Product recommendation method and device, electronic equipment, storage medium and program product

Abstract

The invention provides a product recommendation method and device, electronic equipment, a storage medium and a program product, and the method comprises the steps: obtaining user association information of a target user, and determining a target customer group corresponding to the target user according to the user association information; determining a plurality of recommendation algorithms and a weight coefficient corresponding to each recommendation algorithm according to the target customer group; processing the user associated information and the product information of the plurality of to-be-recommended products according to each recommendation algorithm to obtain a product candidate recommendation result generated by each recommendation algorithm; according to the weight coefficient corresponding to each recommendation algorithm, fusing the plurality of product candidate recommendation results to generate a product recommendation result containing the plurality of to-be-recommended products; and displaying the product recommendation result to recommend the plurality of to-be-recommended products to the target user. According to the embodiment of the invention, the flexibility and accuracy of product recommendation to the user can be improved, and the user experience is improved.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to a product recommendation method and device, electronic equipment, computer-readable storage medium, and computer program product. Background Art

[0002] Traditional product recommendation methods often rely on manual experience, simple rule matching, or analysis of limited-dimensional data before recommending products to users. These methods struggle to conduct in-depth analysis of massive and complex user and product data, making it difficult to understand users' actual needs and provide personalized recommendations, thus reducing user experience. Summary of the Invention

[0003] The present disclosure provides a product recommendation method and apparatus, an electronic device, a computer-readable storage medium, and a computer program product.

[0004] In a first aspect, the present disclosure provides a product recommendation method, comprising:

[0005] Obtaining user association information of a target user, and determining a target customer group corresponding to the target user based on the user association information;

[0006] Determining multiple recommendation algorithms and weight coefficients corresponding to each recommendation algorithm according to the target customer group;

[0007] Processing the user association information and the product information of the multiple products to be recommended according to each recommendation algorithm respectively to obtain product candidate recommendation results generated by each recommendation algorithm;

[0008] According to the weight coefficient corresponding to each recommendation algorithm, the plurality of product candidate recommendation results are integrated to generate a product recommendation result including the plurality of products to be recommended;

[0009] The product recommendation results are displayed to recommend a plurality of the to-be-recommended products to the target user.

[0010] In a second aspect, the present disclosure provides a product recommendation device, comprising:

[0011] an acquisition module configured to acquire user association information of a target user and determine a target customer group corresponding to the target user based on the user association information;

[0012] a determination module configured to determine a plurality of recommendation algorithms and a weight coefficient corresponding to each recommendation algorithm according to the target customer group;

[0013] a processing module configured to process the user association information and the product information of the plurality of products to be recommended according to each recommendation algorithm, respectively, to obtain a product candidate recommendation result generated by each recommendation algorithm;

[0014] a fusion module configured to fuse the plurality of product candidate recommendation results according to the weight coefficient corresponding to each recommendation algorithm to generate a product recommendation result including the plurality of products to be recommended;

[0015] The display module is configured to display the product recommendation results to recommend a plurality of the to-be-recommended products to the target user.

[0016] In a third aspect, the present disclosure provides an electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores one or more computer programs executable by the at least one processor, and the one or more computer programs are executed by the at least one processor to enable the at least one processor to execute the above-mentioned product recommendation method.

[0017] In a fourth aspect, the present disclosure provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the above-mentioned product recommendation method when executed by a processor.

[0018] In a fifth aspect, the present disclosure provides a computer program product comprising a computer-readable code, or a non-volatile computer-readable storage medium carrying the computer-readable code. When the computer-readable code runs in a processor of an electronic device, the processor in the electronic device executes the above-mentioned product recommendation method.

[0019] The product recommendation method provided by the embodiment of the present disclosure can determine the target customer group to which the target user belongs based on the user-related information of the target user, and determine the corresponding multiple recommendation algorithms and the corresponding weight coefficients of the recommendation algorithms based on the target customer group to which the target user belongs, so as to process the user-related information and the product information of multiple products to be recommended in combination with the multiple recommendation algorithms, obtain a product recommendation result containing multiple products to be recommended, and display the generated product recommendation result to the target user to recommend multiple products to be recommended to the target user. This enables the generation of corresponding product recommendation results based on different recommendation algorithms for users belonging to different customer groups, improves the flexibility and accuracy of product recommendations to users, and enhances the user experience.

[0020] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings are used to provide a further understanding of the present disclosure and constitute a part of the specification. Together with the embodiments of the present disclosure, they are used to explain the present disclosure and do not constitute a limitation of the present disclosure. The above and other features and advantages will become more apparent to those skilled in the art by describing detailed example embodiments with reference to the accompanying drawings. In the accompanying drawings:

[0022] Figure 1 is a flowchart of a product recommendation method provided by an embodiment of the present disclosure;

[0023] Figure 2 is a schematic diagram of determining a recommendation algorithm and weight coefficients provided by an embodiment of the present disclosure;

[0024] Figure 3 This is a flowchart of screening products to be recommended provided by an embodiment of the present disclosure;

[0025] Figure 4 is a processing flow chart of a product recommendation method provided by an embodiment of the present disclosure;

[0026] Figure 5 is a block diagram of a product recommendation device provided by an embodiment of the present disclosure;

[0027] Figure 6 It is a block diagram of an electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0028] To enable those skilled in the art to better understand the technical solutions of the present disclosure, exemplary embodiments of the present disclosure are described below in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered merely exemplary. Therefore, those skilled in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0029] In the absence of conflict, the various embodiments of the present disclosure and the various features therein may be combined with each other.

[0030] As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0031] The terms used herein are only used to describe specific embodiments and are not intended to limit the present disclosure. As used herein, the singular forms "a" and "the" are also intended to include the plural forms, unless the context clearly indicates otherwise. It will also be understood that when the terms "comprising" and / or "made of" are used in this specification, the presence of the features, wholes, steps, operations, elements and / or components is specified, but the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or groups thereof is not excluded. Similar words such as "connected" or "connected" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect.

[0032] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art. It will also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant art and the present disclosure, and will not be interpreted as having an idealized or overly formal meaning unless expressly defined as such herein.

[0033] In the technical solutions disclosed herein, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals. The use of user data in this technical solution complies with relevant national laws and regulations (for example, the "Information Security Technology Personal Information Security Specification", etc.). For example: corresponding prescribed measures are taken to control access to personal information; the display of personal information is subject to prescribed restrictions; the purpose of using personal information does not exceed the scope of direct or reasonable connection; when using personal information, clear identity reference is eliminated to avoid precise positioning of specific individuals.

[0034] With the continuous accumulation of household wealth and the increasing diversification of investment channels, the number of individual investors continues to grow, and their investment needs are becoming increasingly complex and diverse. Users are no longer satisfied with standardized product recommendations that are one-size-fits-all, but instead desire personalized recommendations that precisely match their investment goals, risk tolerance, investment horizon, and market perspectives. However, the traditional product recommendation methods currently used in the industry mostly rely on manual experience, simple rule matching, or analysis based on limited-dimensional data. These methods struggle to comprehensively and deeply mine and analyze massive and complex customer data, product data, and rapidly changing market data. Consequently, they are unable to accurately understand users' true needs, nor can they provide personalized product recommendation services that align with market dynamics. The limitations of traditional methods are becoming increasingly prominent in an increasingly competitive market environment, and have become a key factor hindering the improvement of user service quality and market competitiveness.

