Exchange recommendation method, device, equipment and storage medium based on Shopify

Abstract

The present invention relates to the field of e-commerce, and discloses a method, device, equipment and storage medium for recommending exchange products based on Shopify. The method comprises: receiving an exchange application request; obtaining basic data associated with the exchange application identification through Shopify's API according to the exchange application identification; obtaining scores of exchange products on preset key indicators through a trained large model according to the basic data, and calculating the recommended weight index of the exchange products according to the scores and preset indicator weights; obtaining the priority of the exchange products according to the recommended weight index, and sorting the exchange products according to the priority, and recommending the exchange products according to the sorting. In an embodiment of the present invention, when selecting an exchange, the exchange product recommendation is realized.

Description

Technical Field

[0001] The present invention relates to the field of e-commerce, and in particular to a method, device, equipment and storage medium for recommending goods exchange based on Shopify. Background Art

[0002] Currently, Shopify, as a popular e-commerce platform, provides merchants with a range of functions through its SaaS (software as a service) model, including order processing, payment, inventory management and customer relationship management.

[0003] However, in terms of returns and exchanges, although Shopify has launched an online order processing exchange function to simplify the return and exchange process and improve customer experience, merchants still face some challenges when processing exchanges. Overall, the Shopify platform lacks an exchange function based on an intelligent recommendation system, which means that consumers cannot get personalized product recommendations when choosing to exchange. Summary of the invention

[0004] The main purpose of the present invention is to solve the technical problem of how to recommend returned goods.

[0005] A first aspect of the present invention provides a method for recommending a product replacement based on Shopify, and the method for recommending a product replacement based on Shopify comprises:

[0006] receiving a product exchange application request, wherein the product exchange application request carries a product exchange application identifier;

[0007] According to the exchange application identifier, obtain basic data associated with the exchange application identifier through Shopify's API;

[0008] Based on the basic data, the scores of the exchanged goods on various preset key indicators are obtained through the trained large model, and the recommended weight index of the exchanged goods is calculated based on the scores and the preset indicator weights;

[0009] The priority of the exchange commodity is obtained according to the recommendation weight index, and the exchange commodity is sorted according to the priority, and the exchange commodity is recommended according to the sort.

[0010] Optionally, in a first implementation of the first aspect of the present invention, obtaining scores of the exchange product on preset key indicators through a trained large model based on the basic data, and calculating the recommendation weight index of the exchange product based on the scores and preset indicator weights includes:

[0011] Based on the basic data, the scores of the exchanged goods on various preset key indicators are obtained through the trained large model;

[0012] Obtaining the user's preference data through the trained recommendation model, and obtaining the preferred exchange product based on the preference data;

[0013] According to the score and the preset indicator weight, the recommendation weight index of the preferred exchange product is calculated.

[0014] Optionally, in a second implementation of the first aspect of the present invention, obtaining the user's preference data through the trained recommendation model, and obtaining the preferred exchange product according to the preference data includes:

[0015] Acquire user behavior data, user feature data, product data, and context data, wherein the context data refers to data on factors that influence product purchases;

[0016] Obtaining user preference data through a trained recommendation model according to the user behavior data, the user feature data, the product data and the context data;

[0017] According to the preference data, a preferred exchange product is obtained.

[0018] Optionally, in a third implementation of the first aspect of the present invention, obtaining the user preference data through a trained recommendation model based on the user behavior data, the user feature data, the product data, and the context data includes:

[0019] Converting discrete features in the user feature data into numerical features to obtain a user feature vector, and normalizing the product data to obtain a product feature vector;

[0020] generating an interaction feature according to the user feature vector and the product feature vector;

[0021] According to the interaction features, the user behavior data and the context data, the user preference data is obtained through a trained recommendation model.

[0022] Optionally, in a fourth implementation of the first aspect of the present invention, before obtaining the scores of the exchanged goods on the preset key indicators through the trained large model based on the basic data, the method further includes:

[0023] Convert each preset key indicator into a numerical feature, and use a multi-layer perceptron to splice and input multiple numerical features into different initial large models;

[0024] Divide the training set and test set by cross-validation method;

[0025] According to the training set, the different initial large models are trained, and according to the test set, the training effect of the large model is evaluated through A / B testing to select the optimal large model.

