Personalized financial product recommendation system and method based on user profiles

Abstract

The present application relates to the field of finance, and particularly relates to a personalized financial product recommendation system and method based on user profiles. The method comprises: acquiring user profile data of a target user, wherein the user profile data comprises basic personal information, financial information, historical transaction records and behavior data; on the basis of the user profile data, constructing a dynamic preference model of the user; screening out from a product pool a financial product set matching the dynamic preference model; on the basis of features of financial products and historical behavior data of the user, determining recommendation priorities; and recommending to the target user a financial product having the highest recommendation priority. By means of updating the preferences and requirements of users in real time, the present invention improves the accuracy and real-time performance of recommending financial products, thereby improving the user experience and recommendation effect.

Description

A Personalized Financial Product Recommendation System and Method Based on User Profile Technical Field

[0001] This invention belongs to the fields of financial technology and artificial intelligence technology, specifically a personalized financial product recommendation system and method based on user profiles. Background Technology

[0002] With the rapid development of fintech, financial product recommendation technology has gradually become an important research direction in the financial field due to its ability to efficiently match user needs with financial products. However, existing financial product recommendation systems and methods still have some shortcomings in terms of the accuracy of user profiling, the diversity of recommendation algorithms, and real-time performance, which affect the recommendation effect and user experience.

[0003] A search revealed a method and system for recommending financial products, with publication number CN118212041B and publication date July 12, 2024. This patent acquires user profile data, holding data, and product pool data of the target user, constructs a decision model based on the user profile data, and recommends the first decision data with the highest similarity to the target user. This technical solution improves the stability of the recommendation and reduces the loss of returns caused by users frequently changing financial products. However, the construction of the decision model in this technical solution is relatively simple and does not fully consider the dynamic behavior of users and market changes, which may lead to insufficient real-time performance and accuracy of the recommendation results. Furthermore, this method mainly relies on the data held by the user, which may not fully reflect the user's actual needs and preferences.

[0004] A search revealed a method and apparatus for recommending financial products, with publication number CN112801803B and publication date February 2, 2024. This patent determines the recommended neighbor set and generates a list of recommended financial products by acquiring customers' historical operation data on financial products, the set of financial products to be recommended, and a threshold for the number of common attributes. This technical solution accurately identifies customer attributes and target markets, improving service efficiency and customer satisfaction. However, in this technical solution, the determination of the recommended neighbor set is mainly based on historical operation data, without fully considering users' dynamic behavior and market changes, which may lead to insufficient real-time performance and accuracy of the recommendation results. In addition, the analysis of customer behavior by this method is relatively simplistic and may not fully reflect the actual needs and preferences of users. Technical issues

[0005] The aforementioned problems indicate that existing financial product recommendation systems and methods still have certain shortcomings in terms of the accuracy of user profiling, the diversity of recommendation algorithms, and real-time performance. Therefore, this invention provides a personalized financial product recommendation system and method based on user profiling, aiming to optimize user profiling construction, improve the diversity and real-time performance of recommendation algorithms, thereby enhancing recommendation effectiveness and user experience, and meeting the financial sector's demand for efficient and accurate recommendation systems. Technical solutions

[0006] This invention provides a personalized financial product recommendation system and method based on user profiles, which addresses the shortcomings of existing technologies in terms of the accuracy of user profiles, the diversity of recommendation algorithms, and real-time performance, thereby improving the accuracy of financial product recommendations and user experience.

[0007] To address the aforementioned problems, one embodiment of the present invention provides a method for recommending personalized financial products based on user profiles, comprising:

[0008] Obtain user profile data of the target users; wherein, the user profile data includes the user's basic personal information, financial information, historical transaction records and behavioral data;

[0009] A dynamic preference model for users is constructed based on the user profile data; wherein, the dynamic preference model is used to update user preferences and needs in real time;

[0010] Select a set of financial products from the product pool that match the dynamic preference model;

[0011] The recommendation priority is determined based on the characteristics of the financial products in the aforementioned financial product set and the user's historical behavior data;

[0012] Recommend the highest priority financial products to the target users.

