For example, when there are 3*3*3=27 possible combinations of service resources, if the terminal's intent matching degree is greater than the intent matching degree threshold and the cost of recognizing the first intent is greater than the cost of recognizing the second intent, the terminal adjusts the service provider's service resource combination. That is, it selects any one of the 26 service resource combinations to replace the current service provider's service resource combination. If the current intent matching degree determined based on the adjusted service intent graph is greater than the previous intent matching degree, the adjustment continues until the current intent matching degree determined based on the adjusted service intent graph is less than the previous intent matching degree. If the current intent matching degree determined based on the adjusted service intent graph is less than the previous intent matching degree, the adjustment continues until all the adjusted service intent graphs corresponding to the 26 service resource combinations are exhausted, and the minimum intent matching degree is obtained. At this point, the intent matching is determined to be complete.

[0085] Step 404: When the intent matching degree is greater than the intent matching degree threshold and the cost of recognizing the first intent is less than the cost of recognizing the second intent, adjust the combination of behavioral resources of the service demander to change the service intent graph to reduce the intent matching degree until the intent matching degree no longer decreases, and determine that the intent matching is complete.

[0086] For example, when there are 3*3*3=27 possible combinations of behavioral resources, if the terminal's intent matching degree is greater than the intent matching degree threshold and the cost of recognizing the first intent is less than the cost of recognizing the second intent, the terminal adjusts the behavioral resource combination of the service requester. That is, it selects any one of the 26 behavioral resource combinations to replace the current behavioral resource combination of the service requester. If the current intent matching degree determined based on the adjusted service intent graph is greater than the previous intent matching degree, the adjustment continues until the current intent matching degree determined based on the adjusted service intent graph is less than the previous intent matching degree. If the current intent matching degree determined based on the adjusted service intent graph is less than the previous intent matching degree, the adjustment continues until all the adjusted service intent graphs corresponding to the 26 behavioral resource combinations are exhausted, and the minimum intent matching degree is obtained. At this point, the intent matching is determined to be complete.

[0087] Step 406: If the intent matching degree is less than the intent matching degree threshold, adjust at least one of the service provider's business resource combination and the service demander's behavioral resource combination to change the service intent graph and reduce the intent matching degree until the intent matching degree no longer decreases, and determine that intent matching is complete.

[0088] For example, when there are 3*3*3=27 combinations of business resources and 3*3*3=27 combinations of behavioral resources, if the terminal's intent matching degree is less than the intent matching degree threshold, it adjusts at least one of the service provider's business resource combination and the service demander's behavioral resource combination. Specifically, it selects only one of the 26 business resource combinations to replace the current service provider's business resource combination, selects only one of the 26 behavioral resource combinations to replace the current service demander's behavioral resource combination, or simultaneously selects one of the 26 business resource combinations to replace the current service provider's business resource combination and selects one of the 26 behavioral resource combinations to replace the current service demander's behavioral resource combination. Choose any combination of behavioral resources from the source combination to replace the current service demander. If the current intent matching degree determined based on the adjusted service intent graph is greater than the previous intent matching degree, continue adjusting until the current intent matching degree determined based on the adjusted service intent graph is less than the previous intent matching degree. If the current intent matching degree determined based on the adjusted service intent graph is less than the previous intent matching degree, continue adjusting until all 728 (26+26+26*26=728) business resource combinations corresponding to the adjusted service intent graph are exhausted, and the minimum intent matching degree is obtained. At this point, the intent matching is determined to be complete.

[0089] In this embodiment, the situation where the intent matching degree is not equal to the intent matching degree threshold is divided into three cases: the intent matching degree is greater than the intent matching degree threshold and the cost of recognizing the first intent is greater than the cost of recognizing the second intent; the intent matching degree is greater than the intent matching degree threshold and the cost of recognizing the first intent is less than the cost of recognizing the second intent; and the intent matching degree is less than the intent matching degree threshold. Different adjustment methods are used to reduce the intent matching degree in these three cases. This achieves the goal of adjusting at least one of the service provider's business resource combination and the service demander's behavioral resource combination when the intent matching degree is not equal to the intent matching degree threshold, so as to change the service intent graph and reduce the intent matching degree. The intent matching is determined to be completed when the intent matching degree no longer decreases.

[0090] In one embodiment, determining a service intent graph based on the service provider's business resource combination, business intent resources, the service demander's behavioral resource combination, and behavioral intent resources includes: determining multiple behavioral resource combinations based on the service demander's historical behavioral resources, wherein historical behavioral resources include historical behavioral data, historical behavioral information, and historical behavioral knowledge; constructing a first service intent graph for the service demander based on the multiple behavioral resource combinations and behavioral intent resources, wherein behavioral intent resources represent the set of historical behavioral intents of the service demander; determining multiple business resource combinations based on the service provider's historical business resources, wherein historical business resources include historical business data, historical business information, and historical business knowledge; constructing a second service intent graph for the service provider based on the multiple business resource combinations and business intent resources, wherein business intent resources represent the set of historical business intents of the service provider; and integrating the first and second service intent graphs to obtain the service intent graph.

[0091] Among them, behavioral intent resources are collections of behavioral intents gathered within a historical time period. Historical behavioral data, historical behavioral information, and historical behavioral knowledge are behavioral data, behavioral information, and behavioral knowledge collected within a historical time period, respectively. Business intent resources are collections of business intents gathered within a historical time period. Historical business data, historical business information, and historical business knowledge are business data, business information, and business knowledge collected within a historical time period, respectively.

[0092] The first service intent graph corresponds to the service demander and is obtained by integrating the service demander's behavioral data graph, behavioral information graph, behavioral knowledge graph, and behavioral intent graph. The second service intent graph corresponds to the service provider and is obtained by integrating the service provider's business data graph, business information graph, business knowledge graph, and business intent graph.

[0093] For example, the terminal determines a combination of various behavioral resources based on the historical behavioral resources of the service demander. The historical behavioral resources include historical behavioral data, historical behavioral information, and historical behavioral knowledge. A behavioral data graph of the service demander is constructed based on the historical behavioral data, a behavioral information graph of the service demander is constructed based on the historical behavioral information, a behavioral knowledge graph of the service demander is constructed based on the historical behavioral knowledge, and a behavioral intent graph of the service demander is constructed based on the historical behavioral intent. The behavioral data graph, behavioral information graph, behavioral knowledge graph, and behavioral intent graph are integrated to obtain the first service intent graph of the service demander.

[0094] The terminal determines a combination of various business resources based on the service provider's historical business resources. These historical business resources include historical business data, historical business information, and historical business knowledge. A business data graph of the service provider is constructed based on the historical business data; a business information graph is constructed based on the historical business information; a business knowledge graph is constructed based on the historical business knowledge; and a business intent graph is constructed based on the historical business intent. These graphs are then integrated to obtain the service provider's second service intent graph.

[0095] The terminal integrates and cleans the first and second service intent graphs to obtain a service intent graph.

[0096] In this embodiment, the purpose of constructing a service intent graph can be achieved by combining the service provider's business resources, business intent resources, the service demander's behavioral resources, and behavioral intent resources.

[0097] In one embodiment, calculating the first intent recognition cost of the target behavioral intent and the second intent recognition cost of the target business intent based on the service intent graph includes: calculating the explicit intent recognition cost corresponding to the target behavioral intent and the explicit intent recognition cost corresponding to the target business intent based on the intent graph in the service intent graph; calculating the first implicit intent recognition cost corresponding to the target behavioral intent and the first implicit intent recognition cost corresponding to the target business intent based on the data graph in the service intent graph; calculating the second implicit intent recognition cost corresponding to the target behavioral intent and the second implicit intent recognition cost corresponding to the target business intent based on the information graph in the service intent graph; and calculating the second implicit intent recognition cost corresponding to the target behavioral intent based on the knowledge graph in the service intent graph. The cost of recognizing the third implicit intent and the cost of recognizing the third implicit intent corresponding to the target business intent; the cost of recognizing the fourth implicit intent corresponding to the target behavioral intent and the cost of recognizing the fourth implicit intent corresponding to the target business intent according to the knowledge graph in the service intent graph; the sum of the cost of recognizing the explicit intent, the cost of recognizing the first implicit intent, the cost of recognizing the second implicit intent, the cost of recognizing the third implicit intent and the cost of recognizing the fourth implicit intent corresponding to the target behavioral intent is taken as the first intent recognition cost of the target behavioral intent; the sum of the cost of recognizing the explicit intent, the cost of recognizing the first implicit intent, the cost of recognizing the second implicit intent, the cost of recognizing the third implicit intent and the cost of recognizing the fourth implicit intent corresponding to the target business intent is taken as the second intent recognition cost of the target business intent.