[0035] Therefore, it is crucial to provide personalized product recommendations to users through algorithmic models and provide customized product recommendation services based on users' personalized goals, preferences and risks.

[0036] Based on this, the embodiments of the present disclosure provide a product recommendation method, a product recommendation device, an electronic device, a computer-readable storage medium, and a computer program product, which are described in detail one by one in the following embodiments.

[0037] The product recommendation method according to the embodiment of the present disclosure can be executed by an electronic device such as a terminal device or a server. The terminal device can be a user equipment (UE), a mobile device, a user terminal, a terminal, a cellular phone, a cordless phone, a personal digital assistant (PDA), a handheld device, a computing device, a vehicle-mounted device, a wearable device, etc. The server can be an independent physical server, a server cluster composed of multiple physical servers, or a cloud server capable of cloud computing. The method can be implemented by a processor calling computer-readable program instructions stored in a memory.

[0038] See also Figure 1 , Figure 1 A flowchart of a product recommendation method according to an embodiment of the present disclosure is shown, which specifically includes the following steps:

[0039] Step 102: Obtain user association information of a target user, and determine a target customer group corresponding to the target user based on the user association information.

[0040] The target user refers to any user for whom a product recommendation is intended. User-related information specifically includes user attribute information and user-related information. User attribute information includes the user's name, age, occupation, marital status, and other information. User-related information includes the user's investment preferences, investment capabilities, investment expectations, risk tolerance, and holdings. The target customer group refers to the customer group to which the target user belongs.

[0041] In practical applications, different users have different investment preferences, risk tolerance, and investment capabilities. Therefore, users can be categorized based on their user association information to obtain different types of customer groups. To improve the accuracy and personalization of product recommendations for different users, the disclosed embodiments obtain the target user's user association information before making product recommendations to the target user, and determine the target customer group to which the target user belongs based on this user association information. This allows targeted product recommendation services to be provided to users in different customer groups in the subsequent process, thereby improving the personalization of product recommendations.

[0042] Step 104: Determine multiple recommendation algorithms and weight coefficients corresponding to each recommendation algorithm according to the target customer group.

[0043] Because user data is massive and complex in real-world applications, the disclosed embodiments utilize multiple recommendation algorithms for subsequent processing to achieve comprehensive and in-depth analysis of user data. Therefore, after determining the target customer group corresponding to the target user, it is necessary to further determine the multiple recommendation algorithms corresponding to the target customer group, as well as the weighting coefficients corresponding to each recommendation algorithm.

[0044] Among them, the recommendation algorithm refers to the algorithm used to provide product recommendation services to users in the target customer group. The weight coefficient is used to characterize the degree of influence of the product candidate recommendation results generated by each recommendation algorithm on the generated product recommendation results. The product candidate recommendation result refers to the recommendation result generated by each recommendation algorithm; the product recommendation result refers to the final recommendation result generated by integrating the product candidate recommendation results generated by multiple recommendation algorithms. The product candidate recommendation results and product recommendation results are specifically the recommendation ranking results of multiple products to be recommended. The product to be recommended refers to the product that needs to be recommended to the target user. In actual applications, the number of recommendation algorithms can also be one, which is determined according to the actual application situation.

[0045] See also Figure 2 , Figure 2 A schematic diagram of determining a recommendation algorithm and weight coefficient according to an embodiment of the present disclosure is shown. Taking the recommendation algorithms including collaborative filtering algorithm, DSSM twin tower model (Deep Structured Semantic Model), DeepFM model (Deep Factorization Machine), large model (LLM), top hot selling rule and marketing rule as examples, the implementation method of determining the recommendation algorithm and weight coefficient corresponding to the target user is explained. Figure 2 As shown, different users are divided into n customer groups. The recommendation algorithms corresponding to customer group 1 include collaborative filtering algorithm and marketing rules. The recommendation algorithms corresponding to customer group 2 include DSSM twin tower model, DeepFM model and large model. The recommendation algorithms corresponding to customer group 3 include DSSM twin tower model, Top hot selling rules and marketing rules, etc. The recommendation algorithms corresponding to other customer groups are not listed here in this embodiment of the disclosure. In the embodiment of the disclosure, each recommendation algorithm has a different weight coefficient, and the weight coefficients of the same recommendation algorithm corresponding to different customer groups may be different. Figure 2, the collaborative filtering algorithm only corresponds to customer group 1, and its corresponding weight coefficient is 0.6; the customer groups corresponding to the DSSM twin-tower model include customer groups 2 and 3. When the DSSM twin-tower model provides calculations for users in customer group 2, its corresponding weight coefficient is 0.5, and when the DSSM twin-tower model provides calculations for users in customer group 3, its corresponding weight coefficient is 0.6; the DeepFM model only corresponds to customer group 2, and its corresponding weight coefficient is 0.3. The weight coefficients corresponding to other recommendation algorithms for different customer groups are as follows: Figure 2 As shown, the embodiments of the present disclosure are not listed here one by one. In actual application, after determining the target customer group to which the target user belongs, the corresponding recommendation algorithm and the weight coefficient corresponding to each recommendation algorithm can be determined according to the target customer group. For example, the target customer group corresponding to the target user is customer group 1, and it can be determined that the recommendation algorithm includes a collaborative filtering algorithm and a marketing rule, where the weight coefficient of the collaborative filtering algorithm is 0.6 and the weight coefficient of the marketing rule is 0.4.

[0046] The embodiment of the present disclosure can determine multiple recommendation algorithms and the weight coefficient corresponding to each recommendation algorithm for subsequent calculations based on the target customer group to which the target user belongs, so as to implement subsequent processing using different recommendation algorithms for different users, thereby improving the accuracy and flexibility of the subsequently generated product recommendation results.

[0047] Step 106: Process the user association information and the product information of the multiple products to be recommended according to each recommendation algorithm to obtain the product candidate recommendation results generated by each recommendation algorithm.

[0048] After determining the recommendation algorithm to be applied and the weight coefficient corresponding to each recommendation algorithm, the user association information and the product information of the products to be recommended to the target user can be processed according to each recommendation algorithm to obtain the product candidate recommendation results generated by each recommendation algorithm.

[0049] The product information refers to the product-related information of the product to be recommended, including but not limited to the product name, product type, investment target, investment style, etc. Before processing the user-related information and the product to be recommended based on the recommendation algorithm, it is necessary to first determine the product to be recommended to the target user.

[0050] In a specific embodiment provided by the present disclosure, the method further comprises:

[0051] Determining a user risk preference level of the target user based on the user association information;

[0052] Screening a plurality of candidate recommended products from a pool of recommended products according to the user's risk preference level, and determining a target allocation ratio corresponding to the target user according to the user's risk preference level;

[0053] According to the target configuration ratio, a plurality of products to be recommended are screened from the plurality of candidate recommended products.

[0054] The user's risk preference level refers to the user's risk tolerance and risk preference assessment level. The recommended product pool includes multiple products that have been listed and can be recommended to users. Candidate recommended products refer to products selected from the recommended product pool based on the target user's risk preference level. The target allocation ratio refers to the long-term target allocation ratio of each major resource category (such as stocks, bonds, cash, commodities, etc.) in the investment portfolio. It is a benchmark ratio set for different resource categories based on factors such as the user's risk tolerance, investment goals, and long-term market expectations.