[0013] As an improvement to the above solution, the step of constructing a dynamic user preference model based on the user profile data includes:

[0014] Feature extraction is performed on the user profile data to obtain the user's static and dynamic features; wherein, the static features include the user's basic personal information and financial information, and the dynamic features include the user's historical transaction records and behavioral data;

[0015] Based on the static and dynamic features, a dynamic preference model for users is constructed using machine learning algorithms; wherein, the dynamic preference model can be updated according to the user's real-time behavioral data.

[0016] As an improvement to the above solution, the step of selecting a set of financial products from the product pool that match the dynamic preference model includes:

[0017] Obtain feature data of all wealth management products in the product pool; wherein, the feature data includes the product's risk level, expected return, investment period, and product type;

[0018] Based on the user's preferences and needs in the dynamic preference model, the matching degree of the feature data is calculated;

[0019] Add financial products with a matching degree higher than a preset threshold to the financial product set.

[0020] As an improvement to the above solution, the step of determining the recommendation priority based on the characteristics of the financial products in the set of financial products and the user's historical behavior data includes:

[0021] For each financial product in the aforementioned financial product set, a recommendation score is calculated based on the user's historical behavior data and product feature data;

[0022] The financial products in the financial product set are sorted according to the recommendation scores.

[0023] The ranking results will be used as the recommendation priority.

[0024] As an improvement to the above scheme, the calculation of the recommendation score includes:

[0025] Obtain users' historical behavior data, including their historical purchase records, browsing history, and feedback records;

[0026] Extract key features from the historical behavioral data, including users' preferences for different product types, risk tolerance, and investment horizon preferences;

[0027] Based on the key features and financial product feature data, a recommendation score is calculated using a preset scoring model; wherein, the preset scoring model can be updated according to the user's real-time behavior data.

[0028] As an improvement to the above solution, recommending the highest-priority financial products to the target user includes:

[0029] Generate a recommendation list and add the top N financial products with the highest recommendation priority to the recommendation list;

[0030] The recommendation list is displayed to the target user via the user terminal;

[0031] Record user feedback data on the recommendation list, including click-through rate, purchase rate, and satisfaction.

[0032] The dynamic preference model and recommendation algorithm are optimized based on the feedback data.

[0033] Accordingly, one embodiment of the present invention also provides a personalized financial product recommendation system based on user profiles, including: a data acquisition module, a dynamic preference model construction module, a product screening module, a recommendation priority determination module, and a recommendation module;

[0034] The data acquisition module is used to acquire user profile data of the target user; wherein, the user profile data includes the user's basic personal information, financial information, historical transaction records and behavioral data;

[0035] The dynamic preference model construction module is used to construct a dynamic preference model for users based on the user profile data; wherein, the dynamic preference model is used to update users' preferences and needs in real time.

[0036] The product screening module is used to select a set of financial products from the product pool that match the dynamic preference model.

[0037] The recommendation priority determination module is used to determine the recommendation priority based on the characteristics of the financial products in the financial product set and the user's historical behavior data;

[0038] The recommendation module is used to recommend the highest-priority financial products to the target user.

[0039] Accordingly, one embodiment of the present invention also provides a computer terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the personalized financial product recommendation method based on user profiles described in the present invention.

[0040] Accordingly, one embodiment of the present invention also provides a computer-readable storage medium, the computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute the personalized financial management product recommendation method based on user profiles described in the present invention. Beneficial effects

[0041] This invention provides a personalized financial product recommendation system and method based on user profiles. By acquiring user profile data of target users, a dynamic preference model is constructed. A set of financial products matching the dynamic preference model is selected from a product pool. Recommendation priorities are determined based on product characteristics and user historical behavior data. Finally, the financial product with the highest recommendation priority is recommended to the target user. This invention improves the accuracy and timeliness of financial product recommendations by updating user preferences and needs in real time, thereby enhancing user experience and recommendation effectiveness. Attached Figure Description

[0042] Figure 1 is a flowchart illustrating a personalized financial product recommendation method based on user profiles provided in an embodiment of the present invention.

