User data clustering methods, personalized recommendation methods and systems

Abstract

This application provides a user data clustering method, a personalized recommendation method, and a system, relating to the field of big data analysis and applicable to the financial and other fields. The user data clustering method includes: generating feature vector data through data preprocessing based on user data provided by different data sources authorized by users; constructing user matrix data with the same scale and distribution based on the feature vector data; obtaining multi-view data by adaptively weighting the user matrix data through preset parameters and a multi-view fusion model; and obtaining user clustering results through cluster analysis based on the multi-view data.

Description

Technical Field

[0001] This application relates to the field of big data analytics and can be used in the financial and other fields, particularly a user data clustering method, a personalized recommendation method, and a system. Background Technology

[0002] Mobile banking refers to the provision of financial services by banking institutions through mobile terminals. With the widespread adoption of smartphones and mobile internet, mobile banking has become the choice of more and more people. Personalized recommendations for mobile banking advertisements are an important promotional method. By analyzing user behavior and preferences, advertisements that better meet user needs are recommended, improving advertising effectiveness and user experience. Recommendation systems based on adaptive weighted multi-view fusion models can more accurately understand user needs and behaviors, achieving more precise personalized recommendation services, and represent an innovative technology for mobile banking advertising promotion.

[0003] Existing personalized recommendation technologies for mobile banking advertisements preprocess data using methods such as PCA dimensionality reduction and manual variable selection, and then utilize regression or clustering models to predict bank user identities. However, in mobile banking personalized recommendation applications, existing technologies are highly sensitive to noise and redundancy in high-dimensional bank user data, affecting model training and prediction performance. Furthermore, treating dense user background data and sparse user-initiated action data as the same data source may lead to insufficient prediction accuracy. Therefore, more efficient and accurate methods are needed to address these issues. Summary of the Invention

[0004] The purpose of this application is to provide a user data clustering method, a personalized recommendation method and system, to overcome the shortcomings of existing mobile data personalized recommendation technologies, making data recommendation simpler, less costly to implement and more accurate.

[0005] To achieve the above objectives, the user data clustering method provided in this application specifically includes: generating feature vector data through data preprocessing based on user data provided by different data sources authorized by the user; constructing user matrix data with the same scale and distribution based on the feature vector data; obtaining multi-view data by performing adaptive weighted multi-view fusion on the user matrix data through preset parameters and a multi-view fusion model; and obtaining user clustering results through cluster analysis based on the multi-view data.

[0006] In the above user data clustering method, optionally, feature vector data is generated by data preprocessing based on user data provided by different data sources authorized by the user, and user matrix data with the same scale and distribution is constructed based on the feature vector data. This includes: de-identifying the user data to obtain de-identified data; converting the de-identified data into feature vector data according to predetermined feature options; and unifying the feature vector data with different scales and distributions into a data matrix with the same scale and distribution through data transformation to obtain user matrix data in matrix form.

[0007] In the above user data clustering method, optionally, converting the de-identified data into feature vector data according to predetermined feature options further includes: performing data cleaning on the de-identified data to obtain prepared data that meets preset integrity requirements; integrating prepared data from different data sources to generate a dataset; and performing feature extraction on the dataset according to one or more predetermined feature options related to clustering to generate feature vector data.

[0008] In the above user data clustering method, optionally, obtaining multi-view data by adaptively weighting multi-view fusion of the user matrix data through preset parameters and a multi-view fusion model includes: generating two or more views from the feature types in the user matrix data through preset parameters and a multi-view fusion model, and analyzing the influence weights of the views; performing fusion processing on the views to generate fused data, and obtaining multi-view data based on the fused data and the influence weights.

[0009] In the above user data clustering method, optionally, the loss function of the multi-view fusion model includes:

[0010]

[0011]

[0012]

[0013] Where X1 is the first view data, X2 is the second view data, and U p Let V be the view-specific basis matrix for learning, and d be the view consensus matrix. p For view weights.

[0014] Optionally, in the above user data clustering method, the method further includes: obtaining accurate clustering data from user feedback based on the user clustering results; adjusting the preset parameters and / or the multi-view fusion model based on the accurate clustering data; and providing the adjusted preset parameters and / or the multi-view fusion model to the next clustering analysis of user data.

