Artificial intelligence-based service recommendation method, apparatus, device, and storage medium
By constructing a heterogeneous graph of user goals and using a heterogeneous graph neural network model for service recommendation, the problem of poor accuracy in customer service recommendations is solved, recommendation efficiency and accuracy are improved, and customer loyalty is enhanced.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINA PING AN PROPERTY INSURANCE CO LTD
- Filing Date
- 2026-03-06
- Publication Date
- 2026-06-05
AI Technical Summary
In existing technologies, customer service recommendations are not very accurate, resulting in low customer activity and satisfaction.
By acquiring multidimensional user data, a target heterogeneous graph is constructed, and a pre-trained heterogeneous graph neural network model is used for service recommendation to generate a list of recommended services.
This improved the accuracy and efficiency of service recommendations, and enhanced customer recognition and reliance on the brand.
Smart Images
Figure CN122153164A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of artificial intelligence technology, and in particular to a service recommendation method, apparatus, device, and storage medium based on artificial intelligence. Background Technology
[0002] With the diversification of market demands, the complexity of modern business scenarios has significantly increased. The financial industry is currently shifting from a single-customer-purchase-of-wealth-product-oriented approach to a customer-operation-oriented one. This can improve customer brand recognition and reliance, while also increasing the number of customer service orders and thus boosting profits. For example, in the auto insurance industry, after purchasing auto insurance, insurance companies can also offer services such as car washing, car maintenance, roadside assistance, and vehicle inspection processing, thereby enhancing customer recognition and loyalty. However, the accuracy of current service recommendations is poor, resulting in difficulty stimulating customer activity and low customer satisfaction.
[0003] Therefore, how to accurately recommend services to customers is a problem that urgently needs to be solved. Summary of the Invention
[0004] The main objective of this application is to provide a service recommendation method, apparatus, device, and storage medium based on artificial intelligence, aiming to improve the efficiency and accuracy of service recommendations.
[0005] Firstly, this application provides an artificial intelligence-based service recommendation method, which includes the following steps: Obtain multidimensional user data of the users to be recommended for service, including behavioral data, sentiment data, and usage scenario data; Based on the behavioral data, the emotional data, and the usage scenario data, construct the user's target heterogeneity graph; A service recommendation model is obtained, wherein the service recommendation model is a heterogeneous graph neural network model pre-trained based on multiple sample data; The target heterogeneous graph is identified and services are predicted using the service recommendation model to obtain a list of recommended services, which includes multiple recommended services.
[0006] Secondly, this application also provides a service recommendation device, which includes an acquisition module, a construction module, and a generation module, wherein: The acquisition module is used to acquire multi-dimensional user data of the users to be recommended for service, including behavioral data, sentiment data, and usage scenario data. The construction module is used to construct the user's target heterogeneous graph based on the behavioral data, the emotional data, and the usage scenario data. The acquisition module is also used to acquire a service recommendation model, which is a heterogeneous graph neural network model pre-trained based on multiple sample data. The generation module is used to identify and predict services in the target heterogeneous graph through the service recommendation model to obtain a list of recommended services, which includes multiple recommended services.
[0007] Thirdly, this application also provides a computer device, the computer device including a processor, a memory, and a computer program stored in the memory and executable by the processor, wherein when the computer program is executed by the processor, it implements the steps of the artificial intelligence-based service recommendation method described above.
[0008] Fourthly, this application also provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, it implements the steps of the artificial intelligence-based service recommendation method described above.
[0009] This application provides a service recommendation method, apparatus, device, and storage medium based on artificial intelligence. The application acquires multidimensional user data of the user to be recommended, including behavioral data, sentiment data, and usage scenario data. Based on the behavioral data, sentiment data, and usage scenario data, a target heterogeneous graph of the user is constructed. A service recommendation model is obtained, which is a heterogeneous graph neural network model pre-trained based on multiple sample data. The service recommendation model identifies and predicts services from the target heterogeneous graph to obtain a list of recommended services, which includes multiple recommended services. By constructing a target heterogeneous graph of the user and identifying and predicting services from the target heterogeneous graph using a service recommendation model, this application can accurately obtain a list of recommended services, thereby greatly improving the efficiency and accuracy of service recommendation. Attached Figure Description
[0010] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0011] Figure 1 A flowchart illustrating an artificial intelligence-based service recommendation method provided in this application embodiment; Figure 2A flowchart illustrating another AI-based service recommendation method provided in this application embodiment; Figure 3 for Figure 1 A flowchart illustrating the sub-steps of an AI-based service recommendation method. Figure 4 A schematic block diagram of a service recommendation device provided in an embodiment of this application; Figure 5 for Figure 4 A schematic block diagram of a submodule of the service recommendation device in the system; Figure 6 This is a schematic block diagram of the structure of a computer device provided in an embodiment of this application.
