WeChat applet commodity recommendation method and system

Abstract

The invention relates to the technical field of commodity recommendation methods, in particular to a WeChat applet commodity recommendation method and system, I, multi-dimensional data collection and integration: collecting user behavior data, collecting search information of a user, analyzing demand intention of the user, acquiring basic information of the user by using WeChat authorization, and performing data integration and storage to obtain a WeChat applet commodity recommendation result; and integrating the collected user data, and establishing a user portrait database. According to the method, a user-based collaborative filtering recommendation algorithm and a feature-based collaborative filtering recommendation algorithm are combined, the similarity of historical data of users or articles is calculated through the combined algorithm, different users and articles are scored and predicted, the preferences of the users are mined, and the recommendation efficiency of the users is improved. And similar articles are recommended according to the preference of the similar user, so that the problems of cold start and data sparseness of a user-based algorithm and the problem of omission of a feature-based algorithm in the aspect of correlation are solved, and the commodity recommendation effect is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of commodity recommendation methods, and specifically to a WeChat applet commodity recommendation method and system. Background Art

[0002] WeChat Mini Programs are a brand-new way to connect users and services. With the globalization of the economy, when users use WeChat Mini Programs, while they gain access to the sharing of various product information resources, they also face a heavy information burden. On the one hand, a large amount of product information is available for users to choose and use, and users cannot quickly obtain all the information. On the other hand, users cannot quickly understand the product information they need. In order for customers to quickly obtain as much product information as possible that suits them, personalized recommendations are needed for individuals.

[0003] However, the current WeChat mini-program recommendation algorithm has certain limitations in its scope of application. The user-based algorithm has problems with cold start and data sparsity, while the feature-based algorithm is prone to omissions in terms of relevance, which makes it easy for the recommended items to be biased.

[0004] In view of this, this application intends to propose a WeChat mini-program product recommendation method and system. Summary of the invention

[0005] The purpose of the present invention is to provide a WeChat mini-program product recommendation method and system to solve the above-mentioned problems.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A WeChat applet product recommendation method, comprising the following steps:

[0008] Ⅰ. Multi-dimensional data collection and integration: By collecting user behavior data, collecting user search information, analyzing user demand intentions, using WeChat authorization to obtain user basic information, and integrating the collected user data through data integration and storage, a user portrait database is established;

[0009] II. Data analysis and mining: Through user-based collaborative filtering algorithms, comprehensive and in-depth analysis of user purchases and personal information in the user portrait database is performed to obtain recommended products;

[0010] III. Implement the recommendation system: recommend products based on similar users through feature-based collaborative filtering algorithms, combine the recommended products in proportion, and finally push the recommendation results to the client's WeChat applet platform;

[0011] IV. Optimize the recommendation system: Establish a real-time data processing mechanism to timely update user behavior data and product information, sort the recommendation results according to the recommendation score calculated by the recommendation algorithm, and regularly evaluate and optimize the recommendation system;

[0012] Ⅴ. User interaction and feedback: By setting up user feedback channels in the mini program, analyzing user feedback data, and utilizing WeChat’s social relationship chain, a social interaction recommendation function is introduced.

[0013] Furthermore, in step I, the multi-dimensional data collection and integration includes the following steps:

[0014] A1. Collect user behavior data: record the user's browsing history in the Mini Program, including the product details viewed, dwell time, page scrolling times, track the user's purchase history, collect the user's search keywords, dwell time, and information on products added to the shopping cart but not purchased, and analyze the user's demand intentions;

[0015] A2. Obtaining basic user information: Using WeChat authorization to obtain basic user information, and encouraging users to complete their personal information in the mini program;

[0016] A3. Data integration and storage: Integrate the collected user data and establish a user portrait database.