[0015] This application also provides a personalized recommendation method applicable to user data clustering methods, the method comprising: obtaining user group characteristics and / or individual user characteristics based on the user clustering results; and performing group data recommendation or individual data recommendation based on the data characteristics of the data to be recommended, using the user group characteristics and / or the individual user characteristics.

[0016] This application also provides a personalized recommendation system, the system comprising a user data clustering device and a recommendation device; the user data clustering device comprises a preprocessing module, a fusion module, and a clustering module; the preprocessing module is used to generate feature vector data through data preprocessing based on user data provided by different data sources authorized by the user, and to construct user matrix data with the same scale and distribution based on the feature vector data; the fusion module is used to perform adaptive weighted multi-view fusion on the user matrix data through preset parameters and a multi-view fusion model to obtain multi-view data; the clustering module is used to obtain user clustering results through clustering analysis based on the multi-view data; the recommendation device is used to obtain user group characteristics and / or individual user characteristics based on the user clustering results; and to perform group data recommendation or individual data recommendation based on the data characteristics of the data to be recommended using the user group characteristics and / or the individual user characteristics.

[0017] This application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described method.

[0018] This application also provides a computer-readable storage medium storing a computer program that performs the above-described methods.

[0019] This application also provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the above-described method.

[0020] The beneficial technical effects of this application are as follows: 1. Improved accuracy of data recommendation: By clustering similar user groups, advertisements that match their interests and needs can be recommended more accurately, thereby improving the accuracy of data recommendation and user experience. 2. Reduced cost of data recommendation: The adaptive weighted multi-view fusion model can effectively utilize existing data and information from multiple perspectives, reducing the data and human resource costs required for data recommendation. 3. Improved advertising effectiveness: Personalized data recommendations can increase the click-through rate and conversion rate of advertisements, thereby improving the effectiveness of advertising and bringing better marketing results to advertisers. 4. Increased user satisfaction: By providing users with advertisements that match their interests and needs, user satisfaction and loyalty can be improved, bringing better reputation and brand value to data providers. Attached Figure Description

[0021] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, do not constitute a limitation thereof. In the drawings:

[0022] Figure 1 This is a flowchart illustrating a user data clustering method provided in an embodiment of this application;

[0023] Figure 2 This is a schematic diagram of the user matrix data generation process provided in an embodiment of this application;

[0024] Figure 3 This is a schematic diagram of the feature vector data generation process provided in an embodiment of this application;

[0025] Figure 4 This is a schematic diagram of the user data preprocessing process provided in an embodiment of this application;

[0026] Figure 5 This is a schematic diagram of the process for acquiring multi-view data according to an embodiment of this application;

[0027] Figure 6 This is a flowchart illustrating the adaptive weighted multi-view fusion stage provided in an embodiment of this application;

[0028] Figure 7 A flowchart illustrating a personalized recommendation method provided in an embodiment of this application;

[0029] Figure 8 This is a schematic diagram of the structure of a personalized recommendation system provided in an embodiment of this application;

[0030] Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0031] The following will describe in detail the implementation methods of this application with reference to the accompanying drawings and embodiments, so as to fully understand how this application uses technical means to solve technical problems and achieve technical effects, and to implement it accordingly. It should be noted that, as long as there is no conflict, the various embodiments and features in each embodiment of this application can be combined with each other, and the resulting technical solutions are all within the protection scope of this application.

[0032] Furthermore, the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0033] It should be noted that the user data clustering method, personalized recommendation method and system disclosed in this application can be used in the field of financial technology, as well as in any field other than financial technology. The application fields of the user data clustering method, personalized recommendation method and system disclosed in this application are not limited.

[0034] Please refer to Figure 1 As shown, this application provides a user data clustering method, the method comprising:

[0035] S101 generates feature vector data through data preprocessing based on user data provided by different data sources authorized by the user, and constructs user matrix data with the same scale and distribution based on the feature vector data;

[0036] S102 performs adaptive weighted multi-view fusion on the user matrix data using preset parameters and a multi-view fusion model to obtain multi-view data;

[0037] S103 Obtains user clustering results through cluster analysis based on the multi-view data.

[0038] Specifically, the user data clustering method provided in this application is mainly divided into two stages: data processing and adaptive weighted multi-view fusion. The data processing stage is performed offline on the backend server after the user data is acquired; the fusion stage is performed offline on the backend server. The specific implementation logic of each stage will be described in detail in subsequent embodiments, and will not be described in detail here.

