Product recommendation method and device, computer equipment and readable storage medium

Abstract

The invention relates to a product recommendation method and device, computer equipment and a readable storage medium. The method comprises the following steps: receiving a product recommendation request; the product recommendation request comprises a target reference product identifier; the target reference product identifier is a product identifier of a target reference product determined based on the user preference information; according to the product type of a target reference product corresponding to the target reference product identifier, obtaining each candidate recommended product from the recommended product set; determining a target candidate recommended product sequence according to the reference product feature data set of the target reference product, the recommended product feature data set corresponding to each candidate recommended product and a cosine similarity algorithm; and obtaining user feature data of the to-be-recommended user, and screening a target recommended product from the target candidate recommended product sequence according to the user feature data. By adopting the method, the accuracy of the product recommendation method can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of application programs, and particularly relates to a product recommendation method, apparatus, computer device, computer-readable storage medium, and computer program product. Background Art

[0002] In the application scenario of bank retail customer relationships, it is necessary to determine the target recommended products that match the user from the bank's recommended product set through a product recommendation method, and recommend the target recommended products to the user.

[0003] The current product recommendation method determines the user portrait of each user according to the user attribute information, historical product purchase data set, and user storage data of each user at a preset time period, and screens the target recommended products to be recommended to the user from each recommended product according to the user portrait of the user.

[0004] However, the current product recommendation method screens the target recommended products according to the user portrait, and the consideration factors are single. Therefore, the accuracy of the current product recommendation method is relatively low. Summary of the Invention

[0005] Based on this, it is necessary to provide a product recommendation method, apparatus, computer device, computer-readable storage medium, and computer program product for the above technical problems.

[0006] In a first aspect, the present application provides a product recommendation method, including:

[0007] Receiving a product recommendation request; the product recommendation request includes a target reference product identifier; the target reference product identifier is the product identifier of the target reference product determined based on user preference information;

[0008] Obtaining each candidate recommended product from the recommended product set according to the product type of the target reference product corresponding to the target reference product identifier;

[0009] Determining a target candidate recommended product sequence according to the reference product feature data set of the target reference product, the recommended product feature data sets corresponding to each candidate recommended product, and the cosine similarity algorithm;

[0010] Obtaining the user feature data of the user to be recommended, and screening the target recommended products from the target candidate recommended product sequence according to the user feature data.

[0011] In one of the embodiments, before receiving the product recommendation request, the method further includes:

[0012] For each product type in the recommended product set, based on the keyword template corresponding to the product type, extract the recommended product keyword sets of each initial recommended product under the product type;

[0013] For each product type in the reference product set, according to the keyword template corresponding to the product type, extract the reference product keyword sets of each reference product under the product type;

[0014] Construct a keyword database based on each of the recommended product keyword sets and each of the reference product keyword sets.

[0015] In one embodiment, the obtaining of each candidate recommended product from the recommended product set according to the product type of the target reference product corresponding to the target reference product identifier includes:

[0016] Query the target product type of the target reference product in the reference product set according to the target reference product identifier;

[0017] Query each candidate recommended product under the target product type in the recommended product set;

[0018] Query the candidate recommended product keyword set of each candidate recommended product in the keyword database, and according to the candidate recommended product keyword set, obtain the recommended product feature data set of the candidate recommended product from the recommended product set.

[0019] In one embodiment, before determining the target candidate recommended product sequence according to the reference product feature data set of the target reference product, the recommended product feature data sets corresponding to each candidate recommended product, and the cosine similarity algorithm, the method further includes:

[0020] Query the reference product keyword set corresponding to the target reference product in the keyword database based on the target reference product identifier;

[0021] Query the target reference product data set of the target reference product in the reference product set according to the target reference product identifier;

[0022] Screen the reference product feature data set corresponding to the target reference product from the target reference product data set according to the reference product keyword set.

[0023] In one embodiment, the determining of the target candidate recommended product sequence according to the reference product feature data set of the target reference product, the recommended product feature data sets corresponding to each candidate recommended product, and the cosine similarity algorithm includes:

[0024] For each of the candidate recommended products, according to the cosine similarity algorithm, perform data operations on the recommended product feature dataset of the candidate recommended product and the reference product feature dataset to obtain the cosine similarity between the candidate recommended product and the target reference product;

[0025] Sort the candidate recommended products in descending order according to each of the cosine similarities to obtain a candidate recommended product sequence;

[0026] Determine a target candidate recommended product sequence from the candidate recommended product sequence according to a preset screening condition.

[0027] In one embodiment, the user feature data includes the preference tags of the user to be recommended and the user's risk tolerance. Screening the target recommended product from the target candidate recommended product sequence according to the user feature data includes:

[0028] Obtain the target candidate recommended product dataset of each target candidate recommended product in the target candidate recommended product sequence; the target candidate recommended product dataset contains the product risk tolerance of the target candidate recommended product;

[0029] Screen each initial target product in each of the target candidate recommended products whose product risk tolerance is lower than and equal to the user's risk tolerance;

[0030] Screen the target recommended product whose product subdivision type conforms to the preference tag from each of the initial target products.

[0031] In a second aspect, the present application also provides a product recommendation device, including:

[0032] A receiving module, configured to receive a product recommendation request; the product recommendation request contains a target reference product identifier; the target reference product identifier is the product identifier of the target reference product determined based on user preference information;

[0033] An obtaining module, configured to obtain each candidate recommended product from the recommended product set according to the product type of the target reference product corresponding to the target reference product identifier;

[0034] A determining module, configured to determine a target candidate recommended product sequence according to the reference product feature dataset of the target reference product, the recommended product feature datasets corresponding to each candidate recommended product, and the cosine similarity algorithm;

[0035] A screening module, configured to obtain the user feature data of the user to be recommended, and screen the target recommended product from the target candidate recommended product sequence according to the user feature data.

[0036] In a third aspect, the present application also provides a computer device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0037] Receive a product recommendation request; the product recommendation request includes a target reference product identifier; the target reference product identifier is the product identifier of a target reference product determined based on user preference information;

[0038] Obtain each candidate recommended product from a recommended product set according to the product type of the target reference product corresponding to the target reference product identifier;

[0039] Determine a target candidate recommended product sequence according to the reference product feature data set of the target reference product, the recommended product feature data sets corresponding to each candidate recommended product, and the cosine similarity algorithm;

[0040] Obtain user feature data of a user to be recommended, and screen a target recommended product from the target candidate recommended product sequence according to the user feature data.