[0055] Specifically, after obtaining the user association information of the target user, the user risk preference level of the target user can be determined based on the user association information, and multiple candidate recommended products can be screened out from the recommendable product pool according to the user risk preference level of the target user, and the target configuration ratio of the target user can be determined according to the user risk preference level of the target user, so as to screen out the recommended products that meet the target configuration ratio of the target user from multiple candidate recommended products according to the target configuration ratio.

[0056] In the disclosed embodiment, the user's risk preference level can be divided into five levels: C1, C2, C3, C4, and C5. C1 can be used to represent a conservative level, C2 can be used to represent a cautious level, C3 can be used to represent a stable level, C4 can be used to represent an active level, and C can be used to represent an aggressive level. To improve the accuracy of the assessment of the user's risk preference level, the user's risk preference level can be adjusted based on the user's age, investment preferences, KYC (Know Your Customer) response results, etc.

[0057] In actual applications, the user's risk preference level can be adjusted according to the user's age. For details, please refer to Table 1 below:

[0058] Table 1

[0059] Age of initial risk level adjustment ≤30 years old (30-40 years old) (40-50 years old) (50-60 years old) >60 years old C1 (conservative) C1 C1 C1 C1 C1 C2 (cautious) C2 C2 C2 C2 C2 C3 (robust type) C3 C3 C3 C3 C3 C4 (active) C4 C4 C4 C4 C3↓ C5 (radical) C5 C5 C4↓ C4↓ C3↓

[0060] As shown in Table 1 above, based on the impact of age on investment, the user risk preference level of users aged 40-50 and 50-60 can be lowered from C5 to C4, and the user risk preference level of users over 60 years old can be lowered from C4 and C5 to C3.

[0061] After adjusting the user's risk preference level based on the user's age, the user's risk preference level is further adjusted based on the user's investment preference. For details, please refer to Table 2 below:

[0062] Table 2

[0063]

[0064] Since a higher proportion of a user's equity resources often indicates a higher risk appetite, while a higher proportion of low-yield resources like currency and bonds often indicates a lower risk appetite, we can adjust the risk appetite of some heavy and equity-focused investors upward, while lowering the risk appetite of some bond and currency-focused investors, as shown in Table 2 above.

[0065] Furthermore, the user's risk preference level can be adjusted based on the user's KYC answer results. The specific adjustment can be based on actual application conditions, and the present embodiment is not limited here. It should be noted that the above description only uses C1-C5 as an example. The user risk preference level is not limited to the above five levels and can be determined based on actual application conditions.

[0066] After determining the target user's risk preference level based on their user-related information, multiple candidate recommended products can be screened from the pool of recommended products based on their risk preference level. The following explains the specific implementation method for screening multiple candidate recommended products.

[0067] In a specific embodiment provided by the present disclosure, a plurality of candidate recommended products are screened from a pool of recommended products according to the user's risk preference level, including:

[0068] Obtaining the product risk level corresponding to each recommended product in the recommended product pool;

[0069] Determine a recommendable product whose product risk level is less than or equal to the user's risk preference level as a first candidate recommended product;

[0070] Based on the user association information, remove the target user's owned products and customized products from the plurality of first candidate recommended products to obtain a plurality of second candidate recommended products;

[0071] A plurality of the second candidate recommendation products are determined as a plurality of candidate recommendation products.

[0072] Among them, already-held products refer to products that have been signed by target users; customized products include regional customized products and user customized products. Regional customized products refer to products that are signed according to regions, and user customized products refer to products that are customized exclusively for users.

[0073] Specifically, in the disclosed embodiment, the product risk levels of the recommended products can be divided into 5 levels, namely R1, R2, R3, R4 and R5. In actual applications, the user can be determined to purchase the contracted products based on the user risk preference level and the product risk level. For example, if the user risk preference level of user A is C5, then user A can purchase any product with a product risk level of R1-R5. If the user risk preference level of user B is C4, then user B can purchase any product with a product risk level of R1-R4, but cannot purchase a product with a product risk level of R5. It should be noted that the above description only takes R1-R5 as an example. The product risk level is not limited to the above 5 levels and can be determined based on the actual application situation.

[0074] Based on this, the product risk level corresponding to each recommendable product in the recommendable product pool is obtained, and the first candidate recommended products that the target user can purchase are screened from the recommendable products based on the product risk level and the user risk preference level of the target user. Specifically, the recommendable products whose product risk level is less than or equal to the user risk preference level can be screened from the recommendable product pool, and the screened recommendable products are determined as the first candidate recommended products. Furthermore, the target user's already held products and customized products are screened from the multiple first candidate recommended products, and the already held products and customized products are eliminated from the first candidate recommended products to obtain multiple second candidate recommended products, and the multiple second candidate recommended products are determined as multiple candidate recommended products.

[0075] In actual applications, it is also necessary to screen the recommended products that meet the target configuration ratio of the target user among multiple candidate recommended products based on the target configuration ratio of the target user. Therefore, after determining multiple candidate recommended products, it is necessary to further determine the target configuration ratio of the target user so as to screen the recommended products based on the target configuration ratio in the subsequent process.

[0076] In the embodiment of the present disclosure, the target allocation ratio of the target user can be determined according to the user risk preference level and the A-share temperature of the target user, as follows:

[0077] Continuing with the example of user risk preference levels C1-C5, if the user risk preference level is C1, no specific rules are set to determine the corresponding target allocation ratio. If the user risk preference level is C2, the target allocation ratio for the target user can be calculated according to the following formula 1.

[0078] Target allocation ratio = 30% * (100-X) / 100 Formula 1

[0079] Where X is the temperature of A-shares. If the user's risk preference level is C3, the target allocation ratio of the target user can be calculated according to the following formula 2.

[0080] Target allocation ratio = 30% + 30% * (100-X) / 100 Formula 2

[0081] When the user risk preference level is C4, the target configuration ratio of the target user can be calculated according to the following formula 3.

[0082] Target allocation ratio = 60% + 40% * (100-X) / 100 Formula 3

[0083] When the user risk preference level is C5, the target allocation ratio of the target user can be calculated according to the following formula 4.

[0084] Target allocation ratio = 40% * (100-X) / 100 Formula 4

[0085] After determining the target allocation ratio of the target user according to the user's risk preference level, the recommended products that can be recommended to the target user can be further screened from multiple candidate recommendation products based on the target allocation ratio.

[0086] Based on this, in a specific embodiment provided by the present disclosure, a plurality of products to be recommended are screened from a plurality of candidate recommended products according to the target configuration ratio, including:

[0087] Obtaining the product configuration ratio corresponding to each candidate recommended product;

[0088] The candidate recommended products whose product configuration ratio is less than or equal to the target configuration ratio are determined as the products to be recommended.

[0089] Specifically, after determining multiple candidate recommended products, the product configuration ratio of each candidate recommended product can be obtained, and the products to be recommended can be further screened based on the target configuration ratio of the determined target user. Specifically, the candidate recommended products whose product configuration ratio is less than or equal to the target configuration ratio can be determined as the products to be recommended.