[0012] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation
[0013] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0014] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.
[0015] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.
[0016] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.
[0017] With the diversification of market demands, the complexity of modern business scenarios has significantly increased. The financial industry is currently shifting from a single-customer-purchase-of-wealth-product-oriented approach to a customer-operation-oriented one. This can improve customer brand recognition and reliance, while also increasing the number of customer service orders and thus boosting profits. For example, in the auto insurance industry, after purchasing auto insurance, insurance companies can also offer services such as car washing, car maintenance, roadside assistance, and vehicle inspection processing, thereby enhancing customer recognition and loyalty. However, the accuracy of current service recommendations is poor, resulting in difficulty stimulating customer activity and low customer satisfaction.
[0018] To address the aforementioned problems, this application provides an artificial intelligence-based service recommendation method, apparatus, device, and storage medium. The AI-based service recommendation method involves: acquiring multi-dimensional user data of a user to be recommended services, including behavioral data, sentiment data, and usage scenario data; constructing a target heterogeneous graph of the user based on the behavioral data, sentiment data, and usage scenario data; acquiring a service recommendation model, which is a heterogeneous graph neural network model pre-trained based on multiple sample data; and using the service recommendation model to identify and predict services on the target heterogeneous graph to obtain a list of recommended services, which includes multiple recommended services.
[0019] This AI-based service recommendation method can be applied to computer devices, such as mobile phones, tablets, laptops, desktop computers, personal digital assistants, and wearable devices.
[0020] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.
[0021] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating an artificial intelligence-based service recommendation method provided as an embodiment of this application.
[0022] like Figure 1 As shown, the AI-based service recommendation method includes steps S101 to S104.
[0023] Step S101: Obtain multi-dimensional user data of the users to be recommended for service, including behavioral data, sentiment data, and usage scenario data.
[0024] The user's multidimensional data includes behavioral data, emotional data, and usage scenario data. The behavioral data refers to the actual actions of the user, the emotional data includes user emotions, evaluations, and social sentiment data, and the usage scenario data refers to the scenarios in which the user uses the service.
[0025] For example, in the auto insurance service industry, behavioral data includes user insurance purchase behavior, claims behavior, renewal behavior, service call behavior, and operation of insurance application behavior, etc.; sentiment data includes user customer service voice messages, user evaluations of auto insurance services, and public opinion, etc.; usage scenario data includes vehicle accident time and location, vehicle usage area, and vehicle usage frequency, etc.
[0026] It should be noted that the user's multidimensional data is the data generated by the user performing basic services. These basic services are the basic services that the user has engaged in, such as vehicle insurance. The recommended services are value-added services derived from vehicle insurance, including but not limited to car washing, car maintenance, roadside assistance, and vehicle annual inspection services.
[0027] In some embodiments, data generated by users during the process of obtaining basic services is acquired to obtain multidimensional user data. This multidimensional user data includes behavioral data, sentiment data, and usage scenario data. By acquiring data generated by users during the process of obtaining basic services, multidimensional user data can be accurately obtained.
[0028] In some embodiments, obtaining multidimensional user data by acquiring data generated by users during the process of handling basic services can be achieved by: acquiring user data from the data generated during the process of handling basic services; classifying and cleaning the user data to obtain multidimensional user data including behavioral data, sentiment data, and usage scenario data. The data cleaning methods include, but are not limited to, abnormal data removal and abnormal data modification. The data classification method includes dimensional data classification.
[0029] Step S102: Construct the user's target heterogeneous graph based on the behavioral data, the emotional data, and the usage scenario data.