[0041] In a fourth aspect, the present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0042] Receive a product recommendation request; the product recommendation request includes a target reference product identifier; the target reference product identifier is the product identifier of a target reference product determined based on user preference information;

[0043] Obtain each candidate recommended product from a recommended product set according to the product type of the target reference product corresponding to the target reference product identifier;

[0044] Determine a target candidate recommended product sequence according to the reference product feature data set of the target reference product, the recommended product feature data sets corresponding to each candidate recommended product, and the cosine similarity algorithm;

[0045] Obtain user feature data of a user to be recommended, and screen a target recommended product from the target candidate recommended product sequence according to the user feature data.

[0046] In a fifth aspect, the present application also provides a computer program product, including a computer program. When the computer program is executed by a processor, the following steps are implemented:

[0047] Receive a product recommendation request; the product recommendation request includes a target reference product identifier; the target reference product identifier is the product identifier of a target reference product determined based on user preference information;

[0048] Obtain each candidate recommended product from the recommended product set according to the product type of the target reference product corresponding to the target reference product identifier;

[0049] Determine a target candidate recommended product sequence according to the reference product feature data set of the target reference product, the recommended product feature data sets corresponding to the candidate recommended products, and the cosine similarity algorithm;

[0050] Obtain the user feature data of the user to be recommended, and screen the target recommended product from the target candidate recommended product sequence according to the user feature data.

[0051] The above product recommendation method, device, computer device, computer-readable storage medium, and computer program product receive a product recommendation request; the product recommendation request includes a target reference product identifier; the target reference product identifier is the product identifier of the target reference product determined based on user preference information; obtain each candidate recommended product from the recommended product set according to the product type of the target reference product corresponding to the target reference product identifier; determine a target candidate recommended product sequence according to the reference product feature data set of the target reference product, the recommended product feature data sets corresponding to the candidate recommended products, and the cosine similarity algorithm; obtain the user feature data of the user to be recommended, and screen the target recommended product from the target candidate recommended product sequence according to the user feature data. By using this method, the target candidate recommended product sequence is determined through the reference product feature data set and each candidate product feature data set, and the target recommended product is screened from the target candidate recommended product sequence through the user feature data, realizing the joint screening of the target recommended product from the user dimension and the product dimension, increasing the screening dimension of the product dimension, and improving the accuracy of the product recommendation method. Brief Description of the Drawings

[0052] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following will briefly introduce the drawings required for the description of the embodiments of the present application or related technologies. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other related drawings can be obtained without creative efforts based on these drawings.

[0053] Figure 1 It is a flowchart of the product recommendation method in an embodiment;

[0054] Figure 2 It is a flowchart of constructing a keyword database in an embodiment;

[0055] Figure 3 It is a flowchart of determining each candidate recommended product in an embodiment;

[0056] Figure 4 Schematic flow chart for determining a reference product feature data set in one embodiment;

[0057] Figure 5 Schematic flow chart for determining a target candidate recommended product sequence in one embodiment;

[0058] Figure 6 Schematic flow chart for screening target recommended products in one embodiment;

[0059] Figure 7 Schematic architecture diagram of a product recommendation method in an exemplary embodiment;

[0060] Figure 8 Structural block diagram of a product recommendation apparatus in one embodiment;

[0061] Figure 9 Internal structure diagram of a computer device in one embodiment. Detailed implementation manners

[0062] In order to make the objectives, technical solutions and advantages of the present application clearer and more understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0063] In one embodiment, as Figure 1 shown, a product recommendation method is provided. In the embodiments of the present application, this method is described by taking its application to a computer device as an example. The embodiments of the present application do not limit the execution device of the product recommendation method, and include the following steps 102 to step 108:

[0064] Step 102, receiving a product recommendation request.

[0065] Among them, the product recommendation request includes a target reference product identifier; the target reference product identifier is the product identifier of the target reference product determined based on user preference information.

[0066] In implementation, the computer device receives a product recommendation request sent by a terminal. Among them, the product recommendation request includes the target reference product identifier of the target reference product and the user identifier of the user to be recommended. The target reference product is the reference product intended by the user to be recommended. The reference product and the initial recommended product are products of the same business. For example, if the initial recommended product is a financial product, the reference product is also a financial product.

[0067] Specifically, the target reference product may be, but is not limited to, an actual or simulated reference product. If the target reference product is an actual reference product, the computer device determines the target reference product from the reference product set according to the user preference information. If the target reference product is a simulated reference product, the computer device generates a simulated target reference product according to the user preference information.

[0068] In an exemplary embodiment, the computer device obtains the user preference information of the user to be recommended according to the user identifier of the user to be recommended, and determines the target reference product from the reference product set according to the user preference information of the user to be recommended. Specifically, the user preference information includes the product data sets of each historical purchased product and the product data sets of each historical browsed product. The computer device determines each historical purchased product and each historical browsed product as each target reference product in the reference product set. For each target reference product, the computer device sends a verification request to the verification terminal of the verification object based on the user to be recommended and the target reference product. When the verification result corresponding to the verification request is verified to be passed, the verification terminal sends a product recommendation request to the computer device based on the target reference product identifier of the target reference product and the product identifier of the product to be recommended. The computer device receives the product recommendation request sent by the verification terminal.

[0069] In an exemplary embodiment, the computer device sends each reference product in the reference product set and the user information of each user to the terminal of the target object, so that the target object determines the target reference product from each reference product and determines the user to be recommended for the product to be recommended by the target object from each user. Then, the terminal of the target object constructs a product recommendation request based on the target reference product identifier of the target reference product and the user identifier of the user to be recommended, and sends the product recommendation request to the computer device. The computer device receives the product recommendation request.

[0070] For example, the initial recommended financial product is a financial product of the target bank, while the reference financial products are financial products of other banks except the target bank. The computer device sends each reference financial product and the user information of each user to the terminal of the marketer, instructing the marketer to determine the target reference financial product from each reference financial product and determine the user to be recommended (for product recommendation) from each user. The terminal of the marketer constructs a product recommendation request based on the target reference product identifier of the target reference financial product and the user identifier of the user to be recommended, and sends the product recommendation request to the computer device. The computer device receives the product recommendation request and calls the backend service interface provided by the service interface conversion module, thereby transmitting the user identifier of the user to be recommended and the target reference product identifier of the target reference financial product to the backend. Then, the service interface conversion process of the client application module in the computer device calls the product recommendation service interface provided by the business real-time processing engine to start the product recommendation method.