[0017] Further, in step II, the following steps are included:

[0018] B1. Determine the user's rating matrix Q for the product:

[0019]

[0020] Where: q r,s represents the rating of r products by the sth user;

[0021] B2. Use the cosine similarity method to determine users who are similar to the target user: For any two users, use cosine similarity to calculate the similarity, and add the average score of the users to the formula. The calculation formula is as follows:

[0022]

[0023] Where: J u,v is the set of products rated by both users u and ν. Product h belongs to this set. u,h and q v,h are the ratings of product h by user u and user ν respectively, and Respectively represent the average results of all rated products of users u and v;

[0024] B3. Sort the results of cosine similarity in descending order and select the top five or six users as similar users;

[0025] B4. Generate recommended products: For the product h to be rated by user u, it can be calculated by the following formula:

[0026]

[0027] Where: N u For the similar user set obtained in step B2, the items to be rated are arranged in descending order, and the first five or six items are set as recommended items according to needs.

[0028] Further, in step III, the following steps are included:

[0029] C1. Use the IG feature selection algorithm to select statistically significant features: For the product set A, the set B classified by category, and the product feature set D, the formula is as follows:

[0030] A={a1,a2,…,a n},B={b1,b2,…,b t},D={d1,d2,…,d s}

[0031] The entropy of a feature c in a product in the product set A is calculated using the following formula:

[0032]

[0033] Where: d i is the possible feature of c, P(d i ) is feature c and feature d i The probability of

[0034] C2. Use the IG information gain method to calculate the information entropy increment of a certain feature. The calculation formula is as follows:

[0035]

[0036] Where: P(c) and represents the proportion of products with and without feature c in the product set A, P(d i c) and Respectively in category b i Products with feature c and products without feature c in category b i The proportion of the number of goods;

[0037] C3. Calculate the features through the IG feature selection algorithm, and sort the information gain of each feature from large to small. According to the system needs, the first six features can be selected as the determined features;

[0038] C4. Use collaborative filtering algorithm to rate products according to features and determine the order of products recommended to users: Determine the rating matrix P of users' product ratings. For the existing m users, they rate the n purchased products, where the rating of the i-th user for the k-th product can be written as p i,k , then the rating matrix P is:

[0039]

[0040] C5. For target user d, the Pearson correlation coefficient is used to calculate the similarity between the product to be rated and the rated product. The calculation formula is as follows:

[0041]

[0042] Where: represents the rating of the target user d on the purchased product ci, is the average value of the ratings given by target user d to the purchased products, The predicted rating for a product that target customer d would like to rate. is the average predicted rating of the target customer for the set of items to be rated. The target user belongs to the existing m user set G;

[0043] C6. Improve the Pearson correlation coefficient by adding the product attributes into the calculation of the Pearson correlation coefficient. The characteristic difference between any two products m and n is C m,n for:

[0044]

[0045] C7. Add the product attributes to the calculation of the Pearson correlation coefficient, and the updated Pearson correlation coefficient is:

[0046] C8. The Pearson correlation coefficient can be used to determine the rating results of the products to be rated. The results of all the products to be rated are arranged in descending order. The top five or more products set in the system can be recommended to the target user, thereby completing the product recommendation;

[0047] C9. Combining the calculation results of the user-based collaborative filtering algorithm and the feature-based collaborative filtering algorithm to recommend products. The calculation method after combining the calculation results of the two algorithms is:

[0048]

[0049] Where: N is the number of products that need to be recommended, α and β are the coefficients of the feature-based collaborative filtering algorithm and the user-based collaborative filtering algorithm, respectively, and S and Y are the number of recommended products of the feature-based collaborative filtering algorithm and the user-based collaborative filtering algorithm, respectively.

[0050] Further, in step IV, the following steps are included:

[0051] D1. By establishing a real-time data processing mechanism, timely updating of user behavior data and product information, and adopting an incremental recommendation algorithm, the recommendation results can be quickly updated when user behavior changes;

[0052] D2. Sort the recommendation results according to the recommendation scores calculated by the recommendation algorithm, consider multiple factors to determine the priority of the recommendation results, and display the recommendation results in the mini program;

[0053] D3. Regularly evaluate and optimize the recommendation system, understand the accuracy of recommendations and user satisfaction through user feedback and A / B testing, and adjust the parameters of the recommendation algorithm, improve data collection methods, and optimize the presentation of recommendation results based on the evaluation results.