[0030] The target heterogeneous graph is a heterogeneous graph of multiple types of nodes constructed based on multidimensional user data.
[0031] In some embodiments, a behavioral heterogeneous graph is obtained by associating user and behavioral data through a behavioral heterogeneous structure; an emotional heterogeneous graph is obtained by associating user and sentiment data through an emotional heterogeneous structure; a scene heterogeneous graph is obtained by associating user and usage scenario data through a scene heterogeneous structure; and a target heterogeneous graph is obtained by associating the behavioral, emotional, and scene heterogeneous graphs through heterogeneous graph relationships. By constructing heterogeneous nodes from behavioral, sentiment, and usage scenario data, the target heterogeneous graph can be accurately obtained.
[0032] In some embodiments, the way to obtain a behavior heterogeneous graph by associating user and behavior data with a behavior heterogeneous structure can be as follows: treat a user as a heterogeneous user node, treat each behavior data as a behavior heterogeneous node, connect the heterogeneous user node and each behavior heterogeneous node respectively to obtain behavior heterogeneous edges between the heterogeneous user node and each behavior heterogeneous node; associate each behavior heterogeneous node with the behavior heterogeneous edge to obtain the behavior heterogeneous graph.
[0033] In some embodiments, the way to obtain an emotional heterogeneous graph by associating users and emotional data with an emotional heterogeneous structure can be as follows: treat each user as a heterogeneous user node and each piece of emotional data as an emotional heterogeneous node, connect the heterogeneous user node and each emotional heterogeneous node respectively to obtain the emotional heterogeneous edge between the heterogeneous user node and each emotional heterogeneous node; associate each emotional heterogeneous node and emotional heterogeneous edge to obtain the emotional heterogeneous graph.
[0034] In some embodiments, the method of associating the user and the usage scenario data with a heterogeneous scenario structure to obtain a heterogeneous scenario graph can be as follows: treating the user as a heterogeneous user node and each scenario data as a heterogeneous scenario node, connecting the heterogeneous user node and each heterogeneous scenario node respectively to obtain heterogeneous scenario edges between the heterogeneous user node and each heterogeneous scenario node; associating each heterogeneous scenario node and the heterogeneous scenario edge to obtain a heterogeneous scenario graph.
[0035] In some embodiments, the target heterogeneous graph can be obtained by associating heterogeneous graph relationships among the behavioral heterogeneous graph, the emotional heterogeneous graph, and the scene heterogeneous graph. This can be achieved by connecting the associated nodes in the behavioral heterogeneous graph, the emotional heterogeneous graph, and the scene heterogeneous graph.
[0036] Step S103: Obtain the service recommendation model, which is a heterogeneous graph neural network model pre-trained based on multiple sample data.
[0037] The service recommendation model is a heterogeneous graph neural network model that has been pre-trained based on multiple sample data.
[0038] In some embodiments, such as Figure 2As shown, the AI-based service recommendation method also includes steps S201 to S205.
[0039] Step S201: Obtain a sample dataset, which includes multiple sample data, including a sample heterogeneity graph and a list of labeled recommendation services.
[0040] The sample data includes a sample heterogeneity graph and a labeled list of recommended services. The sample heterogeneity graph is constructed based on multidimensional sample user data.
[0041] In some embodiments, historical data is acquired, including user multidimensional data and a list of recommended services manually configured for the user; a heterogeneous graph of the user is constructed from the behavioral data, sentiment data, and usage scenario data in the user multidimensional data, and the heterogeneous graph is labeled as a sample heterogeneous graph, and the list of recommended services is labeled as an annotated list of recommended services, thus obtaining a sample data set; the aforementioned steps are repeated to obtain a sample dataset.
[0042] Step S202: Obtain a preset heterogeneous graph neural network model, and select a sample data from the sample dataset as the target sample data.
[0043] Obtain an untrained heterogeneous graph neural network model, and randomly select a sample data from the sample dataset as the target sample data. The target sample data includes the sample heterogeneous graph and the labeled list of recommendation services.
[0044] Step S203: Identify and predict the sample heterogeneous graphs in the target sample data using the preset heterogeneous graph neural network model to obtain a predicted list of recommended services.