[0071] In an optional embodiment, the recommended product set includes various recommended wealth management products. The reference product set includes various reference wealth management products. The recommended wealth management products are wealth management products in the target bank, and the reference wealth management products are wealth management products in other banks. The computer device obtains the user preference information of the user to be recommended according to the user identification of the user to be recommended. Then, based on the user to be recommended and the user preference information of the user to be recommended, the computer device sends a target reference product request to the terminal of the marketer. The terminal receives the target reference product request and displays the user preference information in the target reference product request. The marketer determines the target reference product from the historical purchased products and the historical viewed products in the user preference information and submits the target reference product. Then, based on the target reference product identification of the target reference product and the user identification of the user to be recommended, the terminal constructs a product recommendation request and sends the product recommendation request to the computer device. The computer device receives the product recommendation request.

[0072] Optionally, the target reference product may be a product of the target bank or a product of other banks. The embodiments of the present application do not limit the target reference product.

[0073] In an exemplary embodiment, the computer device obtains the historical view data set and the historical product purchase data set of the user to be recommended from the database according to the user identification of the user to be recommended in the product recommendation request, so as to obtain the user preference information of the user to be recommended. The computer device extracts each historical viewed product from the historical view data set and obtains the product data set of each historical viewed product. At the same time, the computer device extracts each historical purchased product from the historical product purchase data set and obtains the product data set of each historical purchased product. Then, the computer device inputs the product data sets of each historical purchased product and the product data sets of each historical viewed product into the product simulation model, and performs prediction processing on each product data set through the product simulation model to obtain the product data set of the simulated product. The computer device determines the simulated product as the target reference product of the user to be recommended. The computer device sends a verification request to the verification terminal of the verification object based on the user to be recommended and the target reference product. When the verification result corresponding to the verification request is verified to pass, the verification terminal sends a product recommendation request to the computer device based on the target reference product identification of the target reference product and the product identification of the product to be recommended. The computer device receives the product recommendation request sent by the verification terminal. By using the product recommendation method to verify whether the user to be recommended can accept the simulated product, the initial recommended products and the simulated products in the recommended product set can be optimized.

[0074] In an optional embodiment, the computer device obtains the user preference information of each user. The user preference information of a user includes the user's purchase data set and user browsing data set. The computer device inputs the user preference information of each user into a simulation model, and performs prediction processing on the user preference information of each user through the simulation model to obtain a simulation product data set of the simulation product. Then, the computer device determines the simulation product as the target reference product. Then, the computer device sends a verification request to the verification terminal of the verification object based on the user to be recommended and the target reference product. When the verification result corresponding to the verification request is verified to pass, the verification terminal sends a product recommendation request to the computer device based on the target reference product identifier of the target reference product and the product identifier of the product to be recommended. The computer device receives the product recommendation request sent by the verification terminal. The computer device screens the target recommended product from the recommended product set through the target reference product. If there is no target recommended product in the recommended product set, the computer device updates the recommended product set based on the target reference product.

[0075] Step 104, according to the product type of the target reference product corresponding to the target reference product identifier, obtain each candidate recommended product from the recommended product set.

[0076] In implementation, the computer device determines the product type of the target reference product according to the target reference product identifier. Then, the computer device obtains each candidate recommended product of this product type from the recommended product set. The computer device obtains the recommended product feature data set of each candidate recommended product.

[0077] Specifically, the computer device determines the product type of the target reference product in the reference product database according to the target reference product identifier. Then, the computer device obtains each candidate recommended product from the recommended product set according to this product type. The computer device obtains the candidate recommended product keyword set of each candidate recommended product from the keyword database, and obtains the recommended product feature data set of each candidate recommended product from the recommended product set according to each candidate recommended product keyword set.

[0078] Step 106, determine the target candidate recommended product sequence according to the reference product feature data set of the target reference product, the recommended product feature data sets corresponding to each candidate recommended product, and the cosine similarity algorithm.

[0079] In implementation, for each candidate recommended product, the computer device performs data processing on the recommended product feature data set of the candidate recommended product and the reference product feature data set of the target reference product according to the cosine similarity algorithm to obtain the cosine similarity between the target reference product and the candidate recommended product. Then, the computer device determines the target candidate recommended product sequence according to each cosine similarity and each candidate recommended product.

[0080] Specifically, the computer device obtains the reference product keyword set of the target reference product from the keyword database according to the target reference product identifier, and obtains the reference product feature data set of the target reference product from the reference product set according to the reference product keyword set. Then, for each candidate recommended product, the computer device processes the recommended product feature data set of the candidate recommended product and the reference product feature data set of the target reference product according to the cosine similarity algorithm to obtain the cosine similarity between the target reference product and the candidate recommended product. Then, the computer device determines the target candidate recommended product among the candidate recommended products according to each cosine similarity, so as to obtain the target candidate recommended product sequence.

[0081] In an optional embodiment, a similarity model is set in the computer device. The similarity model is used to determine the similarity between two products. For each candidate recommended product, the computer device inputs the recommended product feature data set of the candidate recommended product and the reference product feature data set of the target reference product into the similarity model, and calculates the similarity between the candidate recommended product and the target reference product through the similarity model. Then, the computer device determines the target candidate recommended product sequence according to each similarity and each candidate recommended product.

[0082] Optionally, the similarity model can be but is not limited to a BERT (Bidirectional Encoder Representations from Transformers, a pre-trained language model) model or an XLM-R model (XLM-RoBERTa, a multilingual training model), and the similarity model is not limited in the embodiments of the present application.

[0083] Step 108, obtain the user feature data of the user to be recommended, and screen the target recommended product from the target candidate recommended product sequence according to the user feature data.

[0084] In implementation, the computer device obtains the user feature data from the user database according to the user identifier of the user to be recommended. Then, the computer device screens the target recommended product from the target candidate recommended product sequence with the user feature data as the screening condition.

[0085] In an exemplary embodiment, the computer device obtains the user feature data of the user to be recommended from the user database according to the user identifier of the user to be recommended. Then, for each target candidate recommended product in the target candidate recommended product sequence, the computer device determines the matching degree between the user to be recommended and the target candidate recommended product according to the recommended product feature data set and the user feature data set of the target candidate recommended product. Then, the computer device sorts the target candidate recommended products in descending order of the matching degree to obtain a matching degree sequence. The computer device determines the target recommended product according to the matching degree sequence and the target candidate recommended product sequence.