[0054] Further, in step V, the following steps are included:

[0055] E1. By setting up user feedback channels in the mini program, users can provide feedback on recommended products, and according to user feedback, timely adjust the recommendation results, analyze user feedback data, and understand changes in user needs and preferences;

[0056] E2. Personalization: Provides personalized settings options so that users can customize recommended content according to their needs and preferences, and allows users to adjust the weight of recommendations;

[0057] E3. Utilize WeChat’s social relationship chain and introduce social interaction recommendation function.

[0058] A WeChat applet product recommendation system includes a user information management module, a product management module, a product recommendation module, an order management module, an evaluation management module and an after-sales module:

[0059] The user information management module is used to manage the information of the user after logging in. The user can log in by setting the user name and password. According to the user's role, the administrator can set the relevant function permissions in the background and manage the user's related information;

[0060] The commodity management module is used to manage the commodities sold by merchants;

[0061] The product recommendation module recommends products to users after comprehensively analyzing the products purchased by users and their personal information by combining feature-based collaborative filtering algorithms with user-based collaborative filtering algorithms;

[0062] The order management module is used to manage the ordering and payment of goods, and to track and manage the goods after the order is placed;

[0063] The evaluation management module is used to confirm receipt of the goods and evaluate the goods after the user receives them;

[0064] The after-sales module is used to provide tracking services to users after they receive the goods.

[0065] Furthermore, the product recommendation module includes a data collection module, a data storage module, a data analysis module, an optimization recommendation module and an interaction management module:

[0066] The data collection module is used to collect user behavior data and obtain basic information of users through WeChat authorization;

[0067] The data storage module is used to integrate the user data collected by the data collection module, establish a user portrait database, store the data, encrypt the user data and perform access control;

[0068] The data analysis and recommendation module combines feature-based collaborative filtering algorithms with user-based collaborative filtering algorithms to conduct a comprehensive analysis of user purchases and personal information, and then recommends products based on similar user behaviors;

[0069] The optimization recommendation module establishes a real-time data processing mechanism to timely update user behavior data and product information, and regularly evaluates and optimizes the recommendation system;

[0070] The interaction management module introduces a social interaction recommendation function by setting up a user feedback channel in the mini program, analyzing user feedback data, and utilizing WeChat's social relationship chain.

[0071] Beneficial effects of the present invention:

[0072] 1. In the present invention, the system adopts a combination of a user-based collaborative filtering recommendation algorithm and a feature-based collaborative filtering recommendation algorithm. For the user-based algorithm, the IG feature selection algorithm is used to select product features, and then the improved Pearson correlation coefficient is used to calculate the similarity to ensure the accuracy of the similarity and obtain recommended products. For the feature-based algorithm, the improved cosine similarity is used to calculate user similarity, and products are recommended based on similar users. The recommended products are combined in proportion and finally the products are recommended. The combined algorithm can recommend similar items based on the preferences of similar users, so that users can quickly obtain effective information, solve the cold start and data sparsity problems of the user-based algorithm, and the problem that the feature-based algorithm is prone to omissions in terms of relevance, thereby ensuring the product recommendation effect.

[0073] 2. In the present invention, the optimization recommendation module establishes a real-time data processing mechanism to timely update user behavior data and product information, and can recommend the latest products in real time. It sorts the recommendation results according to the recommendation scores calculated by the recommendation algorithm, and regularly evaluates and optimizes the recommendation system. Through user feedback and A / B testing, the accuracy of recommendations and user satisfaction are understood. According to the evaluation results, the parameters of the recommendation algorithm are adjusted, the data collection method is improved, and the presentation of recommendation results is optimized. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 This is a system schematic diagram of a WeChat mini-program product recommendation method and system of the present invention.