[0039] Please refer to Figure 2 As shown in one embodiment of this application, feature vector data is generated through data preprocessing based on user data provided by different data sources authorized by the user. The user matrix data with the same scale and distribution constructed based on the feature vector data includes:

[0040] S201 performs data anonymization on the user data to obtain anonymized data;

[0041] S202 converts the desensitized data into feature vector data according to predetermined feature options;

[0042] S203 transforms the feature vector data of different scales and distributions into a data matrix of the same scale and distribution to obtain user matrix data in matrix form.

[0043] Please refer to this again. Figure 3 As shown, in the above embodiment, converting the de-identified data into feature vector data according to predetermined feature options further includes:

[0044] S301 performs data cleaning on the de-identified data to obtain prepared data that meets the preset integrity requirements;

[0045] S302 integrates prepared data from different data sources to generate a dataset, and performs feature extraction on the dataset according to one or more predetermined feature options related to clustering to generate feature vector data.

[0046] Specifically, in practical work, the data processing stage of the user data clustering method provided in this application can be referenced. Figure 4 As shown, the data processing stage first involves inputting user data, followed by a series of preprocessing steps, including desensitization, data cleaning, data integration, feature selection, data transformation, and data storage, and finally outputting user data in matrix form.

[0047] The specific preprocessing operations and objectives of each stage are as follows: **Anonymization:** Eliminating sensitive personal information from the data to ensure user information is not leaked. **Data Cleaning:** Performing operations such as deduplication, missing value handling, outlier handling, and data type conversion on the data to ensure data quality and integrity. **Data Integration:** Integrating data from different data sources into a single dataset to prepare for subsequent modeling and analysis. **Feature Selection:** Empirically selecting cluster-related features and transforming the raw data into features usable for modeling using feature extraction techniques. **Manual Labeling:** Experienced professionals label user data by observing each user's feature, using the association strength level as the label. **Data Transformation:** Unifying data from different scales and distributions to the same scale and distribution so that the model can better learn data patterns and regularities. **Data Storage:** Storing the processed data in a database or file system for use in subsequent modeling and analysis tasks.

[0048] Please refer to Figure 5 As shown, in one embodiment of this application, the multi-view data obtained by adaptively weighting the user matrix data through multi-view fusion using preset parameters and a multi-view fusion model includes:

[0049] S501 generates two or more views from the feature types in the user matrix data using preset parameters and a multi-view fusion model, and analyzes the influence weight of the views;

[0050] S502 performs a fusion process on the views to generate fused data, and obtains multi-view data based on the fused data and the influence weights.

[0051] Specifically, the adaptive weighted multi-view fusion stage in practical applications of this application can be referenced. Figure 6As shown, the steps are as follows: First, input user data in matrix form. Then, set parameters based on experience, followed by multi-view fusion, and then clustering. Finally, output the user group data. In machine learning, parameters refer to those that need to be manually set during model building. Different parameter settings can significantly affect the model's performance and generalization ability. Parameter selection usually requires experience and trial and error, making it a crucial task. Setting appropriate parameters can improve model performance, while over-adjusting parameters can lead to overfitting or underfitting. Therefore, techniques such as cross-validation are needed to adjust hyperparameters and evaluate their impact on the model to select the optimal hyperparameter combination. Of course, the hyperparameters set in the clustering stage can also be empirically set to achieve good evaluation results. In practice, semi-supervised learning can be replaced with unsupervised learning, eliminating the need for labeled data. However, while this reduces the cost of data labeling, it significantly impacts prediction accuracy. Therefore, the specific choice can be made according to actual needs, and this application does not impose further limitations here.

[0052] Multi-view Fusion: The adaptive weighted multi-view fusion model is a method for extracting features from two data views. This model mainly includes the following steps: Multi-view Fusion: Selected features are fused to enhance the data's expressive power. Adaptive Weighting: To improve feature quality, an adaptive weighting mechanism is introduced, weighting features in each view to better capture the data's distribution characteristics. This model utilizes data from two views, improving data expressiveness and clustering accuracy through adaptive weighting and feature fusion. Simultaneously, the unsupervised approach automatically discovers latent features in the data without requiring any prior knowledge or labeled data. Therefore, this model has high flexibility and practicality. Clustering: By clustering the multi-view data extracted based on the adaptive weighted multi-view fusion model, different user group characteristics can be discovered, and users can be divided into different groups. This allows for better ad recommendations, as different groups have different preferences and needs, thus enabling more targeted ad recommendations. Furthermore, clustering groups similar users together, helping to better understand user behavior patterns and purchasing preferences, thereby improving marketing strategies and product solutions, enhancing market competitiveness and business revenue.