[0086] Specifically, a matching degree model is set in the computer device. For each target candidate recommended product in the target candidate recommended product sequence, the computer device inputs the recommended product feature data set and the user feature data set of the target candidate recommended product into the matching degree model, and calculates the matching degree between the user to be recommended and the target candidate recommended product through the matching degree model. Then, the computer device sorts the target candidate recommended products in descending order of the matching degree to obtain a matching degree sequence. The computer device determines the matching degree scores of the target candidate recommended products in the matching degree sequence in the order of the matching degree sequence, and determines the similarity scores of the target candidate recommended products in the target candidate recommended product sequence in the order of the target candidate recommended product sequence. Then, the computer device adds the similarity scores and the matching degree scores of the same target candidate recommended product to obtain the total score of the target candidate recommended product. The computer device determines the target candidate recommended product with the highest total score as the target recommended product.

[0087] In an alternative embodiment, the target reference product is an actual recommended product. If the computer device can screen out the target recommended product, the computer device constructs recommendation information based on the recommended product data set of the target recommended product and the user information of the user to be recommended. Then, the computer device sends the recommendation information to the target object so that the target object recommends the target recommended product to the user to be recommended according to the recommendation information.

[0088] If the target recommended product screened by the computer device is empty, the computer device determines that there is no target recommended product in the recommended product set, and pushes the fact that there is no target recommended product in the recommended product set to the target object. The computer device reports the reference product data set of the target reference product to the product designer so that the product designer can identify the lack of the target recommended product similar to the target reference product, thereby supplementing the target recommended product.

[0089] In an optional embodiment, the target reference product is a simulated recommended product. By executing the product recommendation method, the user acceptance of the target reference product can be verified, avoiding the manual verification of the target reference product through research, and improving the verification efficiency. Specifically: If the computer device can screen out the target recommended product, the computer device constructs recommendation information based on the recommended product data set of the target recommended product and the user information of the user to be recommended. Then, the computer device sends the recommendation information to the target object, so that the target object recommends the target recommended product to the user to be recommended according to the recommendation information. Then, the computer device labels the target reference product based on the simulation label and updates the reference product set according to the reference product information of the target reference product. The computer device establishes an association relationship between the user to be recommended and the target reference product and updates the accepted users of the target reference product. If the number of accepted users of the target reference product is greater than the acceptance user threshold, the computer device pushes the reference product data set of the target reference product to the product designer to assist the product designer in verifying the target reference product.

[0090] If the target recommended product screened by the computer device is empty, the computer device determines that there is no target recommended product in the recommended product set and pushes the fact that there is no target recommended product in the recommended product set to the target object. The computer device reports the reference product data set of the target reference product to the product designer so that the product designer can clarify that the target reference product cannot be accepted by the user to be recommended. The computer device constructs a verification result based on the non-acceptance and the target reference product and updates the verification result set.

[0091] In the above product recommendation method, the target candidate recommended product sequence is determined through the reference product feature data set and each candidate product feature data set, and the target recommended product is screened from the target candidate recommended product sequence through the user feature data, realizing the joint screening of the target recommended product from the user dimension and the product dimension, increasing the screening dimension of the product dimension, and improving the accuracy of the product recommendation method.

[0092] In an exemplary embodiment, it can be understood that before receiving the product recommendation, it is necessary to first extract the recommended product keyword sets of each recommended product in the recommended product set and the reference product keyword sets of each reference product in the reference product set to obtain the keyword set. As Figure 2 shown, in step 102, the executor countersigns, and the specific processing process of this product recommendation method further includes steps 202 to 206. Among them:

[0093] Step 202, for each product type in the recommended product set, based on the keyword template corresponding to the product type, extract the recommended product keyword sets of each initial recommended product under the product type.

[0094] Among them, the recommended product set contains the recommended product data sets of each initial recommended product.

[0095] In implementation, keyword templates corresponding to each product type are preset in the database. The computer device obtains the keyword template corresponding to each product type from the database for each product type in the recommended product set. Then, the computer device extracts the recommended product keyword set of each initial recommended product from the recommended product data set of each initial recommended product under the product type according to the keyword template. Among them, each recommended product keyword in the recommended product keyword set is a product element of the initial recommended product. The product elements of different initial recommended products are different.

[0096] In an exemplary embodiment, the initial recommended product is an initial recommended financial product. The initial financial products are divided into bank deposits, bank financial products, funds, bonds, stocks, insurance products, etc. The computer device obtains the keyword template corresponding to each product type from the database for each product type in the recommended product set. The computer device extracts the recommended product keyword set of each initial recommended product from the recommended product data set of each initial recommended product under the product type according to the keyword template. For example, for the product type of bonds, the computer device extracts the product price, issue term, minimum purchase amount, maximum purchase limit, allowable purchase term, and risk level of the initial recommended product from the recommended product data set of each initial recommended product under the product type of bonds, to obtain each recommended product keyword of the initial recommended product type. Then, the computer device constructs the recommended product keyword set of the initial recommended product according to each initial recommended product keyword of the initial recommended product.

[0097] In an alternative embodiment, the computer device extracts the product data of the target bank from the batch data source of the product information of the target bank through batch ETL (ETL is the abbreviation of Extract, Transform, and Load, which is a process of data integration) to form a recommended product set (recommended product database). At the same time, the computer device obtains the reference product data sets of each reference financial product from each data source in the public channel through data crawling or data dimension. Then, the computer device constructs a reference product set according to the reference product data sets of each reference financial product. The computer device extracts the recommended product keyword sets of each initial recommended product in the recommended product set according to the keyword templates of each product type.

[0098] Optionally, the keyword templates of each product type are determined according to the product attribute information of each product type, and the embodiments of the present application do not limit the keyword templates.

[0099] Step 204 , for each product type in the reference product set, extract the reference product keyword set of each reference product under the product type according to the keyword template corresponding to the product type.

[0100] The reference product set includes the reference product data sets of each reference product.

[0101] In implementation, keyword templates corresponding to each product type are pre-set in the database. For each product type in the reference product set, the computer device obtains the keyword template corresponding to the product type from the database. Then, based on the keyword template, the computer device extracts the reference product keyword set of the initial reference product from the reference product data set of each initial reference product under the product type. The reference product keyword set is the product element of the initial reference product.

[0102] In an exemplary embodiment, the initial reference product is an initial reference financial product. The initial financial products are divided into bank deposits, bank financial products, funds, bonds, stocks and insurance products, etc. The computer device obtains the keyword template corresponding to each product type in the reference product set from the database. According to the keyword template, the computer device extracts the reference product keyword set of the initial reference product from the reference product data set of each initial reference product under the product type. For example, for the product type of fund, the computer device proposes the product price, issuance period, minimum purchase amount, maximum purchase limit, allowed purchase period, risk level and industry of the initial reference product in the reference product data set of each initial reference product under the product type of fund, and obtains the reference product keywords of the initial reference product type. Then, the computer device constructs the reference product keyword set of the initial reference product according to the initial reference product keywords of the initial reference product. It should be understood that the reference product and the initial recommended product are interbank products. Interbank products refer to various products involved in business cooperation between financial institutions.