[0075] In the figure: 1. User information management module; 2. Product management module; 3. Product recommendation module; 4. Order management module; 5. Evaluation management module; 6. After-sales module; 31. Data collection module; 32. Data storage module; 33. Data analysis module; 34. Optimization recommendation module; 35. Interaction management module. DETAILED DESCRIPTION

[0076] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0077] Example 1: Please refer to Figure 1 ,This design proposes an implementation method, a WeChat applet product recommendation method, including the following steps:

[0078] Ⅰ. Multi-dimensional data collection and integration: By collecting user behavior data, collecting user search information, analyzing user demand intentions, using WeChat authorization to obtain user basic information, and integrating the collected user data through data integration and storage, a user portrait database is established, including the following steps:

[0079] A1. Collect user behavior data: record the user's browsing history in the Mini Program, including the product details viewed, dwell time, page scrolling times, track the user's purchase history, collect the user's search keywords, dwell time, and information on products added to the shopping cart but not purchased, and analyze the user's demand intentions;

[0080] A2. Obtaining basic user information: Using WeChat authorization to obtain basic user information, and encouraging users to complete their personal information in the mini program;

[0081] A3. Data integration and storage: Integrate the collected user data and establish a user portrait database;

[0082] II. Data analysis and mining: Through the user-based collaborative filtering algorithm, a comprehensive and in-depth analysis of the user's purchased products and personal information in the user portrait database is performed to obtain recommended products, including the following steps:

[0083] B1. Determine the user's rating matrix Q for the product:

[0084]

[0085] Where: q r,s represents the rating of r products by the sth user;

[0086] B2. Use the cosine similarity method to determine users who are similar to the target user: For any two users, use cosine similarity to calculate the similarity, and add the average score of the users to the formula. The calculation formula is as follows:

[0087]

[0088] Where: J u,v is the set of products rated by both users u and ν. Product h belongs to this set. u,h and q v,h are the ratings of product h by user u and user ν respectively, and Respectively represent the average results of all rated products of users u and v;

[0089] B3. Sort the results of cosine similarity in descending order and select the top five or six users as similar users;

[0090] B4. Generate recommended products: For the product h to be rated by user u, it can be calculated by the following formula:

[0091]

[0092] Where: N u For the similar user set obtained in step B2, sort the items to be rated in descending order, and set the first five or six items as recommended items according to needs;

[0093] III. Implement the recommendation system: Use the feature-based collaborative filtering algorithm to recommend products based on similar users, combine the recommended products in proportion, and finally push the recommendation results to the client's WeChat applet platform, including the following steps:

[0094] C1. Use the IG feature selection algorithm to select statistically significant features: For the product set A, the set B classified by category, and the product feature set D, the formula is as follows:

[0095] A={a1,a2,…,a n},B={b1,b2,…,b t},D={d1,d2,…,d s}

[0096] The entropy of a feature c in a product in the product set A is calculated using the following formula:

[0097]

[0098] Where: d i is the possible feature of c, P(d i ) is feature c and feature d i The probability of

[0099] C2. Use the IG information gain method to calculate the information entropy increment of a certain feature. The calculation formula is as follows:

[0100]

[0101] Where: P(c) and represents the proportion of products with and without feature c in the product set A, P(d i c) and Respectively in category b i Products with feature c and products without feature c in category b i The proportion of the number of goods;

[0102] C3. Calculate the features through the IG feature selection algorithm, and sort the information gain of each feature from large to small. According to the system needs, the first six features can be selected as the determined features;

[0103] C4. Use collaborative filtering algorithm to rate products according to features and determine the order of products recommended to users: Determine the rating matrix P of users' product ratings. For the existing m users, they rate the n purchased products, where the rating of the i-th user for the k-th product can be written as p i,k , then the rating matrix P is:

[0104]

[0105] C5. For target user d, the Pearson correlation coefficient is used to calculate the similarity between the product to be rated and the rated product. The calculation formula is as follows:

[0106]

[0107] Where: represents the rating of the target user d on the purchased product ci, is the average value of the ratings given by target user d to the purchased products, The predicted rating for a product that target customer d would like to rate. is the average predicted rating of the target customer for the set of items to be rated. The target user belongs to the existing m user set G;

[0108] C6. Improve the Pearson correlation coefficient by adding the product attributes into the calculation of the Pearson correlation coefficient. The characteristic difference between any two products m and n is C m,n for:

[0109]