[0045] The pre-defined heterogeneous graph neural network model includes an input layer, a feature fusion layer, and an output layer. The input layer is used to extract features from the heterogeneous graph, the feature fusion layer is used to fuse the sentiment feature vector and the service relationship feature vector, and the output layer is used to predict the recommendation service based on the target feature vector.
[0046] In some embodiments, the input layer extracts features from the heterogeneous graph of the samples to obtain a predicted relation feature vector, which includes a predicted user sentiment feature vector and a predicted service relationship feature vector. The feature fusion layer fuses the predicted user sentiment feature vector and the predicted service relationship feature vector to obtain a predicted target feature vector. The output layer performs recommendation service prediction on the predicted target feature vector to obtain a predicted list of recommended services.
[0047] In some embodiments, the method of extracting features from the sample heterogeneous graph through the input layer to obtain the predicted relationship feature vector can be as follows: extracting features from the sentiment nodes in the sample heterogeneous graph through the input layer to obtain the predicted user sentiment feature vector; and extracting features from the service nodes in the sample heterogeneous graph through the input layer to obtain the predicted service relationship feature vector.
[0048] In some embodiments, the method of fusing the predicted user sentiment feature vector and the predicted service relationship feature vector through the feature fusion layer to obtain the predicted target feature vector can be as follows: obtaining the first weight parameter corresponding to the user sentiment feature vector and the second weight parameter corresponding to the service relationship feature vector; performing feature vector weighting processing based on the first weight parameter, the second weight parameter, the predicted user sentiment feature vector and the predicted service relationship feature vector to obtain the target feature vector.
[0049] In some embodiments, the method of obtaining a predicted list of recommended services by predicting recommended services on the predicted target feature vector through the output layer can be as follows: predicting service preferences and sentiment on the predicted target feature vector to obtain a predicted recommendation score for each service; sorting the services according to their predicted recommendation scores to generate a predicted list of recommended services, wherein the list of recommended services includes multiple recommended services.
[0050] Step S204: Based on the predicted list of recommended services and the labeled list of recommended services in the target sample data, determine whether the preset heterogeneous graph neural network model has converged. Based on the predicted list of recommended services and the labeled list of recommended services in the target sample data, the loss value of the preset heterogeneous graph neural network model is determined. If the loss value of the preset heterogeneous graph neural network model is less than or equal to the preset loss value, the preset heterogeneous graph neural network model is determined to have converged; if the loss value of the preset heterogeneous graph neural network model is greater than the preset loss value, the preset heterogeneous graph neural network model is determined to have not converged. The preset loss value can be set according to actual conditions, and this embodiment does not impose specific limitations on it. For example, the preset loss value can be set to 0.02.
[0051] In some embodiments, the method for determining the loss value of the preset heterogeneous graph neural network model based on the predicted list of recommended services and the labeled list of recommended services in the target sample data can be as follows: calculate the similarity between the predicted list of recommended services and the labeled list of recommended services to obtain the current similarity value; subtract the current similarity value from the unit value to determine the current loss value; obtain historical loss values, and calculate the average of the current loss value and the historical loss value to obtain the loss value.
[0052] Step S205: If the preset heterogeneous graph neural network model does not converge, the model parameters of the preset heterogeneous graph neural network model are modified, and the step of selecting a sample data from the sample dataset as the target sample data continues to be executed until a converged service recommendation model is obtained.
[0053] If the loss value of the preset heterogeneous graph neural network model is greater than the preset loss value, it is determined that the preset heterogeneous graph neural network model has not converged. Then, the model parameters of the preset heterogeneous graph neural network model are modified, and the process continues to select a sample data from the sample dataset as the target sample data. The preset heterogeneous graph neural network model is used to identify the sample heterogeneous graph in the target sample data and predict services to obtain the predicted list of recommended services. Based on the predicted list of recommended services and the labeled list of recommended services in the target sample data, it is determined whether the preset heterogeneous graph neural network model has converged. This process continues until a converged service recommendation model is obtained.