[0098] Explicit intents are the existing interactive behavior paths and interactive business path information in the service intent graph. Implicit intents are path information inferred from explicit intents, combinations of behavioral resources, and combinations of business resources. They are not intents that exist in the form of direct intents, but rather exist in data, information, knowledge, and intents.

[0099] The cost of identifying explicit intent corresponding to the target behavioral intent ( The search cost of the target behavioral intent on the service intent graph (DIKP, i.e., Data-Information-Knowledge-Purpose) is related to the size of the intent graph in the service intent graph and the size of the set of associations with the target behavioral intent.

[0100] The cost of implicit intent recognition corresponding to the target behavior intent ( The costs are the first, second, third, and fourth implicit intention recognition costs corresponding to the target behavioral intention. Among them, the first implicit intention recognition cost (…) ) is the cost of identifying the implicit intent in the data graph (D) of the service intent graph, and the cost of identifying the second implicit intent corresponding to the target behavioral intent. ) is the cost of identifying the implicit intent in the information graph (I) of the service intent graph, and the cost of identifying the third implicit intent corresponding to the target behavioral intent. ) is the cost of identifying the implicit intent of the target behavior intent on the knowledge graph (K) in the service intent graph, and the cost of identifying the fourth implicit intent corresponding to the target behavior intent. ) is the cost of identifying implicit intents in the intent graph (P) of the service intent graph for the target behavioral intent.

[0101] The explicit intent recognition cost corresponding to the target business intent is the search cost of the target business intent in the Service Intent Graph (DIKP), which is related to the size of the intent graph in the Service Intent Graph and the size of the association set with the target business intent. The implicit intent recognition cost corresponding to the target business intent includes the first, second, third, and fourth implicit intent recognition costs. Specifically, the first implicit intent recognition cost is the cost of recognizing the implicit intent in the data graph of the Service Intent Graph; the second implicit intent recognition cost is the cost of recognizing the implicit intent in the information graph of the Service Intent Graph; the third implicit intent recognition cost is the cost of recognizing the implicit intent in the knowledge graph of the Service Intent Graph; and the fourth implicit intent recognition cost is the cost of recognizing the implicit intent in the intent graph of the Service Intent Graph.

[0102] For example, the terminal calculates the explicit intent recognition cost corresponding to the target behavioral intent based on the intent graph in the service intent graph. Based on the data graph in the service intent graph, calculate the first implicit intent recognition cost corresponding to the target behavioral intent. Based on the information graph in the service intent graph, calculate the second implicit intent recognition cost corresponding to the target behavioral intent. Based on the knowledge graph in the service intent graph, calculate the cost of recognizing the third implicit intent corresponding to the target behavioral intent. Based on the knowledge graph in the service intent graph, calculate the cost of recognizing the fourth implicit intent corresponding to the target behavioral intent. The sum of the first implicit intent recognition cost, the second implicit intent recognition cost, the third implicit intent recognition cost, and the fourth implicit intent recognition cost is taken as the implicit intent recognition cost corresponding to the target behavior intent. The sum of the implicit intent recognition cost corresponding to the target behavior intent and the explicit intent recognition cost corresponding to the target behavior intent is taken as the first intent recognition cost of the target behavior intent. That is:

[0103]

[0104] .

[0105] Similarly, the terminal calculates the explicit intent recognition cost corresponding to the target service intent based on the intent graph in the service intent graph; calculates the first implicit intent recognition cost corresponding to the target service intent based on the data graph in the service intent graph; calculates the second implicit intent recognition cost corresponding to the target service intent based on the information graph in the service intent graph; calculates the third implicit intent recognition cost corresponding to the target service intent based on the knowledge graph in the service intent graph; calculates the fourth implicit intent recognition cost corresponding to the target service intent based on the knowledge graph in the service intent graph; and uses the sum of the explicit intent recognition cost, the first implicit intent recognition cost, the second implicit intent recognition cost, the third implicit intent recognition cost, and the fourth implicit intent recognition cost corresponding to the target service intent as the second intent recognition cost of the target service intent.

[0106] In this embodiment, by using the data graph, information graph, knowledge graph and intent graph in the service intent graph, it is possible to calculate the first intent recognition cost of the target behavioral intent and the second intent recognition cost of the target business intent.

[0107] In one embodiment, the explicit intent recognition cost corresponding to the target behavioral intent is calculated based on the intent graph in the service intent graph, including: determining the intent scale of the intent graph in the service intent graph; determining the association set corresponding to the target behavioral intent and the size of the association set in the intent graph in the service intent graph; and using the sum of the intent scale and the size of the association set as the explicit intent recognition cost corresponding to the target behavioral intent.

[0108] The intent scale within the service intent graph is the size of the intent graph itself, which is the number of nodes it includes. The association set corresponding to the target behavior intent is the set of intents associated with the target behavior intent within the service intent graph; that is, the set of nodes connected to the node representing the target behavior intent. The size of the association set, also known as association frequency, is the number of intents included in the association set, which is the sum of the number of times each node connected to the node representing the target behavior intent appears in the service intent graph.

[0109] For example, such as Figure 5 As shown, the terminal counts the intent scale of the intent graph in the service intent graph, and counts the association set scale of the association set corresponding to the target behavior intent in the intent graph. The explicit intent recognition cost corresponding to the target behavior intent is calculated according to the following formula.

[0110] EXCOST1=SCALE INTENSION_GRAPH+AssociationNum|AssociationSet

[0111] In the above formula, This represents the size of the intent graph within the service intent graph, with a value of 8. This is the size of the associative set, with a value of 3. It is the explicit intent recognition cost corresponding to the target behavioral intent, with a value of 11.

[0112] In this embodiment, by statistically analyzing the intent scale of the intent graph in the service intent graph and the size of the association set corresponding to the target behavioral intent, the explicit intent recognition cost corresponding to the target behavioral intent can be calculated.

[0113] In one embodiment, the explicit intent recognition cost corresponding to the target business intent is calculated based on the intent graph in the service intent graph, including: determining the intent scale of the intent graph in the service intent graph; determining the association set corresponding to the target business intent in the intent graph in the service intent graph, and the association set scale of the association set corresponding to the target business intent; and using the sum of the intent scale and the association set scale of the association set corresponding to the target business intent as the explicit intent recognition cost corresponding to the target business intent.

[0114] In one embodiment, the service intent graph is obtained by integrating a first service intent graph corresponding to the service demander and a second service intent graph corresponding to the service provider. Based on the data graph in the service intent graph, the first implicit intent recognition cost corresponding to the target behavioral intent is calculated, including: calculating the first data frequency of each data in the data graph of the service intent graph appearing in the data graph of the first service intent graph, and the second data frequency of each data appearing in the data graph of the second service intent graph; weighting and summing the first data frequency and the second data frequency to obtain the comprehensive data frequency of each data in the data graph of the service intent graph; determining a first preset condition based on the target behavioral intent; and weighting and summing the comprehensive data frequencies of the data that meet the first preset condition to obtain the first implicit intent recognition cost corresponding to the target behavioral intent.

[0115] The first preset condition is determined based on the target behavioral intent, and this embodiment does not limit this. For example, the first preset condition may be that the overall data frequency is greater than the overall data frequency threshold, where the overall data frequency threshold is preset according to the intent matching requirements, and this embodiment does not limit this.

[0116] Identifying intents within the data graph of the service intent graph involves statistically analyzing the frequency of data associations to obtain a comprehensive data frequency. If the comprehensive data frequency is greater than the comprehensive data frequency threshold (Frequency 0), it is considered an implicit intent, and a set is formed from the nodes in the data graph whose comprehensive data frequency is greater than the threshold. This set is a collection of implicit intentions identified from the data graph.

[0117] Data association frequency includes a first data frequency and a second data frequency. The first data frequency is the frequency of each data point in the data graph of the service intent graph appearing in the data graph of the first service intent graph. The second data frequency is the frequency of each data point in the data graph of the service intent graph appearing in the data graph of the second service intent graph. The combined data frequency is the combined frequency of each data point in the data graph of the service intent graph appearing in the data graph of the service intent graph.

[0118] The service intent graph is obtained by integrating the first service intent graph corresponding to the service demander and the second service intent graph corresponding to the service provider. Therefore, the data graph in the service intent graph includes both the data graph in the first service intent graph (also known as the behavioral data graph) and the data graph in the second service intent graph (also known as the business data graph). Any data in the data graph of the service intent graph may appear only in the data graph of the first service intent graph, only in the data graph of the second service intent graph, or simultaneously in the data graphs of both the first and second service intent graphs.