[0103] Step 206: construct a keyword database based on each recommended product keyword set and each reference product keyword set.

[0104] In implementation, the computer device stores each recommended keyword set and each reference product keyword in a database to obtain a keyword database. The keyword database is constructed based on product keyword sets of products of different product types (reference products or initial recommended products). Each product keyword set represents the product characteristics of the product. The keyword database is used for retrieval by the product real-time calculation recommendation program to determine the reference product feature data set of the target reference product and the recommended product feature data set of the candidate recommended product.

[0105] In this embodiment, by extracting the recommended product keyword set and the reference product keyword set and constructing a keyword database, the product elements of each initial recommended product and each reference product are obtained, which facilitates the subsequent calculation of the similarity between the target reference product and the candidate recommended product. In addition, compared with directly calculating the reference product data set of the target reference product and the recommended product data set of the candidate recommended product, the amount of data calculation is reduced, the data calculation efficiency is improved, and thus the efficiency of the product recommendation method is improved.

[0106] In an exemplary embodiment, as Figure 3 shown, the specific processing process of step 104 includes steps 302 to 306. Among them:

[0107] Step 302, according to the target reference product identifier, query the target product type of the target reference product in the reference product set.

[0108] In implementation, the computer device queries the product type of the target reference product in the reference product set according to the target reference product identifier to obtain the target product type.

[0109] Specifically, the computer device queries the reference product data set of the target reference product in the reference product set according to the target reference product identifier to obtain the target reference product data set. Then, the computer device determines the target product type of the target reference product in the target reference product data set.

[0110] Step 304, query each candidate recommended product under the target product type in the recommended product set.

[0111] In implementation, the computer device screens each candidate recommended product from each initial recommended product in the recommended product set according to the target product type.

[0112] Specifically, for each initial recommended product in the recommended product set, the computer device determines whether the product type of the initial recommended product is the target product type. If the product type of the initial recommended product is the target product type, the computer device determines the initial recommended product as a candidate recommended product.

[0113] In an exemplary embodiment, the initial recommended product is an initial recommended financial product, and the target product type is a fund. For each initial recommended financial product in the recommended product set, the computer device determines whether the initial recommended financial product is a fund. If the initial recommended financial product is a fund, the computer device determines the initial recommended financial product as a candidate recommended product.

[0114] Step 306, query the candidate recommended product keyword set of each candidate recommended product in the keyword database, and obtain the recommended product feature data set of the candidate recommended product from the recommended product set according to the candidate recommended product keyword set.

[0115] In implementation, the computer device obtains the full set of recommended product keywords in the keyword database. For each candidate recommended product, the computer device determines the candidate recommended product keyword set of the candidate recommended product in the full set of recommended product keywords according to the product identifier of the candidate recommended product. The candidate recommended product keyword set contains each candidate recommended product keyword of the candidate recommended product, and the candidate recommended product keyword is a key element of the candidate recommended product. The computer device queries the recommended product data set of the candidate recommended product in the recommended product set according to the product identifier of the candidate recommended product, and filters the recommended product feature data set in the recommended product data set according to the candidate recommended product keyword set.

[0116] Specifically, the candidate recommended product keyword set only contains each candidate recommended product keyword of the candidate recommended product. However, the similarity between the candidate recommended product and the target reference product cannot be determined only based on the candidate recommended product keyword. Therefore, it is necessary to obtain the keyword value corresponding to the candidate recommended product keyword. Therefore, for the product identifier of each candidate recommended product, the computer device queries the recommended product data set of the candidate recommended product in the recommended product set. Then, for each candidate recommended product keyword in the candidate recommended product keyword set, the computer device determines the keyword value corresponding to the candidate recommended product keyword set in the recommended product data set, and adds the keyword value to the candidate recommended product keyword set to obtain the recommended product feature data set of the candidate recommended product.

[0117] In an exemplary embodiment, taking the candidate recommended product as the first fund product as an example, the determination of the recommended product feature data set corresponding to the first fund product is specifically described. Specifically, the computer device determines the candidate recommended product keyword set of the first fund product in the full set of recommended product keywords according to the product identifier of the first fund product. Among them, the candidate recommended product keyword set of the first fund product includes product price, issue term, minimum purchase amount, maximum purchase limit, permitted purchase term, risk level, and industry. The computer device queries the recommended product data set of the first fund product in the recommended product set according to the product identifier of the first fund product. Then, for each candidate recommended product keyword in the candidate recommended product keyword set, the computer device determines the keyword value corresponding to the candidate recommended product keyword set in the recommended product data set, and adds the keyword value to the candidate recommended product keyword set to obtain the recommended product feature data set of the candidate recommended product. For example, for the candidate recommended product keyword of product price, the computer device determines that the price of the first fund product is 2 yuan in the recommended product data set. Then, the computer device updates the keyword value of 2 yuan to the candidate recommended product keyword set.

[0118] In this embodiment, through the product type of the target reference product, the candidate recommended products and the recommended product feature datasets of the candidate recommended products are determined, and a preliminary screening is carried out to obtain the candidate recommended products, which is convenient for subsequent screening of the target recommended products from the product dimension according to the recommended product feature datasets.

[0119] In an exemplary embodiment, it should be understood that before calculating the similarity based on the reference product feature dataset, it is necessary to first obtain the reference product feature dataset of the target reference product. As Figure 4 shown, before step 106 is executed, the specific processing process of this product recommendation method further includes steps 402 to 406. Among them:

[0120] Step 402, based on the target reference product identifier, query the reference product keyword set corresponding to the target reference product in the keyword database.

[0121] In implementation, the computer device obtains the full set of reference product keywords from the keyword database. Then, the computer device determines the reference product keyword set corresponding to the target reference product in each reference product keyword set according to the target reference product identifier.

[0122] In an exemplary embodiment, the target reference product is a target fund product. The computer device obtains the full set of reference product keywords from the keyword database. The computer device determines the reference product keyword set corresponding to the target fund product in each reference product keyword set according to the target reference product identifier of the target fund product. Among them, the reference product keyword set corresponding to the target fund product includes product price, issue term, minimum purchase amount, maximum purchase limit, allowable purchase term, risk level, and industry.