[0110] C7. Add the product attributes to the calculation of the Pearson correlation coefficient, and the updated Pearson correlation coefficient is:

[0111] C8. The Pearson correlation coefficient can be used to determine the rating results of the products to be rated. The results of all the products to be rated are arranged in descending order. The top five or more products set in the system can be recommended to the target user, thereby completing the product recommendation;

[0112] C9. Combining the calculation results of the user-based collaborative filtering algorithm and the feature-based collaborative filtering algorithm to recommend products. The calculation method after combining the calculation results of the two algorithms is:

[0113]

[0114] Where: N is the number of products to be recommended, α and β are the coefficients of the feature-based collaborative filtering algorithm and the user-based collaborative filtering algorithm, respectively, S and Y are the number of recommended products of the feature-based collaborative filtering algorithm and the user-based collaborative filtering algorithm, respectively;

[0115] IV. Optimize the recommendation system: Establish a real-time data processing mechanism to timely update user behavior data and product information, sort the recommendation results according to the recommendation score calculated by the recommendation algorithm, and regularly evaluate and optimize the recommendation system, including the following steps:

[0116] D1. By establishing a real-time data processing mechanism, timely updating of user behavior data and product information, and adopting an incremental recommendation algorithm, the recommendation results can be quickly updated when user behavior changes;

[0117] D2. Sort the recommendation results according to the recommendation scores calculated by the recommendation algorithm, consider multiple factors to determine the priority of the recommendation results, and display the recommendation results in the mini program;

[0118] D3. Regularly evaluate and optimize the recommendation system, understand the accuracy of recommendations and user satisfaction through user feedback and A / B testing, and adjust the parameters of the recommendation algorithm, improve data collection methods, and optimize the presentation of recommendation results based on the evaluation results;

[0119] V. User interaction and feedback: By setting up user feedback channels in the mini program and analyzing user feedback data, using WeChat’s social relationship chain, and introducing social interaction recommendation functions, the following steps are included:

[0120] E1. By setting up user feedback channels in the mini program, users can provide feedback on recommended products, and according to user feedback, timely adjust the recommendation results, analyze user feedback data, and understand changes in user needs and preferences;

[0121] E2. Personalization: Provides personalized settings options so that users can customize recommended content according to their needs and preferences, and allows users to adjust the weight of recommendations;

[0122] E3. Utilize WeChat’s social relationship chain and introduce social interaction recommendation function.

[0123] In this embodiment, the system adopts a combination of a user-based collaborative filtering recommendation algorithm and a feature-based collaborative filtering recommendation algorithm to recommend products. For the user-based algorithm, the IG feature selection algorithm is used to select product features, and then the improved Pearson correlation coefficient is used to calculate the similarity to ensure the accuracy of the similarity and obtain recommended products. For the feature-based algorithm, the improved cosine similarity is used to calculate user similarity, and products are recommended based on similar users. The recommended products are combined in proportion and finally the products are recommended to ensure the product recommendation effect.

[0124] Example 2: Please refer to Figure 1 , this design proposes an implementation method, a WeChat applet product recommendation system, including a user information management module 1, a product management module 2, a product recommendation module 3, an order management module 4, an evaluation management module 5 and an after-sales module 6:

[0125] The user information management module 1 is used to manage the information of the user after logging in. The user can log in by setting the user name and password. According to the user's role, the administrator can set the relevant function permissions in the background and manage the user's related information;

[0126] The commodity management module 2 is used to manage the commodities sold by the merchants;

[0127] The product recommendation module 3 combines the feature-based collaborative filtering algorithm with the user-based collaborative filtering algorithm to comprehensively analyze the user's purchased products and personal information, and recommends products to the user. It includes a data collection module 31, a data storage module 32, a data analysis module 33, an optimization recommendation module 34 and an interaction management module 35:

[0128] The data collection module 31 is used to collect user behavior data and obtain the user's basic information using WeChat authorization;

[0129] The data storage module 32 is used to integrate the user data collected by the data collection module 31, establish a user portrait database, store the data, encrypt the user data and perform access control;