[0119] For example, the terminal statistical service intent graph shows the first data frequency of each data in the data graph of the first service intent graph and the second data frequency of each data in the data graph of the second service intent graph; the comprehensive data frequency of each data in the data graph of the service intent graph is calculated according to the following formula.

[0120]

[0121] In the above formula, It is the highest data frequency in the data graph (DG|serviceNeeder) within the first service intent graph. h1 is the second data frequency of the data appearing in the data graph (DG|serviceProvider) in the second service intent graph; h2 is the difficulty coefficient of identifying the appearance of each data in the data graph of the service intent graph in the data graph of the first service intent graph; and h3 is the difficulty coefficient of identifying the appearance of each data in the data graph of the service intent graph in the data graph of the second service intent graph.

[0122] When the data appears only in the data graph of the first service intent graph, h2 is 0; when the data appears only in the data graph of the second service intent graph, h1 is 0; when the data appears in the data graphs of both the first and second service intent graphs, the values ​​of h1 and h2 are obtained by training a machine learning model, and the sum of h1 and h2 is 1.

[0123] like Figure 6 As shown, the terminal determines the target's behavioral intent ( Figure 6 The target behavioral intent shown is "view resource borrowing information". The first preset condition is that the comprehensive data frequency is greater than the comprehensive data frequency threshold (for example, the comprehensive data frequency threshold is 70). Based on the first preset condition, the set of data nodes that meet the first preset condition is determined to be {"resource borrowing product A", "resource borrowing product B", "personal resource borrowing H"}. The first implicit intent recognition cost corresponding to the target behavioral intent is calculated according to the following formula.

[0124]

[0125] In the above formula, It is the comprehensive data frequency of each data point among the n data points that satisfy the first preset condition. It is a set of nodes in the data graph of the service intent graph that represent data that meet the first preset condition. It is the weighting coefficient of the comprehensive data frequency of each data point. It can be a preset constant or obtained through training. It is the cost of recognizing the first implicit intent corresponding to the target behavior intent, and its value is 108.8 (i.e. 0.4*120 + 0.4*112 + 0.2*80 = 108.8).

[0126] In this embodiment, by calculating the comprehensive data frequency of each data in the data graph of the service intent graph, determining the first preset condition according to the target behavior intent, and weighting and summing the comprehensive data frequencies of the data that meet the first preset condition, the purpose of calculating the first implicit intent recognition cost corresponding to the target behavior intent can be achieved.

[0127] In one embodiment, the service intent graph is obtained by integrating a first service intent graph corresponding to the service demander and a second service intent graph corresponding to the service provider. Based on the data graph in the service intent graph, the first implicit intent recognition cost corresponding to the target business intent is calculated, including: calculating the first data frequency of each data in the data graph of the service intent graph appearing in the data graph of the first service intent graph, and the second data frequency of each data appearing in the data graph of the second service intent graph; weighting and summing the first data frequency and the second data frequency to obtain the comprehensive data frequency of each data in the data graph of the service intent graph; determining a fifth preset condition based on the target business intent; and weighting and summing the comprehensive data frequencies of the data that meet the fifth preset condition to obtain the first implicit intent recognition cost corresponding to the target business intent.

[0128] The fifth preset condition is determined based on the target business intent, and this application embodiment does not limit it. For example, the fifth preset condition may be that the comprehensive data frequency is greater than the comprehensive data frequency threshold, where the comprehensive data frequency threshold is preset according to the intent matching requirements, and this application embodiment does not limit it.

[0129] In one embodiment, the service intent graph is obtained by integrating a first service intent graph corresponding to the service demander and a second service intent graph corresponding to the service provider. Based on the information graph in the service intent graph, the second implicit intent recognition cost corresponding to the target behavioral intent is calculated, including: counting the frequency of each piece of information in the information graph of the service intent graph appearing in the information graph of the first service intent graph, and the frequency of each piece of information appearing in the information graph of the second service intent graph; weighting and summing the first and second information frequencies to obtain the comprehensive information frequency of each piece of information appearing in the information graph of the service intent graph; determining a second preset condition based on the target behavioral intent; and weighting and summing the comprehensive information frequencies of the information that meet the second preset condition to obtain the second implicit intent recognition cost corresponding to the target behavioral intent.

[0130] The second preset condition is determined based on the target behavioral intent, and this embodiment does not limit this. For example, the second preset condition may be that the comprehensive information frequency is greater than the comprehensive information frequency threshold, where the comprehensive information frequency threshold is preset according to the intent matching requirements, and this embodiment does not limit this.

[0131] Identifying intents within the information graph of a service intent graph involves statistically analyzing the frequency of information interactions. This process, performed at the behavioral interaction level, analyzes dynamic data to obtain a comprehensive information frequency (Interaction). If the comprehensive information frequency Interaction is greater than the comprehensive information frequency threshold Interaction0, it is considered an implicit intent, and a set is formed from the nodes in the information graph whose comprehensive information frequency is greater than the threshold. This set is a set of implicit intentions identified from the information graph.

[0132] The frequency of information interaction includes a first information frequency and a second information frequency. The first information frequency is the frequency of each piece of information in the information graph of the service intent graph appearing in the information graph of the first service intent graph. The second information frequency is the frequency of each piece of information in the information graph of the service intent graph appearing in the information graph of the second service intent graph. The combined information frequency is the combined frequency of each piece of information in the information graph of the service intent graph appearing in the information graph of the service intent graph.

[0133] The service intent graph is obtained by integrating the first service intent graph corresponding to the service demander and the second service intent graph corresponding to the service provider. Therefore, the information graph in the service intent graph includes both the information graph in the first service intent graph (also known as the behavioral information graph) and the information graph in the second service intent graph (also known as the business information graph). Any piece of information in the information graph of the service intent graph may appear only in the information graph of the first service intent graph, only in the information graph of the second service intent graph, or simultaneously in the information graphs of both the first and second service intent graphs.

[0134] For example, the terminal counts the frequency of each piece of information in the information graph of the service intent graph, the frequency of each piece of information appearing in the information graph of the first service intent graph, and the frequency of each piece of information appearing in the information graph of the second service intent graph; the comprehensive frequency of each piece of information appearing in the information graph of the service intent graph is calculated according to the following formula.

[0135]

[0136] In the above formula, It is the frequency of the information appearing in the information graph (IG|serviceNeeder) within the first service intent graph. h1 is the frequency of the second piece of information appearing in the information graph (IG|serviceProvider) of the second service intent graph; h2 is the difficulty coefficient of identifying the appearance of each piece of information in the information graph of the first service intent graph; and h3 is the difficulty coefficient of identifying the appearance of each piece of information in the information graph of the second service intent graph.

[0137] When the information appears only in the information graph of the first service intent graph, h2 is 0; when the information appears only in the information graph of the second service intent graph, h1 is 0; when the information appears in the information graphs of both the first and second service intent graphs, the values ​​of h1 and h2 are obtained by training a machine learning model, and the sum of h1 and h2 is 1.

[0138] like Figure 7 As shown, the terminal determines the target's behavioral intent ( Figure 7 The target behavioral intent shown is "view resource borrowing information". The second preset condition is that the comprehensive information frequency is greater than the comprehensive information frequency threshold (e.g., the comprehensive information frequency threshold is 75). Based on the second preset condition, the set of nodes that meet the second preset condition is determined as {"apply for resource borrowing product A", "view the interest rate of resource borrowing product A", "view the status of personal resource borrowing H"}. The second implicit intent recognition cost corresponding to the target behavioral intent is calculated according to the following formula.

[0139]

[0140] In the above formula, It is the comprehensive information frequency of each of the n pieces of information that satisfy the second preset condition. It is a set of nodes in the information graph of the service intent graph that represent information that meets the second preset condition. It is the weighting coefficient of the comprehensive information frequency of each piece of information. It can be a preset constant or obtained through training. It is the cost of recognizing the second implicit intent corresponding to the target behavior intent, and its value is 108.8 (i.e. 0.4*120 + 0.4*112 + 0.2*80 = 108.8).

[0141] In this embodiment, by calculating the comprehensive information frequency of each piece of information in the information graph of the service intent graph, determining the second preset condition according to the target behavior intent, and weighting and summing the comprehensive information frequencies of the information that meet the second preset condition, the purpose of calculating the second implicit intent recognition cost corresponding to the target behavior intent can be achieved.

[0142] In one embodiment, the service intent graph is obtained by integrating the first service intent graph corresponding to the service demander and the second service intent graph corresponding to the service provider. Based on the information graph in the service intent graph, the second implicit intent recognition cost corresponding to the target business intent is calculated, including: counting the frequency of each piece of information in the information graph of the service intent graph appearing in the information graph of the first service intent graph and the frequency of each piece of information appearing in the information graph of the second service intent graph; weighting and summing the first and second information frequencies to obtain the comprehensive information frequency of each piece of information appearing in the information graph of the service intent graph; determining a sixth preset condition based on the target business intent; and weighting and summing the comprehensive information frequencies of the information that meet the sixth preset condition to obtain the first implicit intent recognition cost corresponding to the target business intent.