[0123] For example, the product recommendation service interface call engine inside the business real-time processing engine in the computer device calls the keyword retrieval module. The keyword retrieval module extracts the reference product keyword set of the target reference product from the keyword database based on the target reference product identifier. Then, the product recommendation service interface call feature extraction module in the business real-time processing engine of the computer device. The feature extraction module extracts the reference product feature dataset of the target reference product from the reference product set according to the target reference product identifier and the reference product keyword set.

[0124] Step 404, according to the target reference product identifier, query the target reference product dataset of the target reference product in the reference product set.

[0125] In implementation, the computer device queries the reference product dataset of the target reference product in the reference product dataset according to the target reference product identifier, and determines the reference product dataset of the target reference product as the target reference product dataset.

[0126] In an exemplary embodiment, the target reference product is a target fund product. The computer device queries the reference product data set of the target fund product in the reference product data set according to the target reference product identifier. Then, the computer device determines the reference product data set as the target reference product data set.

[0127] Step 406: According to the reference product keyword set, screen the reference product feature data set corresponding to the target reference product in the target reference product data set.

[0128] In implementation, for each reference product keyword in the reference product keyword set, the computer device determines the keyword value corresponding to the reference product keyword in the target reference product data set, and updates the keyword value to the reference product keyword set to obtain the reference product feature data set corresponding to the target reference product.

[0129] In an exemplary embodiment, for each reference product keyword in the reference product keyword set, the computer device determines the keyword value corresponding to the candidate recommended product keyword set in the target reference product data set, and adds the keyword value to the reference product keyword set to obtain the reference product feature data set of the target reference product. For example, for the candidate recommended product keyword of the minimum purchase amount, the computer device determines that the price of the target fund product is 100 yuan in the target reference product data set. Then, the computer device updates the keyword value of 100 yuan to the reference product keyword set.

[0130] In this embodiment, by determining the reference product keyword set of the target reference product and determining the reference product feature data set corresponding to the target reference product according to the reference product keyword set, the features of the target reference product are obtained, which is convenient for subsequent screening of target recommended products from the product dimension according to the reference product feature data set.

[0131] In an exemplary embodiment, as Figure 5 shown, the specific processing process of step 106 includes steps 502 to 506. Among them:

[0132] Step 502: For each candidate recommended product, according to the cosine similarity algorithm, perform data operations on the recommended product feature data set and the reference product feature data set of the candidate recommended product to obtain the cosine similarity between the candidate recommended product and the target reference product.

[0133] Among them, the cosine similarity algorithm is the text cosine similarity algorithm. The cosine similarity is the similarity probability between the candidate recommended product and the target reference product.

[0134] In implementation, for each candidate recommended product, the computer device calculates data on the recommended product feature data set and the reference product feature data set of the candidate recommended product according to the text cosine similarity algorithm, and obtains the cosine similarity between the candidate recommended product and the target reference product. The higher the cosine similarity, the more similar the candidate recommended product and the target reference product are; the lower the cosine similarity, the more different the candidate recommended product and the target reference product are.

[0135] Optionally, the text cosine similarity algorithm can be but is not limited to the bag-of-words model or TF–IDF (Term Frequency–Inverse Document Frequency). The embodiments of the present application do not limit the text cosine similarity algorithm.

[0136] Step 504: Sort the candidate recommended products in descending order of each cosine similarity to obtain a candidate recommended product sequence.

[0137] In implementation, the computer device sorts the candidate recommended products in descending order of the cosine similarity of each candidate recommended product to obtain a candidate recommended product sequence.

[0138] In an exemplary embodiment, each candidate recommended product is each candidate recommended fund product. The computer device sorts the candidate recommended fund products in descending order of each cosine similarity to obtain a candidate recommended fund product sequence.

[0139] Step 506: Determine a target candidate recommended product sequence from the candidate recommended product sequence according to a preset screening condition.

[0140] In implementation, the computer device pre-sets a screening condition. The computer device determines a target candidate recommended product sequence from the candidate recommended product sequence according to the screening condition.

[0141] Specifically, the screening condition is a recommended quantity threshold. The computer device determines the candidate recommended products with the recommended quantity threshold in the candidate recommended product sequence, and constructs a target candidate recommended product sequence according to each candidate recommended product.

[0142] In an alternative embodiment, the screening condition is a cosine similarity threshold. The computer device determines, in the order of the candidate recommended product sequence, whether the cosine similarity of each candidate recommended product in the candidate recommended product sequence reaches the cosine similarity threshold. If the cosine similarity of the candidate recommended product reaches the cosine similarity threshold, the computer device determines the candidate recommended product as a target candidate recommended product. The computer device constructs a target candidate recommended product sequence based on each target candidate recommended product.

[0143] In this embodiment, by calculating the cosine similarities between the reference product feature dataset and each candidate product feature dataset, and determining the target candidate recommended product sequence based on each cosine similarity, the target candidate recommended products obtained are screened from the product dimension, adding the product dimension as a screening dimension and improving the accuracy of the product recommendation method.

[0144] In an exemplary embodiment, the user feature data includes the preference labels of the user to be recommended and the user's risk tolerance. As Figure 6 shown, the specific processing procedure for screening the target recommended products from the target candidate recommended product sequence according to the user feature data in step 108 includes steps 602 to 606. Among them:

[0145] Step 602, obtain the target candidate recommended product datasets of each target candidate recommended product in the target candidate recommended product sequence.

[0146] Among them, the target candidate recommended product dataset contains the product risk tolerance of the target candidate recommended product.

[0147] In implementation, for each target candidate recommended product in the target candidate recommended product sequence, the computer device queries the recommended product dataset of the target candidate recommended product in the recommended product set according to the product identifier of the target candidate recommended product, and determines this recommended product dataset as the target candidate recommended product dataset.

[0148] Optionally, the product identifier may be, but is not limited to, the ID (Identity Document, unique code) or name of the target candidate recommended product. The embodiments of the present application do not limit the product identifier.

[0149] Step 604, screen out the initial target products whose product risk tolerance is lower than and equal to the user risk tolerance among each target candidate recommended product.

[0150] In implementation, for each target candidate recommended product, the computer device determines whether the user risk tolerance is greater than or equal to the product risk tolerance of the target candidate recommended product. If the user risk tolerance is greater than or equal to the product risk tolerance, the computer device determines this target candidate recommended product as the initial target product. If the user risk tolerance is less than the product risk tolerance, the computer device screens out this target candidate recommended product. It should be noted that if the user risk tolerance is less than the product risk tolerance, it means that the user to be recommended is not suitable for this target candidate recommended product and the user to be recommended has no intention of owning this target candidate recommended product.