[0053] In the above embodiments, multi-view data fusion can merge user background data and operational data as two different views, and can support more views. This multi-view data fusion method helps to improve the dimensionality and diversity of data, and more comprehensively reflects user behavior and preferences. Background data and operational data can be user account data and user operation data. User account information includes: basic account information, financial status, risk preference, financial capacity, investment history and style, etc.; user ID. User operation information includes: user browsing time history on different financial product advertising pages, user click history on different financial product advertising pages, records of financial products consulted by the user in customer service conversations, etc.; user ID; specific indicators can be flexibly changed according to privacy and collection levels. It is worth noting that the user information in the embodiments of this application is obtained through legal and compliant means, and the acquisition, storage, use, and processing of user information have been authorized and agreed upon by the customer.

[0054] In one embodiment of this application, the loss function of the multi-view fusion model includes:

[0055]

[0056]

[0057]

[0058] Where X1 is the first view data, X2 is the second view data, and U p Let V be the view-specific basis matrix for learning, and d be the view consensus matrix. p For view weights.

[0059] Specifically, in practical applications, the multi-view fusion model provided in this application ensures real-time performance and accuracy during the testing phase, playing a role in dual-weighting of samples and features, incomplete learning, and semi-supervised clustering. The total loss function O of this model is as follows:

[0060]

[0061]

[0062] in

[0063] First item It is based on an adaptive weighted multi-view fusion model. X1 represents the first view data, X2 represents the second view data; U p V is the view-specific basis matrix used to indicate the central features of the model cluster; V is the view consensus matrix, enabling the model to learn the latent features of individual users from both views; d pThis is for view weights. The purpose of adding this is to enable the model to adaptively learn information from different views, so that the model can correctly cluster user groups.

[0064] Second item It is a smooth constraint of the basis matrix unique to the view, which helps to prevent overfitting in the model. β is a hyperparameter used to adjust the model balance.

[0065] The constraint is This serves to constrain the sum of view weights to 1, limiting the weight sum to remain unchanged and preventing the model from overfitting. This is an adaptive learning method for view weights.

[0066] Once the total loss function is determined, optimization can be performed using the conventional multiplicative update rule, which will not be elaborated here.

[0067] In one embodiment of this application, the method may further include: obtaining accurate clustering data from user feedback based on the user clustering results; adjusting the preset parameters and / or the multi-view fusion model based on the accurate clustering data; and providing the adjusted preset parameters and / or the multi-view fusion model to the next clustering analysis of user data.

[0068] Specifically, in practical work, to ensure the accuracy and stability of multi-view fusion clustering, a subsequent monitoring phase is required. This monitoring phase mainly involves the following actions: Monitoring algorithm operation: Establishing a real-time algorithm monitoring system on the platform to monitor the algorithm's operational status, promptly identify and address problems. Supervising data quality: Labeled data used for supervised learning needs to undergo quality checks to ensure accuracy and completeness, thereby improving algorithm accuracy. Continuously updating the model: User preferences may change over time, necessitating continuous model updates and optimization to maintain accuracy. Providing data visualization capabilities: Providing data visualization capabilities on the platform to allow users to more intuitively understand model operation and prediction results, facilitating supervision and optimization. Providing feedback channels: Establishing user feedback channels to collect user feedback and opinions, promptly identify and address problems, thereby improving model accuracy and user experience. Since the monitoring components can be implemented using existing technologies, this application will not elaborate on them further.

[0069] Please refer to Figure 7 As shown, in one embodiment of this application, a personalized recommendation method incorporating the aforementioned user data clustering method is also provided, the method comprising:

[0070] S701 obtains user group characteristics and / or individual user characteristics based on the user clustering results;

[0071] S702 performs group data recommendation or individual data recommendation based on the data characteristics of the data to be recommended, using the user group characteristics and / or the individual user characteristics.