[0130] The data analysis and recommendation module 33 conducts a comprehensive analysis of the user's purchased products and personal information by combining a feature-based collaborative filtering algorithm with a user-based collaborative filtering algorithm, and then recommends products based on the user's similar behaviors;

[0131] The optimization recommendation module 34 establishes a real-time data processing mechanism to timely update user behavior data and product information, and regularly evaluates and optimizes the recommendation system;

[0132] The interaction management module 35 introduces a social interaction recommendation function by setting up a user feedback channel in the mini program and analyzing the user's feedback data, using the social relationship chain of WeChat;

[0133] The order management module 4 is used to manage the ordering and payment of goods, and to track and manage the goods after the order is placed;

[0134] The evaluation management module 5 is used to confirm receipt of the goods and evaluate them after the user receives the goods;

[0135] The after-sales module 6 is used to provide tracking services to users after they receive the goods.

[0136] In the present invention, by setting up a product recommendation module 3, the data collection module 31 collects user behavior data, collects user search information, analyzes user demand intentions, obtains user basic information through WeChat authorization, integrates the collected user data through data integration and storage, and establishes a user portrait database; the data analysis and recommendation module 33 uses a user-based collaborative filtering algorithm to conduct a comprehensive and in-depth analysis of user purchases and personal information in the user portrait database to obtain recommended products, and recommends products based on similar users through a feature-based collaborative filtering algorithm, combines the recommended products in proportion, and finally pushes the recommendation results to the WeChat applet platform of the client;

[0137] The system adopts a combination of user-based collaborative filtering recommendation algorithm and feature-based collaborative filtering recommendation algorithm to recommend products. For the user-based algorithm, the IG feature selection algorithm is used to select product features, and then the improved Pearson correlation coefficient is used to calculate the similarity to ensure the accuracy of the similarity and obtain recommended products. For the feature-based algorithm, the improved cosine similarity is used to calculate the user similarity, and products are recommended according to the similar users. The recommended products are combined in proportion and finally recommended. The combined algorithm can calculate the similarity of historical data of users or items, score and predict different users and items, and mine user preferences to recommend similar items according to the preferences of similar users, so that users can quickly obtain effective information. At the same time, the α and β values ​​(the coefficient values ​​of the feature-based collaborative filtering algorithm and the user-based collaborative filtering algorithm) can be adjusted according to the number of products and users in the WeChat recommendation applet and other influencing factors to obtain the best recommendation effect, solve the cold start and data sparsity problems of the user-based algorithm, and the problem that the feature-based algorithm is prone to omissions in terms of relevance, and ensure the product recommendation effect.

[0138] At the same time, the optimization recommendation module 34 establishes a real-time data processing mechanism to timely update user behavior data and product information, and can recommend the latest products in real time. According to the recommendation score calculated by the recommendation algorithm, the recommendation results are sorted, and the recommendation system is regularly evaluated and optimized. Through user feedback and A / B testing methods, the accuracy of recommendations and user satisfaction are understood. According to the evaluation results, the parameters of the recommendation algorithm are adjusted, the data collection method is improved, and the display of recommendation results is optimized. The latest technologies and trends in the industry are continuously followed, and new recommendation algorithms and technologies are continuously introduced to improve the performance and competitiveness of the recommendation system.

[0139] In addition, the interactive management module 35 sets up a user feedback channel in the mini program, so that users can provide feedback on the recommended products, and adjust the recommendation results in time according to the user feedback to improve the accuracy of the recommendation; analyze the user's feedback data to understand the changes in user needs and preferences in order to better optimize the recommendation system; by providing personalized setting options, users can customize the recommended content according to their own needs and preferences, allowing users to adjust the weight of the recommendation to obtain recommendation results that better meet their needs; use the social relationship chain of WeChat to introduce a social interactive recommendation function to increase the credibility and fun of the recommendation, while also expanding the user's discovery range, encouraging users to share their favorite products and recommendation results, and expanding the influence and user base of the mini program through social communication.