[0043] Figure 2 is a schematic diagram of the structure of a personalized financial product recommendation system based on user profiles provided in an embodiment of the present invention;

[0044] Figure 3 is a schematic diagram of the structure of a dynamic preference model construction module provided in an embodiment of the present invention;

[0045] Figure 4 is a schematic diagram of the product screening module provided in an embodiment of the present invention;

[0046] Figure 5 is a schematic diagram of the recommendation priority determination module provided in an embodiment of the present invention;

[0047] Figure 6 is a schematic diagram of the structure of a computer terminal device provided in an embodiment of the present invention. Embodiments of the present invention

[0048] This invention provides a personalized financial product recommendation system and method based on user profiles. By acquiring user profile data of target users, a dynamic preference model is constructed. A set of financial products matching the dynamic preference model is selected from a product pool. Recommendation priorities are determined based on product characteristics and user historical behavior data. Finally, the financial product with the highest recommendation priority is recommended to the target user. This invention improves the accuracy and timeliness of financial product recommendations by updating user preferences and needs in real time, thereby enhancing user experience and recommendation effectiveness. The specific embodiments of this invention are described in detail below with reference to Figures 1 to 6.

[0049] First, referring to Figure 1, this embodiment of the invention provides a method for recommending personalized financial products based on user profiles, the steps of which include:

[0050] Step 101: Obtain user profile data for the target user. This user profile data includes the user's basic personal information, financial information, historical transaction records, and behavioral data. Specifically, basic personal information may include the user's age, gender, occupation, and educational background; financial information may include the user's income level, asset status, and debt situation; historical transaction records may include records of past purchases of financial products, stock transactions, and insurance purchases; and behavioral data may include the user's browsing behavior, search behavior, and click behavior. This data can be obtained through the user's registration information and transaction records with financial institutions, as well as behavioral records on the financial institution's website or mobile application.

[0051] Step 102: Construct a dynamic preference model for the user based on the user profile data. In specific implementation, step 102 includes the following sub-steps:

[0052] Step 1021: Extract features from the user profile data to obtain the user's static and dynamic features. The static features include the user's basic personal information and financial information, while the dynamic features include the user's historical transaction records and behavioral data. Feature extraction methods can employ traditional statistical analysis methods or feature selection algorithms from machine learning, such as Principal Component Analysis (PCA) and feature importance analysis (e.g., feature importance in the Random Forest algorithm).

[0053] Step 1022: Based on the aforementioned static and dynamic features, construct a dynamic preference model for the user using machine learning algorithms. The core of a dynamic preference model lies in its ability to be updated based on real-time user behavior data to maintain model accuracy. Commonly used machine learning algorithms include, but are not limited to, Support Vector Machines (SVM), Decision Trees, Random Forests, and Neural Networks. When constructing a dynamic preference model, the user's static features can be used as input variables, and the user's dynamic features as adaptive update variables. Through the combined effect of these variables, the user's preferences and needs are dynamically adjusted.

[0054] Step 103: Select a set of financial products from the product pool that match the dynamic preference model. In specific implementation, step 103 includes the following sub-steps:

[0055] Step 1031: Obtain the feature data of all wealth management products in the product pool. The feature data includes the product's risk level, expected return, investment period, and product type. This data can be obtained through internal databases of financial institutions, third-party data platforms, etc., and stored in the system's product pool.

[0056] Step 1032: Calculate the matching degree of the feature data based on the user's preferences and needs in the dynamic preference model. The matching degree calculation can be implemented using various algorithms, such as methods based on cosine similarity or Euclidean distance. Specifically, the user's preferences and needs can be quantified into a series of values, and then these values ​​can be compared with the feature data of the financial products to calculate the matching degree of each financial product. The formula for calculating the matching degree can be expressed as:

[0057] Match degree = ∑i=1nwi⋅∣pi−ui∣ Match degree = ∑i=1n​wi​⋅∣pi​−ui​∣

[0058] Where wiwi​ represents the weight, pipi​ represents the feature data of the financial product, uiui​ represents the user preference and demand data, and nn represents the dimension of the feature data.

[0059] Step 1033: Add wealth management products with a matching degree higher than a preset threshold to the wealth management product set. The preset threshold can be adjusted according to the actual application scenario. For example, it can be set to 0.7, meaning that only wealth management products with a matching degree of more than 70% will be included in the recommendation range.