[0054] It should be noted that during the training of the service recommendation model, the target services selected by the user based on the recommended service list are obtained; based on the target services, the user's activity index is determined, and the model parameters of the service recommendation model are updated according to the activity index. By obtaining the target services selected by the user from the recommended service list, and then determining the user's activity index based on the target services and updating the model parameters of the service recommendation model, the accuracy of the model can be effectively improved.
[0055] Step S104: Identify and predict services for the target heterogeneous graph using the service recommendation model to obtain a list of recommended services, which includes multiple recommended services.
[0056] The service recommendation model consists of an input layer, a feature fusion layer, and an output layer.
[0057] In some embodiments, such as Figure 3 As shown, step S104 includes sub-steps S1041 to S1043.
[0058] Sub-step S1041: Extract features from the target heterogeneous graph through the input layer to obtain a relation feature vector, which includes a user sentiment feature vector and a service relationship feature vector.
[0059] In some embodiments, feature extraction is performed on the sentiment nodes in the target heterogeneous graph through the input layer to obtain the user sentiment feature vector; feature extraction is also performed on the service nodes in the target heterogeneous graph through the input layer to obtain the service relationship feature vector. Feature extraction through the target heterogeneous graph can accurately obtain the user sentiment feature vector and the service relationship feature vector.
[0060] Sub-step S1042: The user emotion feature vector and the service relationship feature vector are fused through the feature fusion layer to obtain the target feature vector.
[0061] In some embodiments, a first weight parameter corresponding to the user emotion feature vector and a second weight parameter corresponding to the service relationship feature vector are obtained; the feature vectors are weighted according to the first weight parameter, the second weight parameter, the user emotion feature vector, and the service relationship feature vector to obtain the target feature vector. The first and second weight parameters can be set according to actual conditions, and this embodiment does not impose specific limitations on them. For example, the first weight parameter can be set to 0.7, and the second weight parameter can be set to 0.3.
[0062] Sub-step S1043: Predict recommendation services for the target feature vector through the output layer to obtain the recommendation service list.
[0063] In some embodiments, service preferences and sentiment predictions are performed on the target feature vector to obtain a recommendation score for each service. The services are then ranked according to their recommendation scores to generate a list of recommended services, which includes multiple recommended services. By predicting services from the target feature vector, recommendation scores can be accurately obtained. Then, by ranking multiple services based on their recommendation scores, a list of recommended services can be accurately generated, significantly improving the accuracy of service recommendations.
[0064] In some embodiments, the target service selected by the user based on the recommended service list is obtained; based on the target service, the user's activity index is determined, and the model parameters of the service recommendation model are updated according to the activity index. By increasing the activity level of the user's selected services, the accuracy of service recommendations can be effectively improved.
[0065] For example, the system acquires multidimensional user data related to car insurance services, including car insurance behavior data, car insurance sentiment data, and car insurance usage scenario data. Based on these data, a user target heterogeneity graph is constructed. A service recommendation model is then obtained. The service recommendation model is used to identify and predict services within the target heterogeneity graph, resulting in a recommendation score of 90 for car washing, 60 for car maintenance, 70 for roadside assistance, and 30 for vehicle inspection services. Each service is then ranked according to its recommendation score to generate a recommended service list, which includes car washing, roadside assistance, car maintenance, and vehicle inspection services.
[0066] The AI-based service recommendation method provided in the above embodiments obtains multi-dimensional user data of the user to be recommended, including behavioral data, sentiment data, and usage scenario data; constructs a target heterogeneous graph of the user based on the behavioral data, sentiment data, and usage scenario data; obtains a service recommendation model, which is a heterogeneous graph neural network model pre-trained based on multiple sample data; and identifies and predicts services using the service recommendation model to obtain a list of recommended services, which includes multiple recommended services. By constructing a target heterogeneous graph of the user and identifying and predicting services using the service recommendation model, this application can accurately obtain a list of recommended services, thereby greatly improving the efficiency and accuracy of service recommendation.