[0026] Optionally, in a fifth implementation manner of the first aspect of the present invention, obtaining the priority of the exchange commodity according to the recommendation weight index, and sorting the exchange commodity according to the priority, and recommending the exchange commodity according to the sorting includes:

[0027] For products with clear reasons for return, select similar products corresponding to the returned products according to the categories of the returned products;

[0028] By extracting and splitting keywords from the titles of the similar products, key features of the similar products are obtained, and based on the key features, a list of alternative products with the highest relevance is queried through fuzzy matching search;

[0029] Obtaining the priority of the replacement product according to the replacement product list and the recommendation weight index;

[0030] The exchange commodities are sorted according to the priorities, and the exchange commodities are recommended according to the sorting.

[0031] Optionally, in a sixth implementation of the first aspect of the present invention, after sorting the exchange commodities according to the priorities and recommending the exchange merchants according to the sorting, the method further includes:

[0032] Obtain user feedback data and market change data;

[0033] According to the user feedback data and the market change data, the indicator weights are adjusted and the parameters of the large model are adjusted and updated.

[0034] A second aspect of the present invention provides a Shopify-based exchange recommendation device, comprising: a memory and at least one processor, wherein instructions are stored in the memory, and the memory and the at least one processor are interconnected via a line; the at least one processor calls the instructions in the memory so that the Shopify-based exchange recommendation device executes the above-mentioned Shopify-based exchange recommendation method.

[0035] A third aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned Shopify-based exchange recommendation method.

[0036] In an embodiment of the present invention, a request for exchange application is received, wherein the request for exchange application carries an exchange application identifier; based on the exchange application identifier, basic data associated with the exchange application identifier is obtained through Shopify's API; based on the basic data, the scores of the exchanged goods on various preset key indicators are obtained through a trained large model, and the recommended weight index of the exchanged goods is calculated based on the scores and the preset indicator weights; based on the recommended weight index, the priority of the exchanged goods is obtained, and the exchanged goods are sorted according to the priority, and the exchanged goods are recommended according to the sorting. In the present invention, based on the basic data and various preset key indicators, the recommended weight index of the exchanged goods is calculated through a trained large model, and the exchanged goods are recommended according to the priority according to the recommended weight index, so that when selecting an exchange, the exchanged goods can be recommended, so as to effectively recommend goods that meet user needs for the target market. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a schematic diagram of an embodiment of a method for recommending goods exchange based on Shopify in an embodiment of the present invention;

[0038] Figure 2 Schematic diagram of an embodiment of a device for recommending goods for replacement based on Shopify in an embodiment of the present invention;

[0039] Figure 3 It is a schematic diagram of an embodiment of a replacement recommendation device based on Shopify in an embodiment of the present invention. DETAILED DESCRIPTION

[0040] Embodiments of the present invention provide a method, device, equipment and storage medium for recommending exchange products based on Shopify.

[0041] The embodiments disclosed in the present invention will be described in more detail below with reference to the accompanying drawings. Although certain embodiments disclosed in the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as being limited to the embodiments described herein, which are instead provided for a more thorough and complete understanding of the present invention. It should be understood that the drawings and embodiments disclosed in the present invention are only for exemplary purposes and are not intended to limit the scope of protection disclosed in the present invention.

[0042] In the description of the embodiments disclosed in the present invention, the term "including" and similar terms should be understood as open inclusion, that is, "including but not limited to". The term "based on" should be understood as "based at least in part on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The terms "first", "second", etc. may refer to different or the same objects. Other explicit and implicit definitions may also be included below.

[0043] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 , an embodiment of the exchange recommendation method based on Shopify in the embodiment of the present invention includes:

[0044] S100, receiving a product exchange request.

[0045] In this embodiment, after receiving the goods, if the user is not satisfied with the quality of the goods, he can apply for a replacement. The user needs to click and select the "Replacement" option on the order details page of the Shopify platform, and send a replacement application request to the Shopify platform. The Shopify platform receives the replacement application request and processes the replacement process, wherein the replacement application request carries a replacement application identifier, which is used to identify all goods that can be replaced.

[0046] S200, according to the exchange application identifier, obtain basic data associated with the exchange application identifier through the Shopify API.