[0060] Step 104: Determine the recommendation priority based on the characteristics of the financial products in the aforementioned financial product set and the user's historical behavior data. In specific implementation, step 104 includes the following sub-steps:

[0061] Step 1041: For each financial product in the aforementioned financial product set, calculate a recommendation score based on the user's historical behavior data and product feature data. The recommendation score can be calculated using various methods, such as weighted average, collaborative filtering algorithms, and deep learning models. Specifically, key features can be extracted from the user's historical purchase records, browsing history, and feedback records, including the user's preferences for different product types, risk tolerance, and investment period preferences. The formula for calculating the recommendation score can be expressed as:

[0062] Recommendation Score = w1⋅(Purchase History Preference) + w2⋅(Browsing History Preference) + w3⋅(Feedback History Preference) + w4⋅(Risk Tolerance Match) + w5⋅(Investment Term Preference Match)

[0063] Among them, w1, w2, w3, w4, w5 are the weights of each feature, which can be adjusted according to the actual application scenario.

[0064] Step 1042: Sort the financial products in the financial product set according to the recommendation scores. Algorithms such as quicksort and heapsort can be used to sort the recommendation scores and generate a recommendation priority list.

[0065] Step 1043: Use the ranking results as the recommendation priority. The top N financial products in the recommendation priority list will be recommended to the user. The specific value of N can be set according to the actual application needs, such as recommending the top 5 financial products.

[0066] Step 105: Recommend the highest priority financial products to the target users. In practice, Step 105 includes the following sub-steps:

[0067] Step 1051: Generate a recommendation list, adding the top N financial products with the highest recommendation priority to the list. The recommendation list can be displayed in the form of a table or list for easy viewing and selection by users.

[0068] Step 1052: Display the recommendation list to the target user through the user terminal. The user terminal can be a PC, mobile device (such as a smartphone, tablet), etc., and the recommendation list can be displayed to the user through the website of a financial institution, mobile application, etc.

[0069] Step 1053: Record user feedback data on the recommendation list, including click-through rate, purchase rate, and satisfaction level. This feedback data can be automatically collected through the user's terminal log system and stored in the system's backend database for subsequent optimization analysis.

[0070] Step 1054: Optimize the dynamic preference model and recommendation algorithm based on the feedback data. Specific optimization methods may include, but are not limited to, adjusting model parameters, updating the recommendation algorithm, and adding new behavioral features. Through continuous optimization, the accuracy of recommendations and user experience can be further improved.

[0071] Referring to Figure 2, an embodiment of the present invention also provides a personalized financial product recommendation system based on user profiles. This system includes a data acquisition module 201, a dynamic preference model construction module 202, a product screening module 203, a recommendation priority determination module 204, and a recommendation module 205. The functions of each module are as follows:

[0072] The data acquisition module 201 is used to acquire user profile data of the target users. Specifically, this module can obtain users' basic personal information, financial information, historical transaction records, and behavioral data from multiple channels, including internal databases of financial institutions, third-party data platforms, and user terminals. This data acquisition can be achieved through API interfaces, database queries, and other methods, ensuring the real-time nature and accuracy of the data.

[0073] The dynamic preference model construction module 202 is used to construct a dynamic preference model for users based on the user profile data. Specifically, this module first extracts features from the user profile data to obtain the user's static and dynamic features, and then uses machine learning algorithms to construct a dynamic preference model based on these features. The detailed process of feature extraction is shown in Figure 3. In Figure 3, the dynamic preference model construction module 202 includes a feature extraction submodule 301 and a model construction submodule 302. The feature extraction submodule 301 is responsible for extracting static and dynamic features from the user profile data, while the model construction submodule 302 is responsible for constructing the dynamic preference model using machine learning algorithms. The specific implementation of the feature extraction submodule 301 can employ traditional statistical analysis methods or feature selection algorithms in machine learning, such as principal component analysis (PCA) and feature importance analysis (such as feature importance in the random forest algorithm). The model construction submodule 302 can use algorithms such as support vector machines (SVM), decision trees, random forests, and neural networks. The model constructed using these algorithms can be updated according to the user's real-time behavior data, dynamically adjusting the user's preferences and needs.