[0072] Specifically, after completing user clustering, similar user groups can be identified, and data recommendations can be made using the following methods:

[0073] Group data recommendation: Targeted data delivery to all users within a specific group. Representative data recommendation: Select a representative group of users from the group and deliver data to them to maximize the satisfaction of the entire group's needs. Personalized data recommendation: Recommend the most suitable data for each individual user based on their group and individual characteristics. These three methods can be combined to achieve more accurate data recommendations. Furthermore, A / B testing and other methods can be used to evaluate and optimize the recommendation performance.

[0074] Please refer to Figure 8 As shown, this application also provides a personalized recommendation system, which includes a user data clustering device and a recommendation device. The user data clustering device includes a preprocessing module, a fusion module, and a clustering module. The preprocessing module is used to generate feature vector data through data preprocessing based on user data provided by different data sources authorized by the user, and to construct user matrix data with the same scale and distribution based on the feature vector data. The fusion module is used to perform adaptive weighted multi-view fusion on the user matrix data through preset parameters and a multi-view fusion model to obtain multi-view data. The clustering module is used to obtain user clustering results through clustering analysis based on the multi-view data. The recommendation device is used to obtain user group characteristics and / or individual user characteristics based on the user clustering results. Based on the data characteristics of the data to be recommended, the user group characteristics and / or individual user characteristics are used to perform group data recommendation or individual data recommendation. The specific implementation process of each component in this system can be referred to the relevant descriptions in the foregoing embodiments, and will not be detailed here.

[0075] The beneficial technical effects of this application are as follows: 1. Improved accuracy of data recommendation: By clustering similar user groups, advertisements that match their interests and needs can be recommended more accurately, thereby improving the accuracy of data recommendation and user experience. 2. Reduced cost of data recommendation: The adaptive weighted multi-view fusion model can effectively utilize existing data and information from multiple perspectives, reducing the data and human resource costs required for data recommendation. 3. Improved advertising effectiveness: Personalized data recommendations can increase the click-through rate and conversion rate of advertisements, thereby improving the effectiveness of advertising and bringing better marketing results to advertisers. 4. Increased user satisfaction: By providing users with advertisements that match their interests and needs, user satisfaction and loyalty can be improved, bringing better reputation and brand value to data providers.

[0076] This application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described method.

[0077] This application also provides a computer-readable storage medium storing a computer program that performs the above-described methods.

[0078] This application also provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the above-described method.

[0079] like Figure 9 As shown, the electronic device 600 may also include: a communication module 110, an input unit 120, an audio processor 130, a display 160, and a power supply 170. It is worth noting that the electronic device 600 does not necessarily need to include these components. Figure 9 All components shown; in addition, the electronic device 600 may also include Figure 9 For components not shown, please refer to existing technologies.

[0080] like Figure 9 As shown, the central processing unit 100, sometimes also referred to as a controller or operating control, may include a microprocessor or other processor device and / or logic device. The central processing unit 100 receives inputs and controls the operation of various components of the electronic device 600.

[0081] The memory 140 may be, for example, one or more of a cache, flash memory, hard drive, removable media, volatile memory, non-volatile memory, or other suitable devices. It may store the aforementioned failure-related information, and also store a program for executing that information. The central processing unit 100 may execute the program stored in the memory 140 to perform information storage or processing, etc.

[0082] Input unit 120 provides input to central processing unit 100. Input unit 120 may be, for example, a keypad or touch input device. Power supply 170 provides power to electronic device 600. Display 160 displays images and text. Display may be, for example, an LCD display, but is not limited thereto.

[0083] The memory 140 can be a solid-state memory, such as a read-only memory (ROM), random access memory (RAM), a SIM card, etc. It can also be a memory that retains information even when power is off, can be selectively erased, and contains more data; examples of this type of memory are sometimes referred to as EPROMs. The memory 140 can also be some other type of device. The memory 140 includes a buffer memory 141 (sometimes referred to as a buffer). The memory 140 may include an application / function storage unit 142 for storing application programs and function programs or processes for executing the operation of the electronic device 600 via the central processing unit 100.

[0084] The memory 140 may also include a data storage unit (data 143) for storing data, such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The driver storage unit (driver 144) of the memory 140 may include various drivers for the electronic device's communication functions and / or for performing other functions of the electronic device (such as messaging applications, address book applications, etc.).

[0085] The communication module 110 is a transmitter / receiver 110 that transmits and receives signals via antenna 111. The communication module (transmitter / receiver) 110 is coupled to the central processing unit 100 to provide input signals and receive output signals, which can be the same as in a conventional mobile communication terminal.