[0090] Furthermore, the following Figure 3 This section describes in detail how to filter products to be recommended from the recommended product pool. Figure 3 , Figure 3 FIG. 1 shows a flow chart of screening products to be recommended according to an embodiment of the present disclosure. Figure 3As shown, based on the obtained user association information of the target user, the user risk preference level of the target user is determined, and the product risk level corresponding to each recommendable product in the recommendable product pool is obtained. According to the user risk preference level and the product risk level, the recommendable product whose product risk level is less than or equal to the user risk preference level is determined as the first candidate recommended product. Furthermore, the target user's already held products and customized products (including regional customized products and user customized products) are eliminated from the multiple first candidate recommended products to obtain multiple second candidate recommended products. The target configuration ratio of the target user is determined based on the user risk preference level and the A-share temperature, and the second candidate product whose product configuration ratio is equal to the target configuration ratio is determined as the product to be recommended to the target user.

[0091] The embodiment of the present disclosure screens the recommendable products in the recommendable product pool according to the user's risk preference level and the target configuration ratio respectively, so as to screen out the recommended products that are more suitable for the target user, so that in the subsequent process of recommending to the target user based on the screened recommended products, the user experience of the target user is improved and personalized product recommendations are achieved.

[0092] After determining the products to be recommended to the target user, the user association information and the product information of the products to be recommended can be processed according to the determined multiple recommendation algorithms to obtain product candidate results provided by each recommendation algorithm.

[0093] In a specific embodiment provided by the present disclosure, the recommendation algorithm includes a first recommendation model;

[0094] The user association information and the product information of the multiple products to be recommended are processed according to each recommendation algorithm to obtain the product candidate recommendation results generated by each recommendation algorithm, including:

[0095] Inputting the user association information and the product information of the plurality of products to be recommended into the first recommendation model to obtain a first matching degree between the user association information and each of the products to be recommended;

[0096] A first matching degree between the user association information and each of the to-be-recommended products is determined as a first product candidate recommendation result generated by the first recommendation model.

[0097] The first recommendation model is specifically a model with a small number of model parameters, such as the DSSM twin-tower model, DeepFM model, DNN model (Deep Neural Network), LR model (Logistic Regression), etc. The first matching degree is the matching degree between the user association information output by the first recommendation model and each product to be recommended.

[0098] Specifically, user association information and product information of multiple products to be recommended are input into the first recommendation model to obtain the first matching degree between the user association information output by the first recommendation model and each product to be recommended. The first matching degree between the user association information and each product to be recommended is the first product candidate recommendation result generated by the first recommendation model.

[0099] Taking the DSSM dual-tower model as an example, the first recommendation model consists of two independent deep neural networks: the User Tower and the Item Tower. In actual applications, user association information is input into the User Tower, and product information of multiple products to be recommended is input into the Product Tower. The User Tower and the Product Tower respectively extract high-order features through a multi-layer fully connected network (MLP) or a feature cross layer (such as FM or CrossNet) to generate a low-dimensional dense vector (such as 128 dimensions). See the following formulas 5 and 6:

[0100] UserEmbedding=f user (X user ) Formula 5

[0101] ItemEmbedding=f item (X item ) Formula 6

[0102] The cosine similarity is then used to measure the matching degree between the user and the product, i.e., the first matching degree, as shown in the following formula 7:

[0103]

[0104] Since the first recommendation model is the basis for generating the first product candidate recommendation result, before processing the user association information and the product to be recommended based on the first recommendation model, the first recommendation model needs to be trained first to generate the first product candidate recommendation result using the trained first recommendation model.

[0105] Based on this, in a specific embodiment provided by the present disclosure, the first recommendation model is trained based on the following method:

[0106] Acquire sample user association data and sample product data, as well as sample tags between the sample user association data and the sample product data;

[0107] Inputting the sample user association data and the sample product data into the first recommendation model to obtain a predicted matching degree between the sample user association data and the sample product data;

[0108] Calculating a loss value of the first recommendation model according to the sample label and the predicted matching degree;

[0109] The first recommendation model is trained according to the loss value.

[0110] The sample user-related data and sample product data are sample training data used to train the first recommendation model. The sample label refers to the actual label of the sample training data, which is used to indicate whether the user has signed up for a product. The predicted match degree refers to the match degree between the sample user-related data and the sample product data output by the first recommendation model.

[0111] Specifically, sample user association data and sample product data for training the first recommendation model, as well as sample labels between the sample user association data and the sample product data, are obtained, the sample user association data and the sample product data are input into the first recommendation model, and the predicted matching degree between the sample user association data and the sample product data output by the first recommendation model is obtained. Since the first recommendation model at this time is an untrained model, the model parameters of the first recommendation model need to be adjusted. Therefore, the loss value of the first recommendation model is calculated based on the sample labels and the predicted matching degree, and the model parameters of the first recommendation model are adjusted based on the loss value, and the first recommendation model is continued to be trained until the model training stop condition is reached. In an embodiment of the present disclosure, the loss function used to calculate the loss value of the first recommendation model may be a cross entropy loss function, a square loss function, etc. The present disclosure prefers the cross entropy loss function as the loss function used to calculate the loss value. Based on this, the specific method for calculating the loss value can be found in the following formula 8:

[0112]

[0113] Where L is the loss value of the first recommendation model; y is the sample label, which can be 0 or 1. 0 is used to indicate that the user has not signed a contract for a product, and 1 is used to indicate that the user customer has signed a contract for a product. is the predicted matching degree. When y=1, if If it is close to 1, the loss is small; on the contrary, if If it is close to 0, the loss is large. Similarly, when y=0, if If it is close to 0, the loss is small.

[0114] The embodiment of the present disclosure can improve the accuracy of the first recommendation model in subsequent applications by training the first recommendation model based on sample user association data and sample product data.

[0115] Furthermore, in order to improve the efficiency and interpretability of training the first recommendation model, the embodiment of the present disclosure may also optimize the sample training data used to train the first recommendation model before training the first recommendation model to obtain more representative information.

[0116] In a specific embodiment provided by the present disclosure, obtaining sample user association data and sample product data includes:

[0117] Obtaining a first user dataset and a first product dataset;

[0118] Respectively removing abnormal data from the first user dataset and the first product dataset to obtain a second user dataset and a second product dataset;

[0119] extracting user features from the second user dataset and product features from the second product dataset;

[0120] combining multiple user features in the second user data set to obtain a composite user feature, and combining multiple product features in the second product data set to obtain a composite product feature;

[0121] The composite user feature is determined as sample user association data, and the composite product feature is determined as sample product data.

[0122] The first user dataset refers to the directly accessible original user dataset; the first product dataset refers to the directly accessible original product dataset. The second user dataset refers to the user dataset obtained by removing abnormal data from the first user dataset; the second product dataset refers to the product dataset obtained by removing abnormal data from the first product dataset. Composite user features refer to new user features obtained by combining multiple original user features; composite product features refer to new product features obtained by combining multiple original product features.

[0123] Specifically, a first user dataset and a first product dataset are obtained. The first user dataset can be obtained by processing intelligent return visit data and chat record data with users through a large model; the first product dataset can be obtained by analyzing product background information, such as product configuration information, product investment style, and other data through a large model. Accordingly, the first user dataset and the first product dataset can also be obtained through manual labeling. After obtaining the first user dataset and the first product dataset, data cleaning is performed on the first user dataset and the first product dataset, specifically by removing abnormal data from the first user dataset and the first product dataset, verifying the legitimacy of the data in the first user dataset and the first product dataset, modifying the field format of the data in the dataset, etc. Missing data can also be filled to ensure data integrity, thereby obtaining a second user dataset and a second product dataset.