[0074] Product selection module 203 is used to select a set of financial products from the product pool that match the dynamic preference model. Specifically, this module first obtains the feature data of all financial products in the product pool, then calculates the matching degree of these feature data according to the user's preferences and needs in the dynamic preference model, and finally adds financial products with a matching degree higher than a preset threshold to the financial product set. The specific structure of product selection module 203 is shown in Figure 4. In Figure 4, product selection module 203 includes a data acquisition submodule 401, a matching degree calculation submodule 402, and a set generation submodule 403. Data acquisition submodule 401 is responsible for obtaining the feature data of financial products from the product pool, matching degree calculation submodule 402 is responsible for calculating the matching degree of each financial product, and set generation submodule 403 is responsible for generating a set of financial products. The matching degree can be calculated using various algorithms, such as the calculation method based on cosine similarity, the calculation method based on Euclidean distance, etc. Specifically, the user's preferences and needs can be quantified into a series of values, and then these values ​​can be compared with the feature data of financial products to calculate the matching degree of each financial product. The formula for calculating the matching degree can be expressed as:

[0075] Match degree = ∑i=1nwi⋅∣pi−ui∣ Match degree = ∑i=1n​wi​⋅∣pi​−ui​∣

[0076] Where wiwi represents the weight, pipi represents the feature data of the financial product, uiui represents the user preference and demand data, and nn represents the dimension of the feature data. The preset threshold can be adjusted according to the actual application scenario. For example, it can be set to 0.7, meaning that only financial products with a matching degree of more than 70% will be included in the recommendation range.

[0077] The recommendation priority determination module 204 is used to determine the recommendation priority based on the characteristics of the financial products in the financial product set and the user's historical behavior data. Specifically, this module first calculates a recommendation score for each financial product in the financial product set based on the user's historical behavior data and product feature data, and then sorts the financial products according to the recommendation scores to generate a recommendation priority list. The specific structure of the recommendation priority determination module 204 is shown in Figure 5. In Figure 5, the recommendation priority determination module 204 includes a score calculation submodule 501, a sorting submodule 502, and a priority generation submodule 503. The score calculation submodule 501 is responsible for calculating the recommendation score for each financial product, the sorting submodule 502 is responsible for sorting the recommendation scores, and the priority generation submodule 503 is responsible for generating the recommendation priority list. The recommendation score can be calculated using various methods, such as weighted average method, collaborative filtering algorithm, deep learning model, etc. Specifically, key features can be extracted from the user's historical purchase records, browsing records, and feedback records, including the user's preference for different product types, risk tolerance, and investment period preference. The formula for calculating the recommendation score can be expressed as:

[0078] Recommendation Score = w1⋅(Purchase History Preference) + w2⋅(Browsing History Preference) + w3⋅(Feedback History Preference) + w4⋅(Risk Tolerance Match) + w5⋅(Investment Term Preference Match)

[0079] Where w1, w2, w3, w4, w5 represent the weights of each feature, which can be adjusted according to the actual application scenario. The financial products in the financial product set are sorted according to the recommendation scores to generate a recommendation priority list.

[0080] The recommendation module 205 recommends the highest-priority financial products to the target user. Specifically, this module first generates a recommendation list, adds the top N highest-priority financial products to the list, and then displays the recommendation list to the target user through the user terminal. The user terminal can be a PC, mobile device (such as a smartphone or tablet), etc., and the recommendation list can be displayed to the user through the financial institution's website, mobile application, etc. The specific structure of the recommendation module 205 is shown in Figure 6. In Figure 6, the recommendation module 205 includes a recommendation list generation submodule 601 and a display submodule 602. The recommendation list generation submodule 601 is responsible for generating the recommendation list, and the display submodule 602 is responsible for displaying the recommendation list through the user terminal. The recommendation list can be displayed in the form of tables, lists, etc., for easy viewing and selection by the user. At the same time, the recommendation module 205 is also responsible for recording user feedback data on the recommendation list, including click-through rate, purchase rate, and satisfaction. This feedback data can be automatically collected through the user terminal's log system and stored in the system's backend database for subsequent optimization analysis. The dynamic preference model and recommendation algorithm are optimized based on the feedback data. Specific optimization methods may include, but are not limited to, adjusting model parameters, updating recommendation algorithms, and adding new behavioral features. Through continuous optimization, the accuracy of recommendations and user experience can be further improved.