[0151] Step 606, screen out the target recommended products whose product subdivision types conform to the preference labels among each initial target product.

[0152] Among them, the preference tags contain various target product subdivision types preferred by the user to be recommended.

[0153] In implementation, for each initial target product, the computer device determines whether the product subdivision type of the initial target product is a target product subdivision type. If the product subdivision type of the initial target product is a target product subdivision type, the computer device determines the initial target product as a target recommended product. If the product subdivision type of the initial target product is not a target product subdivision type, it indicates that the user to be recommended is not interested in the target initial product, and the computer device filters out the initial target product.

[0154] In an optional embodiment, the computer device determines whether the target recommended product is a main recommended product according to the recommended product data set of the target recommended product. If the target recommended product is a main recommended product, the computer device marks the target recommended product and sends the recommended product data set of the target recommended product and the user information of the user to be recommended to the target object, so that the target object recommends the target recommended product to the user to be recommended.

[0155] In this embodiment, by screening target recommended products from the target candidate recommended product sequence through user feature data, the goal of jointly screening target recommended products from the user dimension and the product dimension is achieved, adding the product dimension as a screening dimension and improving the accuracy of the product recommendation method.

[0156] In an exemplary embodiment, Figure 7 is a schematic architecture diagram of the product recommendation method in an exemplary embodiment. As Figure 7 shown, the computer device extracts the product data sets of each initial recommended product from the business batch data source of the target bank to obtain a recommended product library (bank-owned product information library). At the same time, the computer device obtains the reference product data sets of each reference product from the peer product information data source to obtain a reference product library (peer product information library). Then, the computer device extracts the keywords of each product from the recommended product library and the reference product library to obtain a keyword library (product feature keyword library). The recommended product library, the reference product library, and the keyword library are all in the MPP (Massively Parallel Processing) database.

[0157] The computer device receives a product recommendation request. The product recommendation information includes the product information of the target reference product (peer product) selected by the target object and the user identifier of the user to be recommended. The computer device performs service interface conversion and invokes the product recommendation service interface based on the in-memory real-time processing engine. The product recommendation server interface invokes the keyword retrieval module to retrieve the reference product keyword set of the target reference product. Then, the product recommendation service interface performs feature extraction according to the reference keyword set to obtain the reference product feature data set of the target reference product. The computer device performs product feature retrieval to obtain each candidate recommended product and the recommended product feature data set of each candidate recommended product. For each candidate recommended product, the computer device performs data operations on the recommended product feature data set and the reference product feature data set of the candidate recommended product according to the cosine similarity algorithm to obtain the cosine similarity between the candidate recommended product and the target reference product. Then, the computer device sorts each candidate recommended product according to each cosine similarity to obtain the target candidate recommended product sequence. The computer device filters out the main promoted products in the target candidate recommended product sequence and filters again according to the user feature data of the user to be recommended to obtain the target recommended products. The computer device displays the target recommended products and instructs the target object to recommend the target recommended products to the user to be recommended. In the case where the user to be recommended selects to place an order for the target recommended products, the product order placement operation process is entered.

[0158] By using MPP database, in-memory computing, and cosine similarity calculation technologies, similarity calculations are performed using the feature keyword data of the target reference product and each candidate recommended product. Based on the similarity calculation results, the main promoted products, the preferences of the user to be recommended for products, and the risk tolerance of the user to be recommended are matched to generate personalized product recommendations for the user to be recommended that can be benchmarked against peer products, solving the problem that marketers lack a peer product benchmark reference when making product recommendations in the product marketing scenario of the user to be recommended.

[0159] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are displayed in sequence according to the arrows, these steps do not necessarily have to be executed in the order indicated by the arrows. Unless clearly stated in this article, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, at least some of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages do not necessarily have to be executed at the same time, but can be executed at different times. The execution order of these steps or stages does not necessarily have to be sequential, but can be executed alternately or in turn with at least some of the steps or stages in other steps or other steps.

[0160] Based on the same inventive concept, an embodiment of the present application further provides a product recommendation device for implementing the product recommendation method involved above. The implementation solutions provided by this device to solve problems are similar to those recorded in the above method. Therefore, the specific limitations in one or more embodiments of the product recommendation device provided below can refer to the limitations on the product recommendation method in the above text, and will not be elaborated here.

[0161] In an exemplary embodiment, as Figure 8 shown, a product recommendation device 800 is provided, including: a receiving module 801, an obtaining module 802, a determining module 803, and a screening module 804, where:

[0162] The receiving module 801 is configured to receive a product recommendation request; the product recommendation request includes a target reference product identifier; the target reference product identifier is the product identifier of the target reference product determined based on user preference information.

[0163] The obtaining module 802 is configured to obtain each candidate recommended product from the recommended product set according to the product type of the target reference product corresponding to the target reference product identifier.

[0164] The determining module 803 is configured to determine a target candidate recommended product sequence according to the reference product feature data set of the target reference product, the recommended product feature data sets corresponding to each candidate recommended product, and the cosine similarity algorithm.

[0165] The screening module 804 is configured to obtain the user feature data of the user to be recommended, and screen the target recommended product from the target candidate recommended product sequence according to the user feature data.

[0166] In an exemplary embodiment, the product recommendation device 800 further includes:

[0167] The first extraction module is configured to extract a recommended product keyword set of each initial recommended product under the product type based on the keyword template corresponding to the product type for each product type in the recommended product set.

[0168] The second extraction module is configured to extract a reference product keyword set of each reference product under the product type according to the keyword template corresponding to the product type for each product type in the reference product set.

[0169] The first construction module is configured to construct a keyword database according to each recommended product keyword set and each reference product keyword set.

[0170] In an exemplary embodiment, the obtaining module 802 includes:

[0171] The first query sub-module is used to query the target product type of the target reference product in the reference product set according to the target reference product identifier.

[0172] The second query sub-module is used to query each candidate recommended product under the target product type in the recommended product set.

[0173] The third query sub-module is used to query the candidate recommended product keyword set of each candidate recommended product in the keyword database, and obtain the recommended product feature data set of the candidate recommended product from the recommended product set according to the candidate recommended product keyword set.

[0174] In an exemplary embodiment, the product recommendation device 800 further includes:

[0175] The first query module is used to query the reference product keyword set corresponding to the target reference product in the keyword database based on the target reference product identifier.