[0143] The sixth preset condition is determined based on the target business intent, and this application embodiment does not limit it. For example, the sixth preset condition may be that the comprehensive information frequency is greater than the comprehensive information frequency threshold, where the comprehensive information frequency threshold is preset according to the intent matching requirements, and this application embodiment does not limit it.

[0144] In one embodiment, the service intent graph is obtained by integrating the first service intent graph corresponding to the service demander and the second service intent graph corresponding to the service provider. Based on the knowledge graph in the service intent graph, the third implicit intent recognition cost corresponding to the target behavioral intent is calculated, including: statistically analyzing the first path confidence and the second path confidence of each path in the knowledge graph of the service intent graph, wherein the first path confidence represents the frequency of the path appearing in the knowledge graph of the first service intent graph, and the second path confidence represents the frequency of the path appearing in the knowledge graph of the second service intent graph; weighting and summing the first path confidence and the second path confidence to obtain the comprehensive path confidence of each path; determining the third preset condition based on the target behavioral intent; and weighting and summing the comprehensive path confidence of the paths that satisfy the third preset condition to obtain the third implicit intent recognition cost corresponding to the target behavioral intent.

[0145] The third preset condition is determined based on the target behavior intent, and this embodiment does not limit it. For example, the third preset condition may be that the comprehensive path confidence is greater than the comprehensive path confidence threshold, where the comprehensive path confidence threshold is preset according to the intent matching requirements, and this embodiment does not limit it.

[0146] Identifying intent within the knowledge graph of the service intent graph involves mathematical calculations based on path confidence during the knowledge reasoning process. This utilizes various relational and reasoning techniques to perform human-like reasoning, resulting in a comprehensive path confidence score. If this comprehensive path confidence score is greater than the threshold confidence0, it is considered an implicit intent. A set is then formed from paths in the knowledge graph whose comprehensive path confidence score is greater than the threshold confidence score. This set is a set of implicit intentions identified from the knowledge graph.

[0147] Path confidence includes first path confidence and second path confidence. First path confidence represents the frequency of a path's occurrence in the knowledge graph of the first service intent graph, while second path confidence represents the frequency of a path's occurrence in the knowledge graph of the second service intent graph. Comprehensive path confidence represents the combined frequency of a path's occurrence in the knowledge graph of the service intent graph.

[0148] The service intent graph is obtained by integrating the first service intent graph corresponding to the service demander and the second service intent graph corresponding to the service provider. Therefore, the knowledge graph in the service intent graph includes both the knowledge graph in the first service intent graph (also known as the behavioral knowledge graph) and the knowledge graph in the second service intent graph (also known as the business knowledge graph). Any knowledge in the knowledge graph of the service intent graph may appear only in the knowledge graph of the first service intent graph, only in the knowledge graph of the second service intent graph, or simultaneously in the knowledge graphs of both the first and second service intent graphs.

[0149] For example, the first path confidence and the second path confidence of each path in the knowledge graph of the terminal statistics service intent graph are calculated according to the following formula to obtain the comprehensive path confidence of each path.

[0150]

[0151] In the above formula, It is the first path confidence score of each path in the knowledge graph of the service intent graph, that is, the frequency of the path appearing in the knowledge graph (KG|serviceNeeder) of the first service intent graph. h1 is the second path confidence of each path in the knowledge graph of the service intent graph, that is, the frequency of the path appearing in the knowledge graph (KG|serviceProvider) of the second service intent graph. h2 is the difficulty coefficient of identifying the appearance of each path in the knowledge graph of the service intent graph in the knowledge graph of the first service intent graph.

[0152] When the path appears only in the knowledge graph of the first service intent graph, h2 is 0; when the path appears only in the knowledge graph of the second service intent graph, h1 is 0; when the path appears in the knowledge graphs of both the first and second service intent graphs, the values ​​of h1 and h2 are obtained by training a machine learning model, and the sum of h1 and h2 is 1.

[0153] like Figure 8 As shown, the terminal determines the target's behavioral intent ( Figure 8 The target behavior intent shown is "view resource borrowing information". The third preset condition is that the overall path confidence score is greater than the overall path confidence score threshold (e.g., the overall path confidence score threshold is 110). Based on the third preset condition, the set of paths that satisfy the third preset condition is determined as {"view resource borrowing information -> apply for resource borrowing product A -> fill in the application for resource borrowing information A", "view resource borrowing information -> view the interest rate of resource borrowing product A -> apply for resource borrowing product A"} (wherein, the overall path confidence score of the path "view resource borrowing information -> apply for resource borrowing product A -> fill in the application for resource borrowing information A" is determined by...). Figure 8 The weighted sum of 100 and 140 is given, for example, 100*0.2 + 140*0.8 = 132; the comprehensive path confidence of the path "View resource borrowing information -> View the interest rate of resource borrowing product A -> Apply for resource borrowing product A" is determined by... Figure 8 The weighted sum of 110 and 114 shown is, for example, 110 * 0.2 + 114 * 0.8 = 113.2); the third implicit intent recognition cost corresponding to the target behavioral intent is calculated according to the following formula.

[0154]

[0155] In the above formula, It is the comprehensive path confidence of each of the n paths that satisfy the third preset condition. It is the set of paths in the knowledge graph of the service intent graph that satisfy the third preset condition. It is the weighting coefficient of the overall path confidence of each path, which can be a preset constant or obtained through training. It is the cost of recognizing the third implicit intent corresponding to the target behavior intent, and its value is 122.6 (i.e. 0.5*132 + 0.5*113.2 = 122.6).

[0156] In this embodiment, by calculating the comprehensive path confidence of each path, determining the third preset condition based on the target behavior intention, and weighting and summing the comprehensive path confidence of the paths that meet the third preset condition, the purpose of calculating the third implicit intent recognition cost corresponding to the target behavior intention can be achieved.

[0157] In one embodiment, the service intent graph is obtained by integrating the first service intent graph corresponding to the service demander and the second service intent graph corresponding to the service provider. Based on the knowledge graph in the service intent graph, the third implicit intent recognition cost corresponding to the target business intent is calculated, including: statistically analyzing the first path confidence and the second path confidence of each path in the knowledge graph of the service intent graph, wherein the first path confidence represents the frequency of the path appearing in the knowledge graph of the first service intent graph, and the second path confidence represents the frequency of the path appearing in the knowledge graph of the second service intent graph; weighting and summing the first path confidence and the second path confidence to obtain the comprehensive path confidence of each path; determining the seventh preset condition based on the target business intent; and weighting and summing the comprehensive path confidence of the paths that satisfy the seventh preset condition to obtain the third implicit intent recognition cost corresponding to the target business intent.

[0158] The seventh preset condition is determined based on the target business intent, and this application embodiment does not limit it. For example, the seventh preset condition may be that the comprehensive path confidence score is greater than the comprehensive path confidence score threshold, where the comprehensive path confidence score threshold is preset based on intent matching requirements, and this application embodiment does not limit it.

[0159] In one embodiment, a path includes at least one edge. The statistical step of calculating the first path confidence of each path in the knowledge graph of the service intent graph includes: calculating the frequency of a first edge in which any two nodes in the knowledge graph of the service intent graph appear simultaneously in the knowledge graph of the first service intent graph, and calculating the first knowledge scale of the knowledge graph of the first service intent graph; using the ratio of the first edge frequency to the first knowledge scale as the first support of any two nodes; calculating the frequency of the first node in the knowledge graph of the first service intent graph among any two nodes; using the ratio of the first node frequency to the first knowledge scale as the first support of the first node; using the ratio of the first support of any two nodes to the first support of the first node as the first confidence of the edge composed of any two nodes; calculating the first confidence of at least one edge included in the path; and performing a weighted summation of the first confidence of at least one edge included in the path to obtain the first path confidence of the path.

[0160] Here, the first edge frequency is the frequency at which any two nodes in the knowledge graph of the service intent graph appear simultaneously in the knowledge graph of the first service intent graph. The first knowledge scale is the scale of the knowledge graph in the first service intent graph, which is the number of nodes included in the knowledge graph of the first service intent graph. The first support of any two nodes is the support of any two nodes in the knowledge graph of the service intent graph in the knowledge graph of the first service intent graph. The first node frequency is the frequency at which the first node (e.g., A) of any two nodes (e.g., A and B) appears in the knowledge graph of the first service intent graph. The first support of the first node is the support of the first node of any two nodes in the knowledge graph of the service intent graph in the knowledge graph of the first service intent graph.