[0140] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A WeChat applet product recommendation method, characterized in that: The following steps are involved: Ⅰ. Multi-dimensional data collection and integration: By collecting user behavior data, collecting user search information, analyzing user demand intentions, using WeChat authorization to obtain user basic information, and integrating the collected user data through data integration and storage, a user portrait database is established; II. Data analysis and mining: Through user-based collaborative filtering algorithms, comprehensive and in-depth analysis of user purchases and personal information in the user portrait database is performed to obtain recommended products; III. Implement the recommendation system: recommend products based on similar users through feature-based collaborative filtering algorithms, combine the recommended products in proportion, and finally push the recommendation results to the client's WeChat applet platform; IV. Optimize the recommendation system: Establish a real-time data processing mechanism to timely update user behavior data and product information, sort the recommendation results according to the recommendation score calculated by the recommendation algorithm, and regularly evaluate and optimize the recommendation system; Ⅴ. User interaction and feedback: By setting up user feedback channels in the mini program, analyzing user feedback data, and utilizing WeChat’s social relationship chain, a social interaction recommendation function is introduced.

2. The WeChat applet product recommendation method according to claim 1, characterized in that: In step I, the multi-dimensional data collection and integration includes the following steps: A1. Collect user behavior data: record the user's browsing history in the Mini Program, including the product details viewed, dwell time, page scrolling times, track the user's purchase history, collect the user's search keywords, dwell time, and information on products added to the shopping cart but not purchased, and analyze the user's demand intentions; A2. Obtaining basic user information: Using WeChat authorization to obtain basic user information, and encouraging users to complete their personal information in the mini program; A3. Data integration and storage: Integrate the collected user data and establish a user portrait database.

3. The WeChat applet product recommendation method according to claim 1, characterized in that: In step II, the following steps are included: B1. Determine the user's rating matrix Q for the product: Where: q r,s represents the rating of r products by the sth user; B2. Use the cosine similarity method to determine users who are similar to the target user: For any two users, use cosine similarity to calculate the similarity, and add the average score of the users to the formula. The calculation formula is as follows: Where: J u,v is the set of products rated by both users u and ν. Product h belongs to this set. u,h and q v,h are the ratings of product h by user u and user ν respectively, and Respectively represent the average results of all rated products of users u and v; B3. Sort the results of cosine similarity in descending order and select the top five or six users as similar users; B4. Generate recommended products: For the product h to be rated by user u, it can be calculated by the following formula: Where: N u For the similar user set obtained in step B2, the items to be rated are arranged in descending order, and the first five or six items are set as recommended items according to needs.

4. The WeChat applet product recommendation method according to claim 3, characterized in that: In step III, the following steps are included: C1. Use the IG feature selection algorithm to select statistically significant features: For the product set A, the set B classified by category, and the product feature set D, the formula is as follows: A={a1,a2,…,a n },B={b1,b2,…,b t },D={d1,d2,…,d s } The entropy of a feature c in a product in the product set A is calculated using the following formula: Where: d i is the possible feature of c, P(d i ) is feature c and feature d i probability; C2. Use the IG information gain method to calculate the information entropy increment of a certain feature. The calculation formula is as follows: Where: P(c) and represents the proportion of products with and without feature c in the product set A, P(d i c) and Respectively in category b i Products with feature c and products without feature c in category b i The proportion of the number of goods; C3. Calculate the features through the IG feature selection algorithm, and sort the information gain of each feature from large to small. According to the system needs, the first six features can be selected as the determined features; C4. Use collaborative filtering algorithm to rate products according to features and determine the order of products recommended to users: Determine the rating matrix P of users' product ratings. For the existing m users, they rate the n purchased products, where the rating of the i-th user for the k-th product can be written as p i,k , then the rating matrix P is: C5. For target user d, the Pearson correlation coefficient is used to calculate the similarity between the product to be rated and the rated product. The calculation formula is as follows: Where: represents the rating of the target user d on the purchased product ci, is the average value of the ratings given by target user d to the purchased products, The predicted rating for a product that target customer d would like to rate. is the average predicted rating of the target customer for the set of items to be rated. The target user belongs to the existing m user set G; C6. Improve the Pearson correlation coefficient by adding the product attributes into the calculation of the Pearson correlation coefficient. The characteristic difference between any two products m and n is C m,n for: C7. Add the product attributes to the calculation of the Pearson correlation coefficient, and the updated Pearson correlation coefficient is: C8. The Pearson correlation coefficient can be used to determine the rating results of the products to be rated. The results of all the products to be rated are arranged in descending order. The top five or more products set in the system can be recommended to the target user, thereby completing the product recommendation; C9. Combining the calculation results of the user-based collaborative filtering algorithm and the feature-based collaborative filtering algorithm to recommend products. The calculation method after combining the calculation results of the two algorithms is: Where: N is the number of products that need to be recommended, α and β are the coefficients of the feature-based collaborative filtering algorithm and the user-based collaborative filtering algorithm, respectively, and S and Y are the number of recommended products of the feature-based collaborative filtering algorithm and the user-based collaborative filtering algorithm, respectively.