[0124] In order to solve the data imbalance phenomenon that exists in the process of training the first recommendation model, the second product data set can be clustered according to feature similarity, and after clustering, the smaller number of clusters in the same category can be excluded so that they do not participate in model training, thereby reducing the interference of data imbalance.

[0125] Furthermore, user features in the second user data set and product features in the second product data set are extracted, and the extracted multiple user features are combined to obtain composite user features, and the multiple product features are combined to obtain composite product features. Specifically, the combination can be based on principal component analysis technology (Principal Component Analysis, PCA), and the combined composite user features are determined as sample user association data, and the composite product features are determined as sample product data, so that the sample user association data and the sample product data are used to train the first recommendation model.

[0126] In practical applications, in order to improve the efficient operation of the model, after extracting user features and product features, the user features and product features can also be evaluated based on the random forest algorithm to select features with higher importance from multiple user features and multiple product features, and the selected features are numerically standardized and normalized before feature combination to obtain composite user features and composite product features.

[0127] Furthermore, in a specific embodiment provided by the present disclosure, the recommendation algorithm includes a second recommendation model, and the model parameter amount of the second recommendation model is greater than the model parameter amount of the first recommendation model;

[0128] The user association information and the product information of the multiple products to be recommended are processed according to each recommendation algorithm to obtain the product candidate recommendation results generated by each recommendation algorithm, including:

[0129] Inputting the user association information and the product information of the plurality of products to be recommended into the second recommendation model to obtain a second matching degree between the user association information and each of the products to be recommended;

[0130] A second matching degree between the user association information and each of the to-be-recommended products is determined as a second product candidate recommendation result generated by the second recommendation model.

[0131] The second recommendation model is specifically a model with a larger number of model parameters, and the number of model parameters of the second recommendation model is greater than that of the first recommendation model. The second recommendation model can specifically be a large model, such as an LLM model. The second matching degree is the matching degree between the user association information output by the second recommendation model and each recommended product.

[0132] Specifically, user association information and product information of multiple products to be recommended are input into the second recommendation model to obtain the second matching degree between the user association information output by the second recommendation model and each product to be recommended. The second matching degree between the user association information and each product to be recommended is the second product candidate recommendation result generated by the second recommendation model.

[0133] The disclosed embodiment is based on the RAG (Retrieval-Augmented Generation) framework to dynamically introduce an external knowledge base to enhance the generation capability of the second recommendation model. By allowing the second recommendation model to access and learn the fund investment advisor's knowledge base (including investment advisor product knowledge, internal system knowledge, investment knowledge, investment advisor strategy knowledge, operational guidance knowledge, etc.), the fund investment advisor business background knowledge is synchronized. Then, the user-related information and the product information of the product to be recommended are converted into natural language form, and through the prompt word engineering of the thinking chain CoT method, the second recommendation model is guided to analyze the user-related information and product information to obtain the second product candidate recommendation result.

[0134] In addition, the embodiment of the present disclosure uses the historical records of users purchasing products as training data for training the second recommendation model, and implements the training of the second recommendation model by performing targeted fine-tuning and then evaluating the second recommendation model.

[0135] After determining multiple recommendation algorithms for processing user association information and product information of products to be recommended, the embodiment of the present disclosure can process the user association information and product information of products to be recommended based on each recommendation algorithm respectively to obtain product candidate recommendation results generated by each recommendation algorithm.

[0136] Step 108: Based on the weight coefficient corresponding to each recommendation algorithm, the plurality of product candidate recommendation results are integrated to generate a product recommendation result including a plurality of the products to be recommended.

[0137] After obtaining the product candidate recommendation results generated by each recommendation algorithm, multiple product candidate recommendation results can be fused according to the weight coefficient corresponding to each recommendation algorithm to generate a fused product recommendation result containing multiple products to be recommended.

[0138] In a specific embodiment provided by the present disclosure, a plurality of the product candidate recommendation results are fused according to the weight coefficient corresponding to each recommendation algorithm to generate a product recommendation result containing a plurality of the products to be recommended, including:

[0139] The first product candidate recommendation result and the second product candidate recommendation result are fused according to a first weight coefficient corresponding to the first recommendation model and a second weight coefficient corresponding to the second recommendation model to generate the product recommendation result.

[0140] The first weight coefficient is the weight coefficient corresponding to the first recommendation model, and the second weight coefficient is the weight coefficient corresponding to the second recommendation model.

[0141] Specifically, according to the first weight coefficient corresponding to the first recommendation model and the second weight coefficient corresponding to the second recommendation model, a weighted calculation is performed on the first product candidate recommendation result generated by the first recommendation model and the second product candidate recommendation result generated by the second recommendation model to achieve the fusion of the first product candidate recommendation result and the second product candidate recommendation result to obtain the fused product recommendation result.

[0142] The disclosed embodiment performs weighted calculation on multiple product candidate recommendation results through the weight coefficient corresponding to each recommendation algorithm, thereby integrating multiple product candidate recommendation results into one product recommendation result, thereby improving the accuracy and flexibility of the generated product recommendation results.

[0143] Step 110: Display the product recommendation results to recommend a plurality of the to-be-recommended products to the target user.

[0144] After generating product recommendations, the generated product recommendations can be displayed on the target user's terminal, recommending multiple products to the target user. To enhance user trust and acceptance of the product recommendations, the second recommendation model can also generate a reason for the product recommendation and display it to the target user along with the product recommendation.

[0145] In actual applications, due to differences in marketing needs or limitations of business rules, the generated product recommendation results can also be adjusted accordingly to meet marketing needs or business rules.

[0146] Based on this, in a specific embodiment provided by the present disclosure, after generating a product recommendation result including a plurality of products to be recommended, the method further includes:

[0147] Adjust the product recommendation result based on preset business rules to generate an adjusted product recommendation result;

[0148] Display the product recommendation results, including:

[0149] The adjusted product recommendation results are displayed.

[0150] Among them, preset business rules refer to pre-set business rules. For example, preset business rules can be recommending the top 10 best-selling products, recommending key marketing products, etc.

[0151] Specifically, the product recommendation results can be adjusted according to preset business rules to generate adjusted product recommendation results. For example, the recommendation order of multiple products to be recommended can be rearranged according to the preset business rules to obtain a rearranged recommendation order. After the adjustment of the product recommendation results is completed, the adjusted product recommendation results are displayed to the target user.

[0152] Furthermore, the embodiments of the present disclosure may also evaluate the product recommendation results. For example, the subsequent progress of the product recommendation results may be tracked so as to more effectively match or recommend products to users.

[0153] Based on this, in a specific embodiment provided by the present disclosure, the method further includes:

[0154] Determining an evaluation indicator based on the product recommendation result;

[0155] The first recommendation model or the second recommendation model is adjusted based on the evaluation indicator.

[0156] Among them, the evaluation metrics include but are not limited to accuracy, ROC (Receiver Operating Characteristic Curve), AUC (Area Under the Curve), etc. The evaluation metrics can be specifically determined according to the signing situation of the target users with the product. Specifically, based on the product recommendation results and the signing situation of the users with the product, the evaluation metrics are determined, and based on the evaluation metrics, the first recommendation model or the second recommendation model is adjusted. Taking the evaluation metric of AUC as an example, AUC is the area enclosed by the ROC curve and the coordinate axes and is an important metric for measuring the classification ability of the model. The value range of AUC is between 0 and 1.