[0161] The first confidence level of an edge is the confidence level of an edge formed by any two nodes in the knowledge graph of the service intent graph in the knowledge graph of the first service intent graph.

[0162] For example, the terminal counts the first edge frequency of any two nodes appearing simultaneously in the knowledge graph of the first service intent graph, and counts the first knowledge scale of the knowledge graph of the first service intent graph. The first support of these two nodes is then calculated according to the following formula.

[0163]

[0164] In the above formula, It is the first side frequency in which any two nodes in the knowledge graph of the service intent graph appear simultaneously in the knowledge graph of the first service intent graph. It is the first knowledge scale of the knowledge graph in the first service intent graph. It is the first support of any two nodes in the knowledge graph of the service intent graph.

[0165] The frequency of the first node appearing in the knowledge graph of any two nodes in the terminal statistical service intent graph is calculated using the following formula to obtain the first support of the first node.

[0166]

[0167] In the above formula, It is the frequency of the first node appearing in the knowledge graph of the first service intent graph among any two nodes in the knowledge graph of the first service intent graph. It is the first knowledge scale of the knowledge graph in the first service intent graph. It is the first support of the first node of any two nodes in the knowledge graph of the service intent graph.

[0168] The terminal calculates the first confidence level of an edge formed by any two nodes in the knowledge graph of the service intent graph according to the following formula.

[0169]

[0170] In the above formula, It is the first support between any two nodes in the knowledge graph of the service intent graph. It is the first support of the first node of any two nodes in the knowledge graph of the service intent graph. It is the first confidence level of an edge formed by any two nodes in the knowledge graph of the service intent graph.

[0171] The first confidence score of at least one edge included in the terminal statistical path is calculated; the first path confidence score of the path is obtained by weighted summation of the first confidence scores of at least one edge included in the path.

[0172] In this embodiment, by calculating the first support of any two nodes in the knowledge graph of the service intent graph and the first support of the first node among any two nodes, the first confidence of the edge formed by any two nodes is obtained. By weighted summing of the first confidence of at least one edge included in the path, the first path confidence of the path can be calculated.

[0173] In one embodiment, a path includes at least one edge. The statistical step of calculating the second path confidence of each path in the knowledge graph of the service intent graph includes: calculating the frequency of a second edge in which any two nodes in the knowledge graph of the service intent graph appear simultaneously in the knowledge graph of the second service intent graph, and calculating the second knowledge scale of the knowledge graph of the second service intent graph; using the ratio of the second edge frequency to the second knowledge scale as the second support of any two nodes; calculating the frequency of a second node in which the first node of any two nodes appears in the knowledge graph of the second service intent graph; using the ratio of the second node frequency to the second knowledge scale as the second support of the first node; using the ratio of the second support of any two nodes to the second support of the first node as the second confidence of the edge composed of any two nodes; calculating the second confidence of at least one edge included in the path; and performing a weighted summation of the second confidence of at least one edge included in the path to obtain the second path confidence of the path.

[0174] Here, the second edge frequency is the frequency at which any two nodes in the knowledge graph of the service intent graph appear simultaneously in the knowledge graph of the second service intent graph. The second knowledge scale is the scale of the knowledge graph in the second service intent graph, that is, the number of nodes included in the knowledge graph of the second service intent graph. The second support of any two nodes is the support of any two nodes in the knowledge graph of the service intent graph in the knowledge graph of the second service intent graph. The second node frequency is the frequency at which the first node (e.g., A) of any two nodes (e.g., A and B) appears in the knowledge graph of the second service intent graph. The second support of the first node is the support of the first node of any two nodes in the knowledge graph of the service intent graph in the knowledge graph of the second service intent graph.

[0175] The second confidence level of an edge is the confidence level of an edge formed by any two nodes in the knowledge graph of the service intent graph in the knowledge graph of the second service intent graph.

[0176] The confidence level of at least one edge included in the path is obtained by weighted summation of the first confidence level and the second confidence level of the edge.

[0177] like Figure 8As shown, the path "View resource borrowing information -> Apply for resource borrowing product A -> Fill in application resource borrowing information A" includes two edges. The confidence level of the first edge "View resource borrowing information -> Apply for resource borrowing product A" is 100, and the confidence level of the second edge "Apply for resource borrowing product A -> Fill in application resource borrowing information A" is 140. Here, 100 is obtained by weighted sum of the first confidence level and the second confidence level of the first edge, and 140 is obtained by weighted sum of the first confidence level and the second confidence level of the second edge.

[0178] In one embodiment, the fourth implicit intent recognition cost corresponding to the target behavioral intent is calculated based on the knowledge graph in the service intent graph, including: calculating the confidence of at least one edge included in each path, calculating the edge frequency of each edge appearing in the intent graph of the service intent graph, summing the product of the confidence of at least one edge included in each path and the corresponding edge frequency to obtain the association quantity of each path; determining a fourth preset condition based on the target behavioral intent; and weighted summing the association quantities of paths that satisfy the fourth preset condition to obtain the fourth implicit intent recognition cost corresponding to the target behavioral intent.

[0179] The fourth preset condition is determined based on the target's behavioral intent, and this embodiment does not limit this. For example, the fourth preset condition may be that the correlation quantity is greater than the correlation quantity threshold, where the correlation quantity threshold is preset based on intent matching requirements, and this embodiment does not limit this.

[0180] In the service intent graph, intent identification is performed by evaluating the frequency of co-occurrence between intents to obtain the path correlation coefficient (Relation). If the path correlation coefficient (Relation) is greater than the path correlation coefficient threshold (Relation 0), it is considered a valid path, and a set is formed from paths in the intent graph whose correlation coefficient (Relation) is greater than the threshold. This set is the set of legal paths identified from the intent graph.

[0181] Referring to the above embodiments, the confidence level of at least one edge included in the path is obtained by weighted summation of the edge's first confidence level and second confidence level. The edge frequency of each edge appearing in the intent graph of the service intent graph is the frequency of the edge formed by any two nodes in the intent graph of the service intent graph appearing in the intent graph of the service intent graph.

[0182] The affinity of a path is the sum of the affinity of the edges of at least one edge included in the path.

[0183] For example, such as Figure 9As shown, the terminal counts the confidence of at least one edge included in each path, counts the edge frequency of each edge in the intent graph of the service intent graph, and calculates the association quantity of each path according to the following formula.

[0184]

[0185]

[0186] In the above formula, It is the confidence score of each edge included in the path. It is the frequency of each edge appearing in the intent graph of the service intent graph. It is the correlation quantity of each edge included in the path. It refers to the correlation quantity of the path. For example... Figure 9 As shown, the path associations include the associations of three paths, which are:

[0187] Relation path1 = 0.9 * 60 + 0.9 * 60 = 108

[0188] Relation path2 = 0.8*50 + 0.8*50 + 0.8*50 = 120

[0189] Relation path3 = 0.7*30 + 0.6*20 + 0.7*30 + 0.7*30 = 75.

[0190] The terminal is based on the target's behavioral intent ( Figure 9 The target behavioral intent shown is from the explicit intent "resource storage intent" to the implicit intent "purchase non-physical resource product intent". The fourth preset condition is that the path association quantity is greater than the path association quantity threshold (e.g., the path association quantity threshold is 100). According to the fourth preset condition, the set of paths that meet the fourth preset condition is determined as {"resource storage intent -> understanding non-physical resource product intent -> purchase non-physical resource product intent", "resource storage intent -> browsing lifestyle store intent -> payment intent -> purchase non-physical resource product intent"}. The fourth implicit intent recognition cost corresponding to the target behavioral intent is calculated according to the following formula.

[0191]

[0192] In the above formula, It is the association quantity of each path among the n paths that satisfy the fourth preset condition. It is the set of paths in the intent graph of the service intent graph that satisfy the fourth preset condition. These are the weight coefficients of the correlation quantities of each path. They can be preset constants or obtained through training. It is the cost of recognizing the fourth implicit intent corresponding to the target behavior intent, and its value is 92 (i.e. 0.5*108 +0.5*120 = 114).

[0193] In this embodiment, by calculating the association quantity of each path, determining the fourth preset condition based on the target behavior intention, and weighting and summing the association quantities of the paths that satisfy the fourth preset condition, the purpose of calculating the fourth implicit intent recognition cost corresponding to the target behavior intention can be achieved.