[0047] In this embodiment, the exchanged goods are obtained according to the exchange application identifier, and then the return rate, best-selling data, profit margin, product relevance, return reason and target market data of the goods are collected by using Shopify's API. Among them, Shopify's API includes Shopify's Return API, Product API, Order API, and Discount API. Return API is a return and exchange interface, which is used to create a return and exchange application and synchronize it to Shopify orders; Product API is a product parameter-related interface, which is used to obtain basic parameters of the goods such as price, product series, relevance, target market, etc.; Order API is an order-related interface, which is used to obtain order-related parameters such as sales, profit margin, return rate, etc.; Discount API is a discount coupon-related interface, which is used to create an exchange application. At the same time, the data is cleaned and sorted. Specifically, the collected data is cleaned, outliers, duplicate data and low sample data are removed, and the accuracy and completeness of the data are ensured. In the process of data cleaning, the data format needs to be unified and standardized. Then, according to the received exchange application identifier, the basic data associated with the exchange application identifier is obtained from the collected basic data.

[0048] S300, based on the basic data, obtains the scores of the exchanged goods on the preset key indicators through the trained large model, and calculates the recommended weight index of the exchanged goods based on the scores and the preset indicator weights.

[0049] In this embodiment, the key indicators include return rate index, best-selling index, profit margin index, commodity relevance index, return reason matching index, and target market matching index, wherein the return rate index is calculated by counting the proportion of the number of returns of a certain commodity in the total number of orders within a specific time; the best-selling index is determined based on the sales ranking or sales speed; the profit margin index needs to combine the cost and the selling price to calculate the profit level of a single commodity; the commodity relevance index is obtained by analyzing the purchase pattern between commodities, such as the combination of commodities that are often purchased together; the return reason matching index is obtained by analyzing the reasons for return, and matches it with the characteristics of the commodity to evaluate the possible problems of the commodity; the target market matching index is to judge whether the commodity meets the characteristics of the target market based on the user portrait and commodity positioning. The basic data associated with the exchange application identification is input into the trained large model, and the scores of the exchanged commodity on the preset key indicators can be obtained, wherein the large model can be a support vector machine model, a deep learning model, etc. The weights of each indicator are assigned according to business needs, such as emphasizing the conversion rate, giving a higher weight to the best-selling index. Finally, the score of each indicator is multiplied by its corresponding weight and summed to obtain the recommended weight index of each exchanged commodity.

[0050] S400, obtaining the priority of the exchange product according to the recommendation weight index, sorting the exchange products according to the priority, and recommending the exchange products according to the sorting.

[0051] In this embodiment, the priority of the exchanged goods is set according to the recommendation weight index. For example, a higher priority is set for goods with low return rate and good sales, or goods with good sales and high profit margin, so that these goods are recommended first. In addition, while recommending high-weight index goods, some novel or long-tail goods are appropriately introduced to increase the diversity of recommendations. Then, according to the priority, the recommended goods set is screened and sorted, and the exchanged goods are recommended according to the sorting. At the same time, the system provides personalized setting options, allowing users to adjust the recommendation results according to their preferences; merchants can customize the weights of relevant indexes and set priority rules. Merchants customize weight adjustment: Merchants can customize the weights of relevant indexes and set priority rules. In this way, merchants can flexibly adjust recommendation strategies according to market trends and their own strategies.

[0052] In an optional implementation of the first aspect of the present invention, based on the basic data, through the trained large model, the scores of the exchange goods on the preset key indicators are obtained, and according to the scores and the preset indicator weights, the recommended weight index of the exchange goods is calculated, including:

[0053] Based on the basic data, the scores of the exchange products on the preset key indicators are obtained through the trained large model; the user's preference data is obtained through the trained recommendation model, and the preferred exchange products are obtained based on the preference data; based on the scores and the preset indicator weights, the recommendation weight index of the preferred exchange products is calculated.

[0054] In this embodiment, the recommendation model mainly uses the SVD++ model. SVD++ is an improved matrix decomposition method that is widely used in recommendation systems. It adds user implicit feedback information to the traditional SVD, thereby improving the accuracy and robustness of the recommendation system. The model formula is as follows:

[0055]

[0056] Among them, μ is the global average rating, which means the average rating of all users for all items, b u represents the bias of user u, reflecting the rating tendency of user u, b u represents the bias of item i, reflecting the popularity of item i; q u is the latent factor vector of item i, which represents the representation of the item in the latent factor space; P u is the latent factor vector of user u, which represents the user's representation in the latent factor space; N(u) is the implicit feedback latent factor vector of item j, which represents the implicit feedback of user u to item j in the latent factor space; is a normalization factor used to balance the impact of different amounts of implicit feedback from users.

[0057] First, the initial recommendation model is trained using historical data to ensure that the model can learn the complex relationship between users and products. The model effect is evaluated through A / B testing, the model parameters are continuously optimized through iteration, and regularization technology is introduced to prevent overfitting and improve the generalization ability of the model. Then, based on the scores of the exchange products on various preset key indicators, the user's preference data is obtained through the trained recommendation model, the user's preferred exchange products are obtained, and the recommendation weight index of the preferred exchange products is calculated to recommend the exchange products.