[0067] Please see Figure 4 , Figure 4 A schematic block diagram of a service recommendation device provided in an embodiment of this application; like Figure 4 As shown, the service recommendation device 300 includes an acquisition module 310, a construction module 320, and a generation module 330, wherein: The acquisition module 310 is used to acquire multi-dimensional user data of the users to be recommended for service, including behavioral data, emotional data and usage scenario data. The construction module 320 is used to construct the user's target heterogeneous graph based on the behavioral data, the emotional data, and the usage scenario data. The acquisition module 310 is also used to acquire a service recommendation model, which is a heterogeneous graph neural network model pre-trained based on multiple sample data. The generation module 330 is used to identify and predict services in the target heterogeneous graph through the service recommendation model to obtain a list of recommended services, which includes multiple recommended services.
[0068] In some embodiments, such as Figure 5 As shown, the generation module 330 includes a first generation submodule 331, a second generation submodule 332, and a third generation submodule 333, wherein: The first generation submodule 331 is used to extract features from the target heterogeneous graph through the input layer to obtain a relation feature vector, wherein the relation feature vector includes a user sentiment feature vector and a service relation feature vector. The second generation submodule 332 is used to perform feature fusion on the user sentiment feature vector and the service relationship feature vector through the feature fusion layer to obtain the target feature vector; The third generation submodule 333 is used to predict recommendation services for the target feature vector through the output layer to obtain the recommendation service list.
[0069] In some embodiments, the first generation submodule 331 is further configured to: The input layer extracts features from the sentiment nodes in the target heterogeneous graph to obtain the user sentiment feature vector. The service nodes in the target heterogeneous graph are extracted through the input layer to obtain service relationship feature vectors.
[0070] In some embodiments, the second generation submodule 332 is further configured to: Obtain the first weight parameter corresponding to the user sentiment feature vector and the second weight parameter corresponding to the service relationship feature vector; The target feature vector is obtained by weighting the feature vectors based on the first weight parameter, the second weight parameter, the user sentiment feature vector, and the service relationship feature vector.
[0071] In some embodiments, the third generation submodule 333 is further configured to: Service preference and sentiment prediction are performed on the target feature vector to obtain a recommendation score for each service; The services are sorted according to their recommendation scores to generate a list of recommended services, which includes multiple recommended services.
[0072] In some embodiments, the construction module 320 is further configured to: The user and the behavior data are correlated to obtain a behavior heterogeneity graph. The user and the sentiment data are correlated using a heterogeneous sentiment structure to obtain a heterogeneous sentiment graph. The user and usage scenario data are correlated to obtain a scenario heterogeneity graph. The target heterogeneous graph is obtained by performing heterogeneous graph relationship association on the behavioral heterogeneous graph, the emotional heterogeneous graph, and the scene heterogeneous graph.
[0073] In some embodiments, the service recommendation device 300 is further configured to: Obtain the target service selected by the user based on the recommended service list; Based on the target service, determine the user's activity index, and update the model parameters of the service recommendation model based on the activity index.
[0074] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the above-mentioned service recommendation device can be referred to the corresponding process in the aforementioned embodiment of the AI-based service recommendation method, and will not be repeated here.
[0075] Please see Figure 6 , Figure 6 This is a schematic block diagram of the structure of a computer device provided in an embodiment of this application.
[0076] like Figure 6 As shown, the computer device 400 includes a processor 402 and a memory 403 connected via a system bus 401, wherein the memory 402 may include a storage medium and internal memory.
[0077] The storage medium may store a computer program. The computer program includes program instructions that, when executed, cause the processor 402 to perform any artificial intelligence-based service recommendation method.
[0078] Processor 402 provides computing and control capabilities to support the operation of the entire computer device.
[0079] Internal memory provides an environment for the execution of computer programs stored in storage media. When these computer programs are executed by a processor, the processor can perform any service recommendation method based on artificial intelligence.
[0080] Those skilled in the art will understand that Figure 6 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.
[0081] It should be understood that processor 402 can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.
[0082] In one embodiment, the processor 402 is configured to run a computer program stored in a memory to perform the following steps: Obtain multidimensional user data of the users to be recommended for service, including behavioral data, sentiment data, and usage scenario data; Based on the behavioral data, the emotional data, and the usage scenario data, construct the user's target heterogeneity graph; A service recommendation model is obtained, wherein the service recommendation model is a heterogeneous graph neural network model pre-trained based on multiple sample data; The target heterogeneous graph is identified and services are predicted using the service recommendation model to obtain a list of recommended services, which includes multiple recommended services.