[0176] The second query module is used to query the target reference product data set of the target reference product in the reference product set according to the target reference product identifier.

[0177] The second screening module is used to screen the reference product feature data set corresponding to the target reference product in the target reference product data set according to the reference product keyword set.

[0178] In an exemplary embodiment, the determination module 803 includes:

[0179] The first arithmetic sub-module is used to perform data arithmetic on the recommended product feature data set and the reference product feature data set of each candidate recommended product according to the cosine similarity algorithm to obtain the cosine similarity between the candidate recommended product and the target reference product.

[0180] The first sorting sub-module is used to sort each candidate recommended product in descending order of each cosine similarity to obtain a candidate recommended product sequence.

[0181] The first screening sub-module is used to determine a target candidate recommended product sequence in the candidate recommended product sequence according to a preset screening condition.

[0182] In an exemplary embodiment, the user feature data includes the preference tags of the user to be recommended and the user's risk tolerance, and the screening module includes a first acquisition sub-module and a second screening sub-module. Among them, the second screening sub-module includes:

[0183] The second acquisition sub-module is used to acquire the target candidate recommended product data set of each target candidate recommended product in the target candidate recommended product sequence; the target candidate recommended product data set includes the product risk tolerance of the target candidate recommended product.

[0184] A third screening sub-module, configured to screen out each initial target product from each target candidate recommended product, where the product risk tolerance is lower than or equal to the user risk tolerance.

[0185] A fourth screening sub-module, configured to screen out a target recommended product from each initial target product, where the product subdivision type conforms to the preference label.

[0186] Each module in the above product recommendation device can be implemented in whole or in part by software, hardware, and their combination. Each of the above modules can be embedded in or independent of a processor in a computer device in the form of hardware, or stored in a memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0187] In an exemplary embodiment, a computer device is provided. The computer device can be a terminal, and its internal structure diagram can be as Figure 9 shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit, and an input device. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface, the display unit, and the input device are connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, near field communication (NFC), or other technologies. When the computer program is executed by the processor, it implements a product recommendation method. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covered on the display screen, or a button, a trackball, or a touchpad provided on the outer shell of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0188] Those skilled in the art can understand that Figure 9The structure shown is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0189] In one embodiment, a computer device is further provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the steps in the above method embodiments are implemented.

[0190] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0191] In one embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0192] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data need to comply with relevant regulations.

[0193] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in the present application can be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, data processing logics based on quantum computing, artificial intelligence (AI) processors, etc., without limitation.

[0194] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in the present application.

[0195] The above-described embodiments merely represent several implementation manners of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation to the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all fall within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. A product recommendation method, characterized in that, The method includes: Receiving a product recommendation request; the product recommendation request includes a target reference product identifier; the target reference product identifier is the product identifier of a target reference product determined based on user preference information; Obtaining each candidate recommended product from a recommended product set according to the product type of the target reference product corresponding to the target reference product identifier; Determining a target candidate recommended product sequence according to the reference product feature data set of the target reference product, the recommended product feature data sets corresponding to each candidate recommended product, and the cosine similarity algorithm; Obtaining the user feature data of the user to be recommended, and screening the target recommended product from the target candidate recommended product sequence according to the user feature data.

2. The method according to claim 1, characterized in that, Before receiving the product recommendation request, the method further includes: For each product type in the recommended product set, extracting a recommended product keyword set of each initial recommended product under the product type based on the keyword template corresponding to the product type; For each product type in the reference product set, extracting a reference product keyword set of each reference product under the product type according to the keyword template corresponding to the product type; Constructing a keyword database according to each recommended product keyword set and each reference product keyword set.

3. The method according to claim 1, characterized in that, The obtaining each candidate recommended product from the recommended product set according to the product type of the target reference product corresponding to the target reference product identifier includes: Querying the target product type of the target reference product in the reference product set according to the target reference product identifier; Querying each candidate recommended product under the target product type in the recommended product set; Querying a candidate recommended product keyword set of each candidate recommended product in the keyword database, and obtaining the recommended product feature data set of the candidate recommended product from the recommended product set according to the candidate recommended product keyword set.

4. The method according to claim 1, wherein Before determining the target candidate recommended product sequence according to the reference product feature data set of the target reference product, the recommended product feature data sets corresponding to each candidate recommended product, and the cosine similarity algorithm, the method further includes: Querying the reference product keyword set corresponding to the target reference product in the keyword database based on the target reference product identifier; Querying the target reference product data set of the target reference product in the reference product set according to the target reference product identifier; Screening the reference product feature data set corresponding to the target reference product from the target reference product data set according to the reference product keyword set.

5. The method according to claim 1, wherein The determining the target candidate recommended product sequence according to the reference product feature data set of the target reference product, the recommended product feature data sets corresponding to each candidate recommended product, and the cosine similarity algorithm includes: For each candidate recommended product, performing data operation on the recommended product feature data set of the candidate recommended product and the reference product feature data set according to the cosine similarity algorithm to obtain the cosine similarity between the candidate recommended product and the target reference product; Sorting each candidate recommended product in descending order according to each cosine similarity to obtain a candidate recommended product sequence; Determine a target candidate recommended product sequence from the candidate recommended product sequences according to preset screening conditions.

6. The method according to claim 1, wherein The user feature data includes the preference tags of the user to be recommended and the user's risk tolerance. Screening the target recommended products from the target candidate recommended product sequence according to the user feature data includes: Obtain the target candidate recommended product data sets of each target candidate recommended product in the target candidate recommended product sequence; the target candidate recommended product data set contains the product risk tolerance of the target candidate recommended product. Screen out the initial target products whose product risk tolerance is lower than and equal to the user's risk tolerance among each of the target candidate recommended products. Screen out the target recommended products whose product sub-types conform to the preference tags among each of the initial target products.

7. A product recommendation device, characterized in that, The device includes: A receiving module, configured to receive a product recommendation request; the product recommendation request contains a target reference product identifier; the target reference product identifier is the product identifier of the target reference product determined based on user preference information. An obtaining module, configured to obtain each candidate recommended product from the recommended product set according to the product type of the target reference product corresponding to the target reference product identifier. A determining module, configured to determine a target candidate recommended product sequence according to the reference product feature data set of the target reference product, the recommended product feature data sets corresponding to each candidate recommended product, and the cosine similarity algorithm. A screening module, configured to obtain the user feature data of the user to be recommended, and screen out the target recommended products from the target candidate recommended product sequence according to the user feature data.

8. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images