[0058] In an optional implementation of the first aspect of the present invention, obtaining the user's preference data through the trained recommendation model, and obtaining the preferred exchange product according to the preference data includes:

[0059] Obtain user behavior data, user feature data, product data and context data, where context data refers to data on factors that affect product purchases; obtain user preference data through a trained recommendation model based on the user behavior data, user feature data, product data and context data; obtain preferred exchange products based on the preference data.

[0060] In this embodiment, in order to better recommend commodities that meet market demand for a specific target market, a recommendation model combining user behavior data, user feature data and context data is adopted. First, user behavior data, user feature data, commodity data and context data are collected, wherein user behavior data includes user clicks, purchases, searches, comments and other behaviors, user feature data includes gender, age, geographic location and the like, commodity data includes commodity category, price, sales volume, evaluation and the like, and context data includes time, weather, holidays and other factors that may affect purchase decisions. Then, users are divided into different groups according to user characteristics (such as geographic location, age, gender), and corresponding features are extracted for each user group, including user features, commodity features and context features. Secondly, according to the above features, the trained recommendation model is used to predict each user's preference for different commodities. Finally, weights are assigned to each indicator according to business needs, and the recommendation weight index of each preferred commodity is calculated. The commodities are sorted according to the recommendation weight index, and the top-ranked commodities are selected and recommended to the user. For example, suppose you want to recommend cosmetics to young female users in a certain city. First, screen out female users between the ages of 18 and 35 in the city, extract these users' purchase history, browsing history, search keywords and other behavioral data, then use the trained recommendation model to predict these users' preferences for different cosmetics. Finally, assign weights to preset key indicators (such as sales, reviews, prices, etc.), calculate the recommendation weight index of each cosmetic, sort them from high to low by recommendation weight index, and select the top 10 cosmetics to recommend to the target users.

[0061] In an optional implementation of the first aspect of the present invention, the user preference data obtained by training the recommendation model based on the user behavior data, the user feature data, the product data and the context data includes:

[0062] The discrete features in the user feature data are converted into numerical features to obtain the user feature vector, and the product data is normalized to obtain the product feature vector; based on the user feature vector and the product feature vector, the interaction features are generated; based on the interaction features, user behavior data and context data, the user preference data is obtained through the trained recommendation model.

[0063] In this embodiment, after collecting user behavior data, user feature data, product data and context data, feature engineering is performed on the user behavior data, user feature data, product data and context data. Specifically, discrete features in the user feature data (such as gender, age and other discrete features) are converted into numerical features, such as using one-hot encoding, to obtain user feature vectors, and continuous features such as price and sales volume of products are normalized to obtain product feature vectors, and then user features and product features are combined to generate interactive features, such as the combination of "age-product category". In addition, feature processing is also performed on the context data to create new features based on time, weather and other factors, such as holiday signs, seasons, etc.

[0064] In an optional implementation of the first aspect of the present invention, before obtaining the scores of the exchanged goods on the preset key indicators through the trained large model based on the basic data, the method further includes:

[0065] Convert each preset key indicator into numerical features, and use a multi-layer perceptron to splice multiple numerical features into different initial large models; divide the training set and test set through the cross-validation method; train different initial large models based on the training set, and evaluate the training effect of the large model through A / B testing based on the test set to select the optimal large model.

[0066] In this embodiment, the data needs to be preprocessed first, which includes steps such as data cleaning, feature selection, and normalization to ensure the quality and consistency of the data. When converting the preset key indicators into numerical features, one-hot encoding can be used to convert the indicators into numerical features. The initial large models are all modeled using multi-layer perceptrons (MLPs), and then multiple numerical features are spliced ​​and input into different initial large models. In order to prevent overfitting, a cross-validation method is used to divide the training set and the test set. Cross-validation improves the generalization ability of the model by dividing the data set multiple times and calculating the average performance index. In addition, online evaluation methods such as A / B testing can be used to compare the effects of different large models and select the optimal solution. According to business goals, the loss function can also be adjusted to optimize specific indicators. For example, if you need to increase the exposure to new users or new products, you can add corresponding weight factors to the loss function to guide the model to pay more attention to these groups.