[0081] To better understand the implementation process of this invention, a specific practical application scenario is used for illustration below. Assume a financial institution uses the personalized financial product recommendation system of this invention to recommend suitable financial products to user A. User A's basic personal information includes age 30, female, occupation: teacher, educational background: undergraduate, etc.; financial information includes monthly income of 10,000 yuan, total assets of 500,000 yuan, and no debt, etc.; historical transaction records include purchasing low-risk financial products and infrequent stock trading, etc.; behavioral data includes frequently browsing the details pages of financial products and recently searching for high-yield financial products, etc. The system first obtains the above information of user A through the data acquisition module 201, and then performs feature extraction and model construction on this information through the dynamic preference model construction module 202. Assume the static and dynamic features of user A obtained through feature extraction are shown in the following table:

[0082] Based on these characteristics, the dynamic preference model building module 202 constructs a dynamic preference model for user A. Assume the initial parameters of this model are as shown in the table below:

[0083] The system retrieves feature data for all wealth management products from the product pool, including risk level, expected return, investment period, and product type. Assuming there are 5 wealth management products in the pool, their feature data are shown in the table below:

[0084] Product screening module 203 calculates the matching degree of these financial products based on user A's dynamic preference model. The matching degree calculation formula is assumed to be:

[0085] Matching degree = 0.5⋅∣p1−u1∣+0.3⋅∣p2−u2∣+0.2⋅∣p3−u3∣

[0086] Where p1p1 represents the product's risk level, u1u1 represents the user's risk tolerance in the current risk preference model; p2p2 represents the product's expected return, u2u2 represents the user's preference for the expected return; p3p3 represents the product's investment period, and u3u3 represents the user's investment period preference. Based on the above formula, the matching degree of each financial product is calculated as shown in the table below:

[0087] Assuming the preset matching threshold is 0.5, the product screening module 203 will include financial products 1, 3 and 5 with a matching degree higher than 0.5 in the recommendation range and generate a set of financial products.

[0088] The recommendation priority determination module 204 calculates a recommendation score for each financial product based on the characteristics of the financial products in the set and user A's historical behavior data. The recommendation score calculation formula is assumed to be:

[0089] Recommendation Score = 0.4 (Purchase History Preference) + 0.3 (Browsing History Preference) + 0.2 (Feedback History Preference) + 0.1 (Risk Tolerance Match)

[0090] The specific values ​​for purchase history preferences, browsing history preferences, and feedback history preferences can be quantified using user behavior records. Based on the above formula, the recommendation score for each financial product is calculated as shown in the table below:

[0091] The financial products are sorted according to their recommendation scores, and a priority list is generated as shown in the table below:

[0092] The recommendation module 205 generates a recommendation list, adding the top three investment products with the highest recommendation priority, and displays them to user A through user A's terminal device. The user A's terminal device interface includes detailed information such as the investment product's name, risk level, expected return, and investment period, facilitating user A's selection. Simultaneously, the recommendation module 205 is also responsible for recording user A's feedback data on the recommendation list, including click-through rate, purchase rate, and satisfaction level. This feedback data is automatically collected through the user terminal's log system and stored in the system's backend database for subsequent optimization and analysis.

[0093] If certain terms are used in the specification and claims to refer to specific components, those skilled in the art will understand that hardware manufacturers may use different names to refer to the same component. This specification and claims do not distinguish components based on differences in name, but rather on differences in function. The term "comprising" as used throughout the specification and claims is an open-ended term and should be interpreted as "comprising but not limited to." "Approximately" means that within an acceptable margin of error, those skilled in the art can solve the technical problem and substantially achieve the technical effect within a certain margin of error.

[0094] It should be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a product or system comprising a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a product or system. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the product or system that includes that element.