[0194] In one embodiment, calculating the fourth implicit intent recognition cost corresponding to the target business intent based on the knowledge graph in the service intent graph includes: calculating the fourth implicit intent recognition cost corresponding to the target business intent based on the knowledge graph in the service intent graph, including: calculating the confidence of at least one edge included in each path, calculating the edge frequency of each edge appearing in the intent graph of the service intent graph, summing the product of the confidence of at least one edge included in each path and the corresponding edge frequency to obtain the association quantity of each path; determining an eighth preset condition based on the target business intent; and weighted summing the association quantities of paths that satisfy the eighth preset condition to obtain the fourth implicit intent recognition cost corresponding to the target business intent.

[0195] The eighth preset condition is determined based on the target business intent, and this embodiment does not limit it. For example, the eighth preset condition may be that the correlation quantity is greater than the correlation quantity threshold, where the correlation quantity threshold is preset based on intent matching requirements, and this embodiment does not limit it.

[0196] In one embodiment, determining the intent matching degree between the target business intent and the target behavioral intent based on the first intent recognition cost and the second intent recognition cost includes: calculating the absolute value of the difference between the first intent recognition cost and the second intent recognition cost; calculating the cost sum of the first intent recognition cost and the second intent recognition cost; and using the ratio of the absolute value of the difference to the cost sum as the intent matching degree between the target business intent and the target behavioral intent.

[0197] For example, the terminal calculates the absolute value of the difference between the first intent recognition cost and the second intent recognition cost according to the following formula.

[0198]

[0199] In the above formula, It is the absolute value of the difference between the cost of recognizing the first intent and the cost of recognizing the second intent. It is the cost of first intent recognition. It is the cost of identifying the second intent.

[0200] The terminal calculates the intent matching degree between the target business intent and the target behavioral intent according to the following formula.

[0201]

[0202] In the above formula, It is the degree of intent matching between the target business intent and the target behavioral intent.

[0203] In this embodiment, the purpose of calculating the intent matching degree between the target business intent and the target behavioral intent can be achieved by using the first intent recognition cost and the second intent recognition cost.

[0204] In one embodiment, such as Figure 10 As shown, an intent matching method is provided, including the following steps:

[0205] S1. Obtain the service provider's business resources (i.e., historical business resources in the above embodiments) and intent resources (i.e., business intent resources in the above embodiments); business resources include business data (i.e., historical business data in the above embodiments), business information (i.e., historical business information in the above embodiments), and business knowledge (i.e., historical business knowledge in the above embodiments), construct a business data graph, a business information graph, a business knowledge graph, and a business intent graph, and construct a DIKP graph corresponding to the service provider (i.e., the first service intent graph in the above embodiments).

[0206] S2. Obtain behavioral resources (i.e., historical behavioral resources in the above embodiments) and intent resources (i.e., behavioral intent resources in the above embodiments) of the financial service demander. Behavioral resources include behavioral data (i.e., historical behavioral data in the above embodiments), behavioral information (i.e., historical behavioral information in the above embodiments), and behavioral knowledge (i.e., historical behavioral knowledge in the above embodiments). Construct a behavioral data graph, a behavioral information graph, a behavioral knowledge graph, and a behavioral intent graph, and construct a DIKP graph corresponding to the service demander (i.e., the second service intent graph in the above embodiments). Integrate the DIKP graph corresponding to the service provider and the DIKP graph corresponding to the service demander to obtain the service intent graph.

[0207] S3. Obtain the interaction intent of the service provider and the demander. The interaction intent includes the target behavioral intent of the service demander and the target business intent of the service provider. Based on the service intent graph, perform intent matching calculation, including: calculating the intent recognition cost COST1 of the target behavioral intent of the service demander in the service intent graph (i.e., the first intent recognition cost in the above embodiment), and calculating the intent recognition cost COST2 of the target business intent of the service provider in the service intent graph (i.e., the second intent recognition cost in the above embodiment); the difference MINUX between COST1 and COST2 is the matching evaluation value, and the intent matching degree MATE is MINUX divided by the sum of COST1 and COST2.

[0208] S4. Circular Reduction of Information Asymmetry: If MATE is greater than the threshold MATE0 and COST1 is greater than COST2, adjust the business resource combination BUSINESS_DIK displayed on the service provider's interactive interface until MATE is less than the threshold MATE0 and MATE is at its minimum value; if MATE is greater than the threshold MATE0 and COST1 is less than COST2, adjust the behavior resource combination ACTION_DIK of the service requester until MATE is less than the threshold MATE0 and MATE is at its minimum value; if MATE is less than the threshold MATE0, continue to try other resource combinations until MATE is at its minimum; if MATE is less than the threshold MATE0, it means that the intent is basically matched, and the information asymmetry (i.e., the asymmetry between the two parties in terms of data, information, knowledge, intent, etc.) is reduced to a controllable level. MATE being at its minimum value means that the intent of the current two parties is the best match, and the information asymmetry is optimally reduced.

[0209] Step S4 further includes: Business resource combination BUSINESS_DIK represents different combinations of business data, business information, and business knowledge, with a total of 3*3*3=27 combination schemes, displayed through an interactive interface; Behavioral resource combination ACTION_DIK represents different combinations of behavioral data, behavioral information, and behavioral knowledge, also with a total of 3*3*3=27 combination schemes, guided through an interactive interface to obtain the adjusted resources; Interactive intents include, but are not limited to, resource storage intents and resource borrowing intents. The process of comparing intent matching degree with intent matching degree threshold is as follows:

[0210]

[0211] In this embodiment, by constructing data graphs, information graphs, and knowledge graphs, management of resources for both service providers and service recipients is achieved. By constructing a service intent graph (DIKP graph), the first intent recognition cost and the second intent recognition cost of the target behavioral intent are calculated based on the service intent graph, thereby realizing the calculation of intent recognition costs and intent matching degrees for both parties in the interaction. Furthermore, intent matching is determined to be complete when the intent matching degree is at its minimum value, thus achieving the goal of accurately matching the intents of both parties in the interaction. The matching of the target behavioral intent of the service provider and the target business intent of the service recipient reduces the asymmetry in data, information, knowledge, intent, and other content between the service provider and the service recipient, and improves the interactive experience of the interface.

[0212] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0213] Based on the same inventive concept, this application also provides an intent matching apparatus for implementing the intent matching method described above. The solution provided by this apparatus is similar to the implementation described in the above method; therefore, the specific limitations in one or more intent matching apparatus embodiments provided below can be found in the limitations of the intent matching method described above, and will not be repeated here.

[0214] In one embodiment, such as Figure 11 As shown, an intent matching device 1100 is provided, including: an acquisition module 1102, a determination module 1104, a calculation module 1106, and an adjustment module 1108, wherein:

[0215] The acquisition module 1102 is used to acquire the target business intent of the service provider and the target behavioral intent of the service demander.

[0216] The determination module 1104 is used to determine the service intent graph based on the service provider's business resource combination, business intent resources, service demander's behavioral resource combination, and behavioral intent resources. The service intent graph includes data graph, information graph, knowledge graph, and intent graph.

[0217] The calculation module 1106 is used to calculate the first intent recognition cost of the target behavioral intent and the second intent recognition cost of the target business intent based on the service intent graph.

[0218] The calculation module 1106 is also used to determine the intent matching degree between the target business intent and the target behavioral intent based on the first intent recognition cost and the second intent recognition cost.

[0219] The adjustment module 1108 is used to adjust at least one of the service provider's business resource combination and the service demander's behavioral resource combination when the intent matching degree is not equal to the intent matching degree threshold, so as to change the service intent graph and reduce the intent matching degree, until the intent matching degree no longer decreases, and then the intent matching is determined to be complete.

[0220] In one embodiment, the adjustment module 1108 is further configured to: when the intent matching degree is greater than the intent matching degree threshold and the first intent recognition cost is greater than the second intent recognition cost, adjust the service provider's business resource combination to change the service intent graph to reduce the intent matching degree until the intent matching degree no longer decreases, and determine that intent matching is complete; when the intent matching degree is greater than the intent matching degree threshold and the first intent recognition cost is less than the second intent recognition cost, adjust the service demander's behavioral resource combination to change the service intent graph to reduce the intent matching degree until the intent matching degree no longer decreases, and determine that intent matching is complete; when the intent matching degree is less than the intent matching degree threshold, adjust at least one of the service provider's business resource combination and the service demander's behavioral resource combination to change the service intent graph to reduce the intent matching degree until the intent matching degree no longer decreases, and determine that intent matching is complete.

[0221] In one embodiment, the determining module 1104 is further configured to: determine a combination of multiple behavioral resources based on the historical behavioral resources of the service demander, wherein the historical behavioral resources include historical behavioral data, historical behavioral information, and historical behavioral knowledge; construct a first service intent graph of the service demander based on the combination of multiple behavioral resources and behavioral intent resources, wherein the behavioral intent resources represent the set of historical behavioral intents of the service demander; determine a combination of multiple business resources based on the historical business resources of the service provider, wherein the historical business resources include historical business data, historical business information, and historical business knowledge; construct a second service intent graph of the service provider based on the combination of multiple business resources and business intent resources, wherein the business intent resources represent the set of historical business intents of the service provider; and integrate the first service intent graph and the second service intent graph to obtain a service intent graph.