[0067] In an optional implementation of the first aspect of the present invention, the priority of the exchange commodity is obtained according to the recommendation weight index, and the exchange commodity is sorted according to the priority, and the exchange commodity is recommended according to the sorting includes:

[0068] For products with clear return reasons, similar products corresponding to the returned products are screened out according to the classification of the returned products; the key features of the similar products are obtained by keyword extraction and splitting of the titles of the similar products, and based on the key features, a list of alternative products with the highest relevance is queried through fuzzy matching search; the priority of the replacement product is obtained based on the list of alternative products and the recommended weight index; the replacement products are sorted according to the priority, and the replacement products are recommended according to the sorting.

[0069] In this embodiment, for commodities with clear reasons for return, similar commodities corresponding to the returned commodities are screened out, and for the screened similar commodities, keywords in their titles are extracted. These keywords may include the main features of the commodities, such as color, size, material, etc. Then, using fuzzy matching technology, through edit distance or other fuzzy matching algorithms, the most relevant alternative commodities to the returned commodities are searched out, and a list of alternative commodities is generated according to the results of fuzzy matching. The commodities in this list should have a high degree of similarity with the returned commodities in key features. Combined with the list of alternative commodities and the recommendation weight index, the priority of the exchange commodities is obtained, and the exchange commodities are sorted according to the priority to ensure that the recommended commodities have a high degree of matching and user satisfaction in all aspects. Finally, the recommended exchange commodities are displayed to the user according to the sorting results.

[0070] In an optional implementation of the first aspect of the present invention, after sorting the exchange commodities according to the priorities and recommending exchangers according to the sorting, the method further includes:

[0071] Obtain user feedback data and market change data; adjust indicator weights and update parameters of the large model based on user feedback data and market change data.

[0072] In this embodiment, user feedback data is continuously collected and analyzed, model parameters are adjusted and updated according to feedback information and market change data, and indicator weights are adjusted to improve recommendation accuracy.

[0073] See also Figure 2 The second aspect of the present invention provides a product exchange recommendation device based on Shopify, and the product exchange recommendation device based on Shopify comprises:

[0074] The request receiving module 10 is used to receive a product exchange application request, wherein the product exchange application request carries a product exchange application identifier;

[0075] The basic data acquisition module 20 is used to acquire the basic data associated with the exchange application identifier through the API of Shopify according to the exchange application identifier;

[0076] The weight index calculation module 30 is used to obtain the scores of the exchange products on various preset key indicators based on the basic data through the trained large model, and calculate the recommended weight index of the exchange products based on the scores and the preset indicator weights;

[0077] The exchange commodity recommendation module 40 is used to obtain the priority of the exchange commodity according to the recommendation weight index, sort the exchange commodities according to the priority, and recommend the exchange commodities according to the sorting.

[0078] In an optional implementation of the second aspect of the present invention, the weight index calculation module 30 is also used to obtain the scores of the exchange goods on various preset key indicators based on the basic data through the trained large model; obtain the user's preference data through the trained recommendation model, and obtain the preferred exchange goods based on the preference data; calculate the recommendation weight index of the preferred exchange goods based on the score and the preset indicator weight.

[0079] In an optional implementation of the second aspect of the present invention, the weight index calculation module 30 is also used to obtain user behavior data, user feature data, product data and context data, where context data refers to data on factors that affect product purchases; based on the user behavior data, user feature data, product data and context data, the user's preference data is obtained through a trained recommendation model; based on the preference data, the preferred exchange product is obtained.

[0080] In an optional implementation of the second aspect of the present invention, the weight index calculation module 30 is also used to convert discrete features in the user feature data into numerical features to obtain a user feature vector, and normalize the product data to obtain a product feature vector; generate interaction features based on the user feature vector and the product feature vector; and obtain user preference data through a trained recommendation model based on the interaction features, user behavior data and context data.

[0081] In an optional implementation of the second aspect of the present invention, the exchange recommendation device based on Shopify also includes a model training module, which is used to convert various preset key indicators into numerical features, and use a multi-layer perceptron to splice multiple numerical features into different initial large models; divide the training set and the test set by a cross-validation method; according to the training set, train different initial large models, and according to the test set, evaluate the training effect of the large model through A / B testing to select the optimal large model.

[0082] In an optional implementation of the second aspect of the present invention, the exchange product recommendation module 40 is also used to screen out similar products corresponding to the returned products with clear return reasons according to the classification of the returned products; obtain key features of the similar products by performing keyword extraction and splitting on the titles of the similar products, and query a list of alternative products with the highest relevance through fuzzy matching search based on the key features; obtain the priority of the exchange product based on the list of alternative products and the recommendation weight index; sort the exchange products according to the priority, and recommend the exchange products according to the sorting.