[0157] When AUC = 1, it indicates that the model has a high classification ability and can accurately distinguish positive and negative class samples in most cases; when 0.5 < AUC < 1, it indicates that the model has a certain classification ability, and the larger the AUC value, the better the performance of the model; when AUC = 0.5, it indicates that the classification ability of the model is the same as random guessing, that is, the model cannot effectively distinguish positive and negative class samples; when AUC < 0.5, it indicates that the classification ability of the model is worse than random guessing, and usually in this case, the model needs to be readjusted or optimized.

[0158] In addition, the business rules, sample labels, etc. can also be adjusted based on the evaluation metrics, which can be specifically determined according to the actual application situation.

[0159] The product recommendation method provided by this disclosure includes: obtaining the user association information of the target user, and determining the target customer group corresponding to the target user according to the user association information; determining multiple recommendation algorithms and the weight coefficient corresponding to each recommendation algorithm according to the target customer group; respectively processing the user association information and the product information of multiple to-be-recommended products according to each recommendation algorithm to obtain the product candidate recommendation results generated by each recommendation algorithm; fusing multiple product candidate recommendation results according to the weight coefficient corresponding to each recommendation algorithm to generate a product recommendation result including multiple to-be-recommended products; and displaying the product recommendation result to recommend multiple to-be-recommended products to the target user.

[0160] The disclosed embodiments achieve the ability to determine the target customer group to which the target user belongs based on the user-related information of the target user, and determine the corresponding multiple recommendation algorithms and weight coefficients corresponding to the recommendation algorithms based on the target customer group to which the target user belongs, so as to process the user-related information and the product information of multiple products to be recommended in combination with the multiple recommendation algorithms, obtain a product recommendation result containing multiple products to be recommended, and display the generated product recommendation result to the target user to recommend multiple products to the target user. This achieves the ability to generate corresponding product recommendation results based on different recommendation algorithms for users belonging to different customer groups, thereby improving the flexibility and accuracy of product recommendations to users and enhancing the user experience.

[0161] The following combination Figure 4 Taking the multiple recommendation algorithms determined to be the first recommendation model and the second recommendation model as an example, the product recommendation method provided by the embodiment of the present disclosure is further explained. Figure 4 A process flow chart of a product recommendation method according to an embodiment of the present disclosure is shown. Figure 4 As shown, after obtaining the user association information of the target user, the target customer group to which the target user belongs is determined based on the user association information, and a plurality of recommendation algorithms and the weight coefficients corresponding to each recommendation algorithm are determined based on the target customer group. Based on the user association information, the user risk preference level of the target user is determined, and the product risk level corresponding to each recommendable product in the recommendable product pool is obtained. According to the user risk preference level and the product risk level, the recommendable products whose product risk level is less than or equal to the user risk preference level are determined as the first candidate recommended products. Furthermore, the target user's already held products and customized products (including regional customized products and user customized products) are eliminated from the multiple first candidate recommended products to obtain multiple second candidate recommended products. The target configuration ratio of the target user is determined based on the user risk preference level and the A-share temperature, and the second candidate product whose product configuration ratio is equal to the target configuration ratio is determined as the product to be recommended to the target user.

[0162] Input the user association information and the product information of multiple products to be recommended into the first recommendation model. For example, the first recommendation model can be a DSSM dual-tower model, a DeepFM model, a DNN model, an LR model, etc., to obtain the first product candidate recommendation result generated by the first recommendation model; input the user association information and the product information of multiple products to be recommended into the second recommendation model, for example, a large model, and combine the second recommendation model with the RAG framework, and guide the second recommendation model to analyze the user association information and the product information of multiple products to be recommended through the prompt word engineering of the thinking chain CoT method, to obtain the second product candidate recommendation result generated by the second recommendation model. According to the first weight coefficient corresponding to the first recommendation model and the second weight coefficient of the second recommendation model, the first product candidate recommendation result and the second product candidate recommendation result are weightedly calculated to achieve the fusion of the first product candidate recommendation result and the second product candidate recommendation result, and obtain the fused product recommendation result, which includes the product ranking results of multiple products to be recommended.

[0163] Furthermore, the product recommendation results can be adjusted according to the preset business rules based on marketing needs, and the product recommendation results can be rearranged to meet the requirements of diversity and dispersion. In addition, the second recommendation model can also generate recommendation reasons for the product recommendation results, and then combine the product recommendation results and recommendation reasons (such as Figure 4 shown) are displayed to the target user together.

[0164] The disclosed embodiments achieve the ability to determine the target customer group to which the target user belongs based on the user-related information of the target user, and determine the corresponding multiple recommendation algorithms and weight coefficients corresponding to the recommendation algorithms based on the target customer group to which the target user belongs, so as to process the user-related information and the product information of multiple products to be recommended in combination with the multiple recommendation algorithms, obtain a product recommendation result containing multiple products to be recommended, and display the generated product recommendation result to the target user to recommend multiple products to the target user. This achieves the ability to generate corresponding product recommendation results based on different recommendation algorithms for users belonging to different customer groups, thereby improving the flexibility and accuracy of product recommendations to users and enhancing the user experience.

[0165] It is understood that the above-mentioned various method embodiments mentioned in this disclosure can be combined with each other to form combined embodiments without violating the principle logic. Due to space limitations, this disclosure will not go into details. It is understood by those skilled in the art that in the above-mentioned methods of specific implementation, the specific execution order of each step should be determined by its function and possible internal logic.

[0166] In addition, the present disclosure also provides a product recommendation device, an electronic device, a computer-readable storage medium, and a computer program product, all of which can be used to implement any product recommendation method provided by the present disclosure. The corresponding technical solutions and descriptions are referred to the corresponding records in the method section and will not be repeated here.

[0167] Figure 5 FIG2 shows a block diagram of a product recommendation device according to an embodiment of the present disclosure. Figure 5 , an embodiment of the present disclosure provides a product recommendation device, the product recommendation device comprising:

[0168] The acquisition module 502 is configured to acquire user association information of a target user and determine a target customer group corresponding to the target user based on the user association information;

[0169] A determination module 504 is configured to determine multiple recommendation algorithms and a weight coefficient corresponding to each recommendation algorithm according to the target customer group;

[0170] The processing module 506 is configured to process the user association information and the product information of the multiple products to be recommended according to each recommendation algorithm, and obtain the product candidate recommendation results generated by each recommendation algorithm;

[0171] A fusion module 508 is configured to fuse the plurality of product candidate recommendation results according to the weight coefficient corresponding to each recommendation algorithm to generate a product recommendation result including the plurality of products to be recommended;

[0172] The display module 510 is configured to display the product recommendation results to recommend a plurality of the to-be-recommended products to the target user.

[0173] Optionally, the recommendation algorithm includes a first recommendation model;

[0174] The processing module 506 is further configured to:

[0175] Inputting the user association information and the product information of the plurality of products to be recommended into the first recommendation model to obtain a first matching degree between the user association information and each of the products to be recommended;

[0176] A first matching degree between the user association information and each of the to-be-recommended products is determined as a first product candidate recommendation result generated by the first recommendation model.