[0222] In one embodiment, the calculation module 1106 is further configured to: calculate, based on the intent graph in the service intent graph, the explicit intent recognition cost corresponding to the target behavioral intent and the explicit intent recognition cost corresponding to the target business intent; calculate, based on the data graph in the service intent graph, the first implicit intent recognition cost corresponding to the target behavioral intent and the first implicit intent recognition cost corresponding to the target business intent; calculate, based on the information graph in the service intent graph, the second implicit intent recognition cost corresponding to the target behavioral intent and the second implicit intent recognition cost corresponding to the target business intent; and calculate, based on the knowledge graph in the service intent graph, the third implicit intent recognition cost corresponding to the target behavioral intent and the third implicit intent recognition cost corresponding to the target business intent. The third implicit intent recognition cost corresponding to the graph; based on the knowledge graph in the service intent graph, calculate the fourth implicit intent recognition cost corresponding to the target behavioral intent and the fourth implicit intent recognition cost corresponding to the target business intent; take the sum of the explicit intent recognition cost, the first implicit intent recognition cost, the second implicit intent recognition cost, the third implicit intent recognition cost and the fourth implicit intent recognition cost corresponding to the target behavioral intent as the first intent recognition cost of the target behavioral intent; take the sum of the explicit intent recognition cost, the first implicit intent recognition cost, the second implicit intent recognition cost, the third implicit intent recognition cost and the fourth implicit intent recognition cost corresponding to the target business intent as the second intent recognition cost of the target business intent.

[0223] In one embodiment, the calculation module 1106 is further configured to determine the intent scale of the intent graph in the service intent graph; determine the association set corresponding to the target behavioral intent and the size of the association set in the intent graph in the service intent graph; and use the sum of the intent scale and the size of the association set as the explicit intent recognition cost corresponding to the target behavioral intent.

[0224] In one embodiment, the calculation module 1106 is further configured to: count the first data frequency of each data in the data graph of the service intent graph appearing in the data graph of the first service intent graph, and the second data frequency of each data appearing in the data graph of the second service intent graph; perform a weighted summation of the first data frequency and the second data frequency to obtain the comprehensive data frequency of each data in the data graph of the service intent graph; determine a first preset condition based on the target behavioral intent; and perform a weighted summation of the comprehensive data frequencies of the data that meet the first preset condition to obtain the first implicit intent recognition cost corresponding to the target behavioral intent.

[0225] In one embodiment, the calculation module 1106 is further configured to: count the frequency of each piece of information appearing in the information graph of the first service intent graph and the frequency of each piece of information appearing in the information graph of the second service intent graph; perform a weighted summation of the first and second information frequencies to obtain the comprehensive information frequency of each piece of information appearing in the information graph of the service intent graph; determine a second preset condition based on the target behavioral intent; and perform a weighted summation of the comprehensive information frequencies of the information that satisfy the second preset condition to obtain the second implicit intent recognition cost corresponding to the target behavioral intent.

[0226] In one embodiment, the calculation module 1106 is further configured to: statistically analyze the first path confidence and the second path confidence of each path in the knowledge graph of the service intent graph; wherein the first path confidence represents the frequency of the path appearing in the knowledge graph of the first service intent graph, and the second path confidence represents the frequency of the path appearing in the knowledge graph of the second service intent graph; weightedly sum the first path confidence and the second path confidence to obtain the comprehensive path confidence of each path; determine a third preset condition based on the target behavioral intent; and weightedly sum the comprehensive path confidence of the paths that satisfy the third preset condition to obtain the third implicit intent recognition cost corresponding to the target behavioral intent.

[0227] In one embodiment, the calculation module 1106 is further configured to: count the frequency of a first edge in which any two nodes in the knowledge graph of the service intent graph appear simultaneously in the knowledge graph of the first service intent graph; count the first knowledge scale of the knowledge graph of the first service intent graph; use the ratio of the first edge frequency to the first knowledge scale as the first support of any two nodes; count the frequency of the first node in the knowledge graph of the first service intent graph; use the ratio of the first node frequency to the first knowledge scale as the first support of the first node; use the ratio of the first support of any two nodes to the first support of the first node as the first confidence of the edge composed of any two nodes; count the first confidence of at least one edge included in the path; and perform a weighted summation of the first confidence of at least one edge included in the path to obtain the first path confidence of the path.

[0228] In one embodiment, the calculation module 1106 is further configured to: calculate the confidence level of at least one edge included in each path; calculate the edge frequency of each edge appearing in the intent graph of the service intent graph; sum the products of the confidence level of at least one edge included in each path and the corresponding edge frequency to obtain the association quantity of each path; determine the fourth preset condition according to the target behavior intent; and perform a weighted summation of the association quantities of the paths that satisfy the fourth preset condition to obtain the fourth implicit intent recognition cost corresponding to the target behavior intent.

[0229] In one embodiment, the determining module 1104 is further configured to calculate the absolute value of the difference between the first intent recognition cost and the second intent recognition cost; calculate the cost sum of the first intent recognition cost and the second intent recognition cost; and use the ratio of the absolute value of the difference to the cost sum as the intent matching degree between the target business intent and the target behavioral intent.

[0230] The modules in the aforementioned intent matching device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can invoke and execute the operations corresponding to each module.

[0231] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 12 As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When executed by the processor, the computer program implements an intent matching method. The display unit is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0232] Those skilled in the art will understand that Figure 12 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0233] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0234] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0235] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0236] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0237] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0238] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0239] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. An intent matching method, characterized in that, The method includes: To obtain the target business intent of the service provider and the target behavioral intent of the service demander; Based on the service provider's business resource combination and business intent resources, the service demander's behavioral resource combination and behavioral intent resources, a service intent graph is determined, which includes a data graph, an information graph, a knowledge graph, and an intent graph. Based on the service intent graph, the first intent recognition cost of the target behavioral intent and the second intent recognition cost of the target business intent are calculated respectively. Based on the first intent recognition cost and the second intent recognition cost, the intent matching degree between the target business intent and the target behavioral intent is determined; If the intent matching degree is not equal to the intent matching degree threshold, at least one of the service provider's business resource combination and the service demander's behavioral resource combination is adjusted to change the service intent graph, thereby reducing the intent matching degree, until the intent matching degree no longer decreases, at which point intent matching is determined to be complete. The step of calculating the first intent recognition cost of the target behavioral intent and the second intent recognition cost of the target business intent based on the service intent graph includes: Based on the intent graph in the service intent graph, calculate the explicit intent recognition cost corresponding to the target behavioral intent and the explicit intent recognition cost corresponding to the target business intent; Based on the data graph in the service intent graph, calculate the first implicit intent recognition cost corresponding to the target behavioral intent and the first implicit intent recognition cost corresponding to the target business intent; Based on the information graph in the service intent graph, calculate the second implicit intent recognition cost corresponding to the target behavioral intent and the second implicit intent recognition cost corresponding to the target business intent; Based on the knowledge graph in the service intent graph, calculate the third implicit intent recognition cost corresponding to the target behavioral intent and the third implicit intent recognition cost corresponding to the target business intent; Based on the knowledge graph in the service intent graph, calculate the fourth implicit intent recognition cost corresponding to the target behavioral intent and the fourth implicit intent recognition cost corresponding to the target business intent; The sum of the explicit intent recognition cost, the first implicit intent recognition cost, the second implicit intent recognition cost, the third implicit intent recognition cost, and the fourth implicit intent recognition cost corresponding to the target behavior intent is taken as the first intent recognition cost of the target behavior intent. The sum of the explicit intent recognition cost, the first implicit intent recognition cost, the second implicit intent recognition cost, the third implicit intent recognition cost, and the fourth implicit intent recognition cost corresponding to the target business intent is taken as the second intent recognition cost of the target business intent.

2. The method according to claim 1, characterized in that, When the intent matching degree is not equal to the intent matching degree threshold, adjusting at least one of the service provider's business resource combination and the service demander's behavioral resource combination to change the service intent graph and reduce the intent matching degree, until the intent matching degree no longer decreases, and then determining that intent matching is complete, includes: If the intent matching degree is greater than the intent matching degree threshold and the first intent recognition cost is greater than the second intent recognition cost, the service provider's business resource combination is adjusted to change the service intent graph to reduce the intent matching degree until the intent matching degree no longer decreases, and the intent matching is determined to be complete. If the intent matching degree is greater than the intent matching degree threshold and the first intent recognition cost is less than the second intent recognition cost, the combination of behavioral resources of the service demander is adjusted to change the service intent graph to reduce the intent matching degree until the intent matching degree no longer decreases, and the intent matching is determined to be complete. If the intent matching degree is less than the intent matching degree threshold, at least one of the service provider's business resource combination and the service demander's behavioral resource combination is adjusted to change the service intent graph, thereby reducing the intent matching degree, until the intent matching degree no longer decreases, and the intent matching is determined to be complete.