[0095] The foregoing description illustrates and describes several preferred embodiments of the present invention. However, as previously stated, it should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of the inventive concept described herein through the foregoing teachings or techniques or knowledge in related fields. Any modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

Claims

1. A user portrait-based personalized financial wealth management product recommendation method, characterized in that, The method comprises the following steps: obtaining user portrait data of a target user; wherein the user portrait data comprises personal basic information, financial information, historical transaction records and behavior data of the user; constructing a dynamic preference model of the user according to the user portrait data; wherein the dynamic preference model is used to update the preference and demand of the user in real time; screening a set of financial products matched with the dynamic preference model from a product pool; determining a recommendation priority according to the characteristics of the financial products in the set of financial products and the historical behavior data of the user; recommending the financial product with the highest recommendation priority to the target user. 2.The user portrait-based personalized financial wealth management product recommendation method according to claim 1, characterized in that, The step of constructing a dynamic preference model of the user according to the user portrait data comprises the following steps: extracting features from the user portrait data to obtain static features and dynamic features of the user; wherein the static features comprise personal basic information and financial information of the user, and the dynamic features comprise historical transaction records and behavior data of the user; constructing a dynamic preference model of the user based on the static features and dynamic features by using a machine learning algorithm; wherein the dynamic preference model can be updated according to real-time behavior data of the user. 3.The user portrait-based personalized financial wealth management product recommendation method according to claim 1, characterized in that, The step of screening a set of financial products matched with the dynamic preference model from a product pool comprises the following steps: obtaining feature data of all financial products in the product pool; wherein the feature data comprises risk level, expected return, investment period and product type of the product; calculating the matching degree of the feature data according to the preference and demand of the user in the dynamic preference model; adding the financial product with a matching degree higher than a preset threshold to the set of financial products. 4.The user portrait-based personalized financial wealth management product recommendation method according to claim 1, characterized in that, The step of determining a recommendation priority according to the characteristics of the financial products in the set of financial products and the historical behavior data of the user comprises the following steps: calculating a recommendation score for each financial product in the set of financial products based on the historical behavior data of the user and the product feature data; sorting the financial products in the set of financial products according to the recommendation score; taking the sorting result as the recommendation priority. 5.The user portrait-based personalized financial wealth management product recommendation method according to claim 4, characterized in that, The step of calculating a recommendation score comprises the following steps: obtaining historical behavior data of the user, including historical purchase records, browsing records and feedback records of the user; extracting key features in the historical behavior data, including the preference of the user for different product types, risk tolerance and investment period preference; calculating a recommendation score based on the key features and the product feature data by using a preset scoring model; wherein the preset scoring model can be updated according to real-time behavior data of the user. 6.The user portrait-based personalized financial wealth management product recommendation method according to claim 1, characterized in that, The step of recommending the financial product with the highest recommendation priority to the target user comprises the following steps: generating a recommendation list and adding the first N financial products with the highest recommendation priority to the recommendation list; displaying the recommendation list to the target user through a user terminal; recording feedback data of the user on the recommendation list, including click rate, purchase rate and satisfaction degree; optimizing the dynamic preference model and the recommendation algorithm according to the feedback data.

7. A personalized financial wealth management product recommendation system based on user profiling, characterized by, The system comprises a data acquisition module, a dynamic preference model construction module, a product screening module, a recommendation priority determination module and a recommendation module. ​ The data acquisition module is configured to acquire user portrait data of a target user, wherein the user portrait data comprises personal basic information, financial information, historical transaction records and behavior data of the user. The dynamic preference model construction module is configured to construct a dynamic preference model of the user according to the user portrait data, wherein the dynamic preference model is used to update the preference and demand of the user in real time. The product screening module is configured to screen a set of financial products matched with the dynamic preference model from a product pool. The recommendation priority determination module is configured to determine a recommendation priority according to the characteristics of the financial products in the set of financial products and the historical behavior data of the user. The recommendation module is configured to recommend the financial product with the highest recommendation priority to the target user.

8. A computer terminal device, characterized by A computer program product comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, the computer program, when executed by the processor, implements the user portrait-based personalized financial product recommendation method according to any one of claims 1 to 6.

Images