[0177] Optionally, the recommendation algorithm includes a second recommendation model, and the model parameter quantity of the second recommendation model is greater than the model parameter quantity of the first recommendation model;

[0178] The processing module 506 is further configured to:

[0179] Inputting the user association information and the product information of the plurality of products to be recommended into the second recommendation model to obtain a second matching degree between the user association information and each of the products to be recommended;

[0180] A second matching degree between the user association information and each of the to-be-recommended products is determined as a second product candidate recommendation result generated by the second recommendation model.

[0181] Optionally, the fusion module 508 is further configured to:

[0182] The first product candidate recommendation result and the second product candidate recommendation result are fused according to a first weight coefficient corresponding to the first recommendation model and a second weight coefficient corresponding to the second recommendation model to generate the product recommendation result.

[0183] Optionally, the device further comprises a screening module configured to:

[0184] Determining a user risk preference level of the target user based on the user association information;

[0185] Screening a plurality of candidate recommended products from a pool of recommended products according to the user's risk preference level, and determining a target allocation ratio corresponding to the target user according to the user's risk preference level;

[0186] According to the target configuration ratio, a plurality of products to be recommended are screened from the plurality of candidate recommended products.

[0187] Optionally, the screening module is further configured to:

[0188] Obtaining the product risk level corresponding to each recommended product in the recommended product pool;

[0189] Determine a recommendable product whose product risk level is less than or equal to the user's risk preference level as a first candidate recommended product;

[0190] Based on the user association information, remove the target user's owned products and customized products from the plurality of first candidate recommended products to obtain a plurality of second candidate recommended products;

[0191] A plurality of the second candidate recommendation products are determined as a plurality of candidate recommendation products.

[0192] Optionally, the screening module is further configured to:

[0193] Obtaining the product configuration ratio corresponding to each candidate recommended product;

[0194] The candidate recommended products whose product configuration ratio is less than or equal to the target configuration ratio are determined as the products to be recommended.

[0195] Optionally, the device further includes an adjustment module configured to:

[0196] Adjust the product recommendation result based on preset business rules to generate an adjusted product recommendation result;

[0197] The display module 510 is further configured to:

[0198] The adjusted product recommendation results are displayed.

[0199] Optionally, the device further includes a training module configured to:

[0200] Acquire sample user association data and sample product data, as well as sample tags between the sample user association data and the sample product data;

[0201] Inputting the sample user association data and the sample product data into the first recommendation model to obtain a predicted matching degree between the sample user association data and the sample product data;

[0202] Calculating a loss value of the first recommendation model according to the sample label and the predicted matching degree;

[0203] The first recommendation model is trained according to the loss value.

[0204] Optionally, the training module is further configured to:

[0205] Obtaining a first user dataset and a first product dataset;

[0206] Respectively removing abnormal data from the first user dataset and the first product dataset to obtain a second user dataset and a second product dataset;

[0207] extracting user features from the second user dataset and product features from the second product dataset;

[0208] combining multiple user features in the second user data set to obtain a composite user feature, and combining multiple product features in the second product data set to obtain a composite product feature;

[0209] The composite user feature is determined as sample user association data, and the composite product feature is determined as sample product data.

[0210] Optionally, the device further comprises an evaluation module configured to:

[0211] Determining an evaluation indicator based on the product recommendation result;

[0212] The first recommendation model or the second recommendation model is adjusted based on the evaluation indicator.

[0213] The product recommendation device provided by the present disclosure includes: an acquisition module, configured to acquire user association information of a target user, and determine a target customer group corresponding to the target user based on the user association information; a determination module, configured to determine multiple recommendation algorithms and a weight coefficient corresponding to each recommendation algorithm based on the target customer group; a processing module, configured to process the user association information and product information of multiple products to be recommended according to each recommendation algorithm, respectively, to obtain a product candidate recommendation result generated by each recommendation algorithm; a fusion module, configured to fuse the multiple product candidate recommendation results according to the weight coefficient corresponding to each recommendation algorithm, to generate a product recommendation result containing multiple products to be recommended; a display module, configured to display the product recommendation result, so as to recommend multiple products to be recommended to the target user.

[0214] The disclosed embodiments achieve the ability to determine the target customer group to which the target user belongs based on the user-related information of the target user, and determine the corresponding multiple recommendation algorithms and weight coefficients corresponding to the recommendation algorithms based on the target customer group to which the target user belongs, so as to process the user-related information and the product information of multiple products to be recommended in combination with the multiple recommendation algorithms, obtain a product recommendation result containing multiple products to be recommended, and display the generated product recommendation result to the target user to recommend multiple products to the target user. This achieves the ability to generate corresponding product recommendation results based on different recommendation algorithms for users belonging to different customer groups, thereby improving the flexibility and accuracy of product recommendations to users and enhancing the user experience.

[0215] Each module in the above-mentioned product recommendation device can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in the computer device in the form of software, so that the processor can call and execute the corresponding operations of each module.

[0216] Figure 6 A block diagram of an electronic device provided in an embodiment of the present disclosure.

[0217] See also Figure 6An embodiment of the present disclosure provides an electronic device 600, which includes: at least one processor 601; at least one memory 602, and one or more I / O interfaces 603 connected between the processor 601 and the memory 602; wherein the memory 602 stores one or more computer programs that can be executed by the at least one processor 601, and the one or more computer programs are executed by the at least one processor 601 so that the at least one processor 601 can execute the above-mentioned product recommendation method.

[0218] Each module in the above-mentioned electronic device can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0219] The present disclosure also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor / processing core, implements the aforementioned product recommendation method. The computer-readable storage medium may be a volatile or non-volatile computer-readable storage medium.

[0220] An embodiment of the present disclosure also provides a computer program product, including a computer-readable code, or a non-volatile computer-readable storage medium carrying the computer-readable code. When the computer-readable code runs in a processor of an electronic device, the processor in the electronic device executes the above-mentioned product recommendation method.

[0221] It will be understood by those skilled in the art that all or some of the steps, systems, and functional modules / units in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. In a hardware implementation, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed by several physical components in cooperation. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or may be implemented as hardware, or may be implemented as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable storage medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or temporary medium).

[0222] As is well known to those skilled in the art, the term computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information (such as computer-readable program instructions, data structures, program modules or other data). Computer storage media includes, but is not limited to, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), static random access memory (SRAM), flash memory or other memory technology, portable compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical disc storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. In addition, as is well known to those skilled in the art, communication media typically contains computer-readable program instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0223] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions to be stored in the computer-readable storage medium in each computing / processing device.

[0224] The computer program instructions for performing the operations of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages such as Smalltalk, C++, and conventional procedural programming languages such as "C" language or similar programming languages. Computer-readable program instructions may be executed entirely on a user's computer, partially on a user's computer, as an independent software package, partially on a user's computer, partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., utilizing an Internet service provider to connect via the Internet). In some embodiments, an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), may be personalized by utilizing the state information of the computer-readable program instructions. The electronic circuit may execute the computer-readable program instructions, thereby realizing various aspects of the present disclosure.

[0225] The computer program product described herein may be implemented in hardware, software, or a combination thereof. In one embodiment, the computer program product is implemented as a computer storage medium. In another embodiment, the computer program product is implemented as a software product, such as a software development kit (SDK).