3. The method according to claim 1, characterized in that, The process of determining the service intent graph based on the service provider's business resource combination, business intent resources, the service demander's behavioral resource combination, and behavioral intent resources includes: Based on the historical behavioral resources of the service demanders, a combination of various behavioral resources is determined, wherein the historical behavioral resources include historical behavioral data, historical behavioral information, and historical behavioral knowledge; Based on the combination of various behavioral resources and behavioral intent resources, a first service intent graph of the service demander is constructed, wherein the behavioral intent resources represent the set of historical behavioral intents of the service demander. Based on the historical business resources of the service provider, a variety of business resource combinations are determined, wherein the historical business resources include historical business data, historical business information, and historical business knowledge; Based on the combination of various business resources and business intent resources, a second service intent graph of the service provider is constructed, wherein the business intent resources represent the set of historical business intents of the service provider; Integrate the first service intent graph and the second service intent graph to obtain a service intent graph.

4. The method according to claim 1, characterized in that, The step of calculating the explicit intent recognition cost corresponding to the target behavioral intent based on the intent graph in the service intent graph includes: Determine the intent scale of the intent graph in the service intent graph; Determine the set of associations in the intent graph of the service intent graph that corresponds to the target behavioral intent, and the size of the set of associations; The sum of the intent size and the associated set size is used as the explicit intent recognition cost corresponding to the target behavioral intent.

5. The method according to claim 1, characterized in that, The service intent graph is obtained by integrating the first service intent graph corresponding to the service demander and the second service intent graph corresponding to the service provider. The step of calculating the first implicit intent recognition cost corresponding to the target behavioral intent based on the data graph in the service intent graph includes: The frequency of each data point in the data graph of the service intent graph is firstly observed in the data graph of the first service intent graph, and secondly observed in the data graph of the second service intent graph. The first data frequency and the second data frequency are weighted and summed to obtain the comprehensive data frequency of each data in the data map of the service intent map; Based on the stated target behavioral intent, a first preset condition is determined; The comprehensive data frequencies of data that meet the first preset conditions are weighted and summed to obtain the first implicit intent recognition cost corresponding to the target behavioral intent.

6. The method according to claim 1, characterized in that, The service intent graph is obtained by integrating the first service intent graph corresponding to the service demander and the second service intent graph corresponding to the service provider. The step of calculating the second implicit intent recognition cost corresponding to the target behavioral intent based on the information graph in the service intent graph includes: The frequency of each piece of information in the information graph of the service intent graph is calculated as the first information frequency in the information graph of the first service intent graph, and the second information frequency in the information graph of the second service intent graph. The first information frequency and the second information frequency are weighted and summed to obtain the comprehensive information frequency of each information in the information graph of the service intent graph; Based on the stated target behavioral intent, determine the second preset condition; The comprehensive information frequencies that satisfy the second preset condition are weighted and summed to obtain the second implicit intent recognition cost corresponding to the target behavioral intent.

7. The method according to claim 1, characterized in that, The service intent graph is obtained by integrating the first service intent graph corresponding to the service demander and the second service intent graph corresponding to the service provider. The step of calculating the third implicit intent recognition cost corresponding to the target behavioral intent based on the knowledge graph in the service intent graph includes: The confidence scores of the first path and the confidence scores of the second path are calculated for each path in the knowledge graph of the service intent graph. The first path confidence score represents the frequency of the path appearing in the knowledge graph of the first service intent graph, and the second path confidence score represents the frequency of the path appearing in the knowledge graph of the second service intent graph. The confidence scores of the first path and the second path are weighted and summed to obtain the comprehensive path confidence score of each path. Based on the target behavioral intent, a third preset condition is determined; the comprehensive path confidence of the paths that satisfy the third preset condition is weighted and summed to obtain the third implicit intent recognition cost corresponding to the target behavioral intent.

8. The method according to claim 7, characterized in that, The path includes at least one edge, and the statistical steps for calculating the first path confidence of each path in the knowledge graph of the service intent graph include: The frequency of the first side occurrence of any two nodes in the knowledge graph of the service intent graph is calculated, and the first knowledge scale of the knowledge graph of the first service intent graph is calculated. The ratio of the first edge frequency to the first knowledge scale is used as the first support of any two nodes. Calculate the frequency of the first node of any two nodes appearing in the knowledge graph of the first service intent graph; The ratio of the frequency of the first node to the size of the first knowledge is used as the first support of the first node. The ratio of the first support of any two nodes to the first support of the first node is used as the first confidence of the edge formed by any two nodes. The first confidence level of at least one edge included in the statistical path; The first path confidence is obtained by weighted summing of the first confidence scores of at least one edge included in the path.

9. The method according to claim 1, characterized in that, The step of calculating the fourth implicit intent recognition cost corresponding to the target behavioral intent based on the knowledge graph in the service intent graph includes: Calculate the confidence score of at least one edge included in each path, and calculate the edge frequency of each edge appearing in the intent graph of the service intent graph. The correlation coefficient of each path is obtained by summing the product of the confidence level of at least one edge included in each path and the corresponding edge frequency. Based on the stated target behavioral intent, a fourth preset condition is determined; The associated quantities of the paths that satisfy the fourth preset condition are weighted and summed to obtain the fourth implicit intent recognition cost corresponding to the target behavioral intent.

10. The method according to any one of claims 1 to 9, characterized in that, The step of determining the intent matching degree between the target business intent and the target behavioral intent based on the first intent recognition cost and the second intent recognition cost includes: Calculate the absolute value of the difference between the first intent recognition cost and the second intent recognition cost; Calculate the sum of the costs of the first intent recognition cost and the second intent recognition cost; The ratio of the absolute value of the difference to the sum of the costs is taken as the degree of intent matching between the target business intent and the target behavioral intent.

11. An intent matching device, characterized in that, The device includes: The acquisition module is used to acquire the target business intent of the service provider and the target behavioral intent of the service demander. The determination module is used to determine a service intent graph based on the service provider's business resource combination, business intent resources, the service demander's behavioral resource combination, and behavioral intent resources. The service intent graph includes a data graph, an information graph, a knowledge graph, and an intent graph. The calculation module is used to calculate the first intent recognition cost of the target behavioral intent and the second intent recognition cost of the target business intent based on the service intent graph. The calculation module is further configured to determine the intent matching degree between the target business intent and the target behavioral intent based on the first intent recognition cost and the second intent recognition cost; The adjustment module is used to adjust at least one of the service provider's business resource combination and the service demander's behavioral resource combination when the intent matching degree is not equal to the intent matching degree threshold, so as to change the service intent graph and reduce the intent matching degree, until the intent matching degree no longer decreases and the intent matching is determined to be completed. The calculation module is further configured to: calculate the explicit intent recognition cost corresponding to the target behavioral intent and the explicit intent recognition cost corresponding to the target business intent based on the intent graph in the service intent graph; calculate the first implicit intent recognition cost corresponding to the target behavioral intent and the first implicit intent recognition cost corresponding to the target business intent based on the data graph in the service intent graph; calculate the second implicit intent recognition cost corresponding to the target behavioral intent and the second implicit intent recognition cost corresponding to the target business intent based on the information graph in the service intent graph; and calculate the third implicit intent recognition cost corresponding to the target behavioral intent and the second implicit intent recognition cost corresponding to the target business intent based on the knowledge graph in the service intent graph. The cost of identifying three implicit intentions is calculated. Based on the knowledge graph in the service intent graph, the cost of identifying the fourth implicit intention corresponding to the target behavioral intention and the cost of identifying the fourth implicit intention corresponding to the target business intention are calculated. The sum of the cost of identifying the explicit intention, the cost of identifying the first implicit intention, the cost of identifying the second implicit intention, the cost of identifying the third implicit intention, and the cost of identifying the fourth implicit intention corresponding to the target behavioral intention is taken as the first intention identification cost of the target behavioral intention. The sum of the cost of identifying the explicit intention, the cost of identifying the first implicit intention, the cost of identifying the second implicit intention, the cost of identifying the third implicit intention, and the cost of identifying the fourth implicit intention corresponding to the target business intention is taken as the second intention identification cost of the target business intention.

12. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 10.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 10.

14. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 10.

Images