[0083] In one embodiment, the service recommendation model includes an input layer, a feature fusion layer, and an output layer; the processor 402, when implementing the process of identifying and predicting services from the target heterogeneous graph using the service recommendation model to obtain a list of recommended services, is configured to: The input layer extracts features from the target heterogeneous graph to obtain a relation feature vector, which includes a user sentiment feature vector and a service relationship feature vector. The target feature vector is obtained by fusing the user sentiment feature vector and the service relationship feature vector through the feature fusion layer. The output layer is used to predict recommendation services for the target feature vector to obtain the recommendation service list.
[0084] In one embodiment, when the processor 402 performs feature extraction on the target heterogeneous graph through the input layer to obtain a relational feature vector, it is used to implement: The input layer extracts features from the sentiment nodes in the target heterogeneous graph to obtain the user sentiment feature vector. The service nodes in the target heterogeneous graph are extracted through the input layer to obtain service relationship feature vectors.
[0085] In one embodiment, when the processor 402 performs feature fusion on the user sentiment feature vector and the service relationship feature vector through the feature fusion layer to obtain the target feature vector, it is configured to: Obtain the first weight parameter corresponding to the user sentiment feature vector and the second weight parameter corresponding to the service relationship feature vector; The target feature vector is obtained by weighting the feature vectors based on the first weight parameter, the second weight parameter, the user sentiment feature vector, and the service relationship feature vector.
[0086] In one embodiment, when the processor 402 performs recommendation service prediction on the target feature vector through the output layer to obtain the recommendation service list, it is configured to: Service preference and sentiment prediction are performed on the target feature vector to obtain a recommendation score for each service; The services are sorted according to their recommendation scores to generate a list of recommended services, which includes multiple recommended services.
[0087] In one embodiment, when the processor 402 constructs the user's target heterogeneous graph based on the behavioral data, the emotional data, and the usage scenario data, it is configured to: The user and the behavior data are correlated to obtain a behavior heterogeneity graph. The user and the sentiment data are correlated using a heterogeneous sentiment structure to obtain a heterogeneous sentiment graph. The user and usage scenario data are correlated to obtain a scenario heterogeneity graph. The target heterogeneous graph is obtained by performing heterogeneous graph relationship association on the behavioral heterogeneous graph, the emotional heterogeneous graph, and the scene heterogeneous graph.
[0088] In one embodiment, the processor 402 is also configured to implement: Obtain the target service selected by the user based on the recommended service list; Based on the target service, determine the user's activity index, and update the model parameters of the service recommendation model based on the activity index.
[0089] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the computer device described above can be referred to the corresponding process in the aforementioned embodiment of the service recommendation method based on artificial intelligence, and will not be repeated here.
[0090] This application also provides a computer-readable storage medium storing a computer program, the computer program including program instructions, and the method implemented when the program instructions are executed can refer to various embodiments of the service recommendation method based on artificial intelligence in this application.
[0091] The computer-readable storage medium can be an internal storage unit of the computer device described in the foregoing embodiments, such as a hard disk or memory of the computer device. The computer-readable storage medium can be non-volatile or volatile. Alternatively, the computer-readable storage medium can be an external storage device of the computer device, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the computer device.
[0092] Furthermore, the computer-readable storage medium may primarily include a program storage area and a data storage area, wherein the program storage area may store the operating system, at least one application required for a function, etc.; and the data storage area may store data created based on the use of blockchain nodes, etc.
[0093] The blockchain referred to in this invention is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms. Essentially, a blockchain is a decentralized database, a chain of data blocks linked together using cryptographic methods. Each data block contains information about a batch of network transactions, used to verify the validity of the information (anti-counterfeiting) and generate the next block. A blockchain can include an underlying blockchain platform, a platform product service layer, and an application service layer.
[0094] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification, unless the context clearly indicates otherwise, the singular forms “a,” “an,” and “the” are intended to include the plural forms.
[0095] It should also be understood that the term "and / or" as used in this specification refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations. It should be noted that, herein, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.
[0096] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments. The above descriptions are merely specific implementations of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.