[0226] Various aspects of the present disclosure are described herein with reference to flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present disclosure. It should be understood that each block of the flowcharts and / or block diagrams, and combinations of blocks in the flowcharts and / or block diagrams, can be implemented by computer-readable program instructions.

[0227] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, thereby producing a machine, so that when these instructions are executed by the processor of the computer or other programmable data processing device, a device is generated that implements the functions / actions specified in one or more blocks in the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium, where these instructions cause the computer, programmable data processing device, and / or other device to operate in a specific manner. Thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing various aspects of the functions / actions specified in one or more blocks in the flowchart and / or block diagram.

[0228] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operational steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to implement the functions / actions specified in one or more blocks in the flowchart and / or block diagram.

[0229] The flow charts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the systems, methods and computer program products according to multiple embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a part of a module, program segment or instruction, and the part of the module, program segment or instruction contains one or more executable instructions for realizing the prescribed logical function. In some alternative implementations, the functions marked in the box can also occur in a sequence different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented by a dedicated hardware-based system that performs the prescribed function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0230] Example embodiments have been disclosed herein, and although specific terms are employed, they are used and should be interpreted only in a general illustrative sense and not for purposes of limitation. In some instances, it will be apparent to those skilled in the art that, unless otherwise expressly indicated, features, characteristics, and / or elements described in conjunction with a particular embodiment may be used alone or in combination with features, characteristics, and / or elements described in conjunction with other embodiments. Therefore, it will be understood by those skilled in the art that various changes in form and detail may be made without departing from the scope of the present disclosure as set forth in the appended claims.

Claims

1. A product recommendation method, characterized in that: include: Obtaining user association information of a target user, and determining a target customer group corresponding to the target user based on the user association information; Determining multiple recommendation algorithms and weight coefficients corresponding to each recommendation algorithm according to the target customer group; Processing the user association information and the product information of the multiple products to be recommended according to each recommendation algorithm respectively to obtain product candidate recommendation results generated by each recommendation algorithm; According to the weight coefficient corresponding to each recommendation algorithm, the plurality of product candidate recommendation results are integrated to generate a product recommendation result including the plurality of products to be recommended; The product recommendation results are displayed to recommend a plurality of the to-be-recommended products to the target user.

2. The method according to claim 1, wherein The recommendation algorithm includes a first recommendation model; The user association information and the product information of the multiple products to be recommended are processed according to each recommendation algorithm to obtain the product candidate recommendation results generated by each recommendation algorithm, including: Inputting the user association information and the product information of the plurality of products to be recommended into the first recommendation model to obtain a first matching degree between the user association information and each of the products to be recommended; A first matching degree between the user association information and each of the to-be-recommended products is determined as a first product candidate recommendation result generated by the first recommendation model.

3. The method according to claim 2, wherein The recommendation algorithm includes a second recommendation model, wherein the model parameter amount of the second recommendation model is greater than the model parameter amount of the first recommendation model; The user association information and the product information of the multiple products to be recommended are processed according to each recommendation algorithm to obtain the product candidate recommendation results generated by each recommendation algorithm, including: Inputting the user association information and the product information of the plurality of products to be recommended into the second recommendation model to obtain a second matching degree between the user association information and each of the products to be recommended; A second matching degree between the user association information and each of the to-be-recommended products is determined as a second product candidate recommendation result generated by the second recommendation model.

4. The method according to claim 3, wherein According to the weight coefficient corresponding to each recommendation algorithm, the plurality of product candidate recommendation results are integrated to generate a product recommendation result including the plurality of products to be recommended, including: The first product candidate recommendation result and the second product candidate recommendation result are fused according to a first weight coefficient corresponding to the first recommendation model and a second weight coefficient corresponding to the second recommendation model to generate the product recommendation result.

5. The method according to claim 1, wherein The method further comprises: Determining a user risk preference level of the target user based on the user association information; Screening a plurality of candidate recommended products from a pool of recommended products according to the user's risk preference level, and determining a target allocation ratio corresponding to the target user according to the user's risk preference level; According to the target configuration ratio, a plurality of products to be recommended are screened from the plurality of candidate recommended products.

6. The method according to claim 5, wherein Based on the user's risk preference level, multiple candidate recommended products are screened from the recommended product pool, including: Obtaining the product risk level corresponding to each recommended product in the recommended product pool; Determine a recommendable product whose product risk level is less than or equal to the user's risk preference level as a first candidate recommended product; Based on the user association information, remove the target user's owned products and customized products from the plurality of first candidate recommended products to obtain a plurality of second candidate recommended products; A plurality of the second candidate recommendation products are determined as a plurality of candidate recommendation products.

7. The method according to claim 5, wherein Screening a plurality of products to be recommended from the plurality of candidate recommended products according to the target configuration ratio includes: Obtaining the product configuration ratio corresponding to each candidate recommended product; The candidate recommended products whose product configuration ratio is less than or equal to the target configuration ratio are determined as the products to be recommended.

8. The method according to claim 1, wherein After generating a product recommendation result including a plurality of products to be recommended, the method further includes: Adjust the product recommendation result based on preset business rules to generate an adjusted product recommendation result; Display the product recommendation results, including: The adjusted product recommendation results are displayed.

9. The method according to claim 2, wherein The first recommendation model is trained based on the following method: Acquire sample user association data and sample product data, as well as sample tags between the sample user association data and the sample product data; Inputting the sample user association data and the sample product data into the first recommendation model to obtain a predicted matching degree between the sample user association data and the sample product data; Calculating a loss value of the first recommendation model according to the sample label and the predicted matching degree; The first recommendation model is trained according to the loss value.

10. The method according to claim 9, wherein Get sample user association data and sample product data, including: Obtaining a first user dataset and a first product dataset; Respectively removing abnormal data from the first user dataset and the first product dataset to obtain a second user dataset and a second product dataset; extracting user features from the second user dataset and product features from the second product dataset; combining multiple user features in the second user data set to obtain a composite user feature, and combining multiple product features in the second product data set to obtain a composite product feature; The composite user feature is determined as sample user association data, and the composite product feature is determined as sample product data.

11. The method according to claim 2 or 3, wherein: The method further comprises: Determining an evaluation indicator based on the product recommendation result; The first recommendation model or the second recommendation model is adjusted based on the evaluation indicator.

12. A product recommendation device, characterized in that: include: an acquisition module configured to acquire user association information of a target user and determine a target customer group corresponding to the target user based on the user association information; a determination module configured to determine a plurality of recommendation algorithms and a weight coefficient corresponding to each recommendation algorithm according to the target customer group; a processing module configured to process the user association information and the product information of the plurality of products to be recommended according to each recommendation algorithm, respectively, to obtain a product candidate recommendation result generated by each recommendation algorithm; a fusion module configured to fuse the plurality of product candidate recommendation results according to the weight coefficient corresponding to each recommendation algorithm to generate a product recommendation result including the plurality of products to be recommended; The display module is configured to display the product recommendation results to recommend a plurality of the to-be-recommended products to the target user.

13. An electronic device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores one or more computer programs executable by the at least one processor. The one or more computer programs are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 11.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the computer program implements the method according to any one of claims 1 to 11.

15. A computer program product, characterized in that The invention comprises a computer-readable code, or a non-volatile computer-readable storage medium carrying the computer-readable code, wherein when the computer-readable code runs in a processor of an electronic device, the processor in the electronic device executes the method according to any one of claims 1 to 11.

Images