[0086] Based on different communication technologies, multiple communication modules 110 can be configured in the same electronic device, such as cellular network modules, Bluetooth modules, and / or wireless LAN modules. The communication module (transmitter / receiver) 110 is also coupled to a speaker 131 and a microphone 132 via an audio processor 130 to provide audio output via the speaker 131 and receive audio input from the microphone 132, thereby enabling typical telecommunications functions. The audio processor 130 may include any suitable buffer, decoder, amplifier, etc. Additionally, the audio processor 130 is coupled to a central processing unit 100, enabling on-device recording via the microphone 132 and on-device playback of stored audio via the speaker 131.

[0087] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0088] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0089] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0090] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0091] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of this application. It should be understood that the above descriptions are merely specific embodiments of this application and are not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method of clustering user data, characterized by, The method comprises: generating feature vector data from user data provided by different data sources authorized by a user through data preprocessing, and constructing user matrix data with the same scale and distribution according to the feature vector data; obtaining multi-view data by adaptive weight multi-view fusion of the user matrix data through a preset parameter and a multi-view fusion model; the loss function of the multi-view fusion model comprises: where X1is the first view data, X2is the second view data, U p is the learned view-specific basis matrix, V is the view-common basis matrix, d p is the view weight, and β is a hyperparameter. obtaining user clustering results by clustering analysis processing according to the multi-view data.

2. The user data clustering method of claim 1, wherein, Generating feature vector data from user data provided by different data sources authorized by a user through data preprocessing, and constructing user matrix data with the same scale and distribution according to the feature vector data comprises: obtaining desensitized data by desensitizing the user data; converting the desensitized data into feature vector data according to a predetermined feature option; unifying the feature vector data with different scales and distributions into a data matrix with the same scale and distribution through data transformation to obtain user matrix data in the form of a matrix.

3. The user data clustering method of claim 2, wherein, Converting the desensitized data into feature vector data according to a predetermined feature option further comprises: obtaining preparation data meeting a preset integrity requirement by data cleaning of the desensitized data; integrating preparation data sets of different data sources to generate a data set, and extracting features from the data set according to one or more predetermined feature options related to clustering to generate feature vector data.

4. The user data clustering method of claim 1, wherein, Obtaining multi-view data by adaptive weight multi-view fusion of the user matrix data through a preset parameter and a multi-view fusion model comprises: generating two or more views from feature types in the user matrix data through a preset parameter and a multi-view fusion model, and analyzing the influence weight of the views; fusing the views to generate fusion data, and obtaining multi-view data according to the fusion data and the influence weight.

5. The user data clustering method of claim 1, wherein, The method further comprises: obtaining clustering accurate data of user feedback according to the user clustering results; adjusting the preset parameter and / or the multi-view fusion model according to the clustering accurate data, and providing the adjusted preset parameter and / or the multi-view fusion model to the clustering analysis of the next user data.

6. A personalized recommendation method suitable for use in the user data clustering method of any one of claims 1 to 5, characterized in that, The method comprises: obtaining user group features and / or user individual features according to the user clustering results; performing group data recommendation or individual data recommendation by using the user group features and / or the user individual features according to the data features of the data to be recommended.

7. A personalized recommendation system characterized by, The system comprises a user data clustering device and a recommendation device; The user data clustering device comprises a preprocessing module, a fusion module and a clustering module; The preprocessing module is used to generate feature vector data from user data provided by different data sources authorized by a user through data preprocessing, and to construct user matrix data with the same scale and distribution according to the feature vector data; The fusion module is used to obtain multi-view data by adaptive weight multi-view fusion of the user matrix data through a preset parameter and a multi-view fusion model; the loss function of the multi-view fusion model comprises: where X1is the first view data, X2is the second view data, U p is the learned view-specific basis matrix, V is the view-common basis matrix, d p is the view weight, and β is a hyperparameter. The clustering module is used to obtain user clustering results by clustering analysis processing according to the multi-view data; The recommendation device is used for obtaining user group features and / or user individual features according to the user clustering result; and performing group data recommendation or individual data recommendation by using the user group features and / or the user individual features according to data features of to-be-recommended data.

8. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor implements the method in any of claims 1 to 6 when executing the computer program.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores the computer program for implementing the method in any of claims 1 to 6 by a computer.

Images