Claims
1. A service recommendation method based on artificial intelligence, characterized in that, include: Obtain multidimensional user data of the users to be recommended for service, including behavioral data, sentiment data, and usage scenario data; Based on the behavioral data, the emotional data, and the usage scenario data, construct the user's target heterogeneity graph; A service recommendation model is obtained, wherein the service recommendation model is a heterogeneous graph neural network model pre-trained based on multiple sample data; The target heterogeneous graph is identified and services are predicted using the service recommendation model to obtain a list of recommended services, which includes multiple recommended services.
2. The service recommendation method based on artificial intelligence as described in claim 1, characterized in that, The service recommendation model includes an input layer, a feature fusion layer, and an output layer; the process of identifying and predicting services for the target heterogeneous graph using the service recommendation model to obtain a list of recommended services includes: The input layer extracts features from the target heterogeneous graph to obtain a relation feature vector, which includes a user sentiment feature vector and a service relationship feature vector. The target feature vector is obtained by fusing the user sentiment feature vector and the service relationship feature vector through the feature fusion layer. The output layer is used to predict recommendation services for the target feature vector to obtain the recommendation service list.
3. The service recommendation method based on artificial intelligence as described in claim 2, characterized in that, The step of extracting features from the target heterogeneous graph through the input layer to obtain a relational feature vector includes: The input layer extracts features from the sentiment nodes in the target heterogeneous graph to obtain the user sentiment feature vector. The service nodes in the target heterogeneous graph are extracted through the input layer to obtain service relationship feature vectors.
4. The service recommendation method based on artificial intelligence as described in claim 2, characterized in that, The step of fusing the user sentiment feature vector and the service relationship feature vector through the feature fusion layer to obtain the target feature vector includes: Obtain the first weight parameter corresponding to the user sentiment feature vector and the second weight parameter corresponding to the service relationship feature vector; The target feature vector is obtained by weighting the feature vectors based on the first weight parameter, the second weight parameter, the user sentiment feature vector, and the service relationship feature vector.
5. The service recommendation method based on artificial intelligence as described in claim 2, characterized in that, The step of predicting recommendation services for the target feature vector through the output layer to obtain the recommendation service list includes: Service preference and sentiment prediction are performed on the target feature vector to obtain a recommendation score for each service; The services are sorted according to their recommendation scores to generate a list of recommended services, which includes multiple recommended services.
6. The service recommendation method based on artificial intelligence as described in claim 1, characterized in that, The step of constructing the user's target heterogeneous graph based on the behavioral data, the emotional data, and the usage scenario data includes: The user and the behavior data are correlated to obtain a behavior heterogeneity graph. The user and the sentiment data are correlated using a heterogeneous sentiment structure to obtain a heterogeneous sentiment graph. The user and usage scenario data are correlated to obtain a scenario heterogeneity graph. The target heterogeneous graph is obtained by performing heterogeneous graph relationship association on the behavioral heterogeneous graph, the emotional heterogeneous graph, and the scene heterogeneous graph.
7. The service recommendation method based on artificial intelligence as described in any one of claims 1-6, characterized in that, The method further includes: Obtain the target service selected by the user based on the recommended service list; Based on the target service, determine the user's activity index, and update the model parameters of the service recommendation model based on the activity index.
8. A service recommendation device, characterized in that, The service recommendation device includes an acquisition module, a construction module, and a generation module, wherein: The acquisition module is used to acquire multi-dimensional user data of the users to be recommended for service, including behavioral data, sentiment data, and usage scenario data. The construction module is used to construct the user's target heterogeneous graph based on the behavioral data, the emotional data, and the usage scenario data; The acquisition module is also used to acquire a service recommendation model, which is a heterogeneous graph neural network model pre-trained based on multiple sample data. The generation module is used to identify and predict services in the target heterogeneous graph through the service recommendation model to obtain a list of recommended services, which includes multiple recommended services.
9. A computer device, characterized in that, The computer device includes a processor, a memory, and a computer program stored in the memory and executable by the processor, wherein when the computer program is executed by the processor, it implements the steps of the artificial intelligence-based service recommendation method as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, it implements the steps of the artificial intelligence-based service recommendation method as described in any one of claims 1 to 7.