5. The WeChat applet product recommendation method according to claim 1, characterized in that: In step IV, the following steps are included: D1. By establishing a real-time data processing mechanism, timely updating of user behavior data and product information, and adopting an incremental recommendation algorithm, the recommendation results can be quickly updated when user behavior changes; D2. Sort the recommendation results according to the recommendation scores calculated by the recommendation algorithm, consider multiple factors to determine the priority of the recommendation results, and display the recommendation results in the mini program; D3. Regularly evaluate and optimize the recommendation system, understand the accuracy of recommendations and user satisfaction through user feedback and A / B testing, and adjust the parameters of the recommendation algorithm, improve data collection methods, and optimize the presentation of recommendation results based on the evaluation results.

6. The WeChat applet product recommendation method according to claim 1, characterized in that: In step V, the following steps are included: E1. By setting up user feedback channels in the mini program, users can provide feedback on recommended products, and according to user feedback, timely adjust the recommendation results, analyze user feedback data, and understand changes in user needs and preferences; E2. Personalization: Provides personalized settings options so that users can customize recommended content according to their needs and preferences, and allows users to adjust the weight of recommendations; E3. Utilize WeChat’s social relationship chain and introduce social interaction recommendation function.

7. A WeChat mini program product recommendation system, applicable to the WeChat mini program product recommendation method according to any one of claims 1 to 6, characterized in that: It includes a user information management module (1), a product management module (2), a product recommendation module (3), an order management module (4), an evaluation management module (5) and an after-sales module (6): The user information management module (1) is used to manage the information of the user after logging in. The user can log in by setting the user name and password. According to the user's role, the administrator can set the relevant function permissions in the background and manage the user's related information at the same time; The commodity management module (2) is used to manage the commodities sold by the merchant; The product recommendation module (3) recommends products to users after comprehensively analyzing the products purchased by users and their personal information by combining a feature-based collaborative filtering algorithm with a user-based collaborative filtering algorithm; The order management module (4) is used to manage the ordering and payment of goods, and to track and manage the goods after the order is placed; The evaluation management module (5) is used to confirm receipt of the goods and evaluate the goods after the user receives them; The after-sales module (6) is used to provide tracking services to users after they receive the goods.

8. The WeChat applet product recommendation system according to claim 7, characterized in that: The commodity recommendation module (3) comprises a data collection module (31), a data storage module (32), a data analysis module (33), an optimization recommendation module (34) and an interaction management module (35): The data collection module (31) is used to collect user behavior data and obtain basic information of users using WeChat authorization; The data storage module (32) is used to integrate the user data collected by the data collection module (31), establish a user portrait database, store the data, encrypt the user data and perform access control; The data analysis and recommendation module (33) combines a feature-based collaborative filtering algorithm with a user-based collaborative filtering algorithm to comprehensively analyze the user's purchased products and personal information, and then recommends products based on similar behaviors of the user; The optimization recommendation module (34) establishes a real-time data processing mechanism to timely update user behavior data and product information, and regularly evaluates and optimizes the recommendation system; The interaction management module (35) introduces a social interaction recommendation function by setting up a user feedback channel in the mini program and analyzing the user's feedback data, utilizing the social relationship chain of WeChat.

Images