[0083] In an optional implementation of the second aspect of the present invention, the exchange recommendation device based on Shopify also includes an adjustment and update module for obtaining user feedback data and market change data; adjusting indicator weights and adjusting and updating parameters of the large model based on the user feedback data and market change data.

[0084] Figure 3 1 is a schematic diagram of the structure of a Shopify-based exchange recommendation device provided by an embodiment of the present invention. The Shopify-based exchange recommendation device 500 may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 510 (for example, one or more processors) and a memory 520, and one or more storage media 530 (for example, one or more mass storage devices) storing application programs 533 or data 532. Among them, the memory 520 and the storage medium 530 may be short-term storage or permanent storage. The program stored in the storage medium 530 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the Shopify-based exchange recommendation device 500. Furthermore, the processor 510 may be configured to communicate with the storage medium 530 to execute a series of instruction operations in the storage medium 530 on the Shopify-based exchange recommendation device 500.

[0085] The Shopify-based exchange recommendation device 500 may also include one or more power supplies 540, one or more wired or wireless network interfaces 550, one or more input and output interfaces 560, and / or, one or more operating systems 531, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will appreciate that Figure 3 The structure of the Shopify-based exchange recommendation device shown does not constitute a limitation of the Shopify-based exchange recommendation device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0086] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the Shopify-based exchange recommendation method.

[0087] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0088] In addition, although each operation is described in a specific order, this should be understood as requiring such operation to be performed in the specific order shown or in a sequential order, or requiring that all illustrated operations should be performed to obtain desired results. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be interpreted as limiting the scope of the present disclosure. Some features described in the context of a separate embodiment can also be implemented in a single implementation in combination. On the contrary, the various features described in the context of a single implementation can also be implemented in multiple implementations individually or in any suitable sub-combination mode.

[0089] Although the subject matter has been described in language specific to structural features and / or methodological logical actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. On the contrary, the specific features and actions described above are merely example forms of implementing the claims.

Claims

1. A method for recommending goods exchange based on Shopify, characterized in that: The recommended method for exchange based on Shopify includes: receiving a product exchange application request, wherein the product exchange application request carries a product exchange application identifier; According to the exchange application identifier, obtain basic data associated with the exchange application identifier through Shopify's API; Based on the basic data, the scores of the exchanged goods on the preset key indicators are obtained through the trained large model, and the recommended weight index of the exchanged goods is calculated based on the scores and the preset indicator weights; The priority of the exchange commodity is obtained according to the recommendation weight index, and the exchange commodity is sorted according to the priority, and the exchange commodity is recommended according to the sort.

2. The method for recommending exchange products based on Shopify according to claim 1, characterized in that: The step of obtaining the scores of the exchanged goods on various preset key indicators through the trained large model according to the basic data, and calculating the recommended weight index of the exchanged goods according to the scores and the preset indicator weights includes: Based on the basic data, the scores of the exchanged goods on various preset key indicators are obtained through the trained large model; Obtaining the user's preference data through the trained recommendation model, and obtaining the preferred exchange product based on the preference data; According to the score and the preset indicator weight, the recommendation weight index of the preferred exchange product is calculated.

3. The method for recommending exchange products based on Shopify according to claim 2, characterized in that: The method of obtaining the user's preference data through the trained recommendation model and obtaining the preferred exchange product according to the preference data includes: Acquire user behavior data, user feature data, product data, and context data, wherein the context data refers to data on factors that influence product purchases; Obtaining user preference data through a trained recommendation model according to the user behavior data, the user feature data, the product data and the context data; According to the preference data, a preferred exchange product is obtained.

4. The method for recommending exchange products based on Shopify according to claim 3, characterized in that: The obtaining of the user preference data by training the recommendation model based on the user behavior data, the user feature data, the product data and the context data includes: Converting discrete features in the user feature data into numerical features to obtain a user feature vector, and normalizing the product data to obtain a product feature vector; generating an interaction feature according to the user feature vector and the product feature vector; According to the interaction features, the user behavior data and the context data, the user preference data is obtained through a trained recommendation model.

5. The method for recommending exchange products based on Shopify according to claim 1, characterized in that: Before obtaining the scores of the exchanged goods on the preset key indicators through the trained large model based on the basic data, the following steps are also included: Convert each preset key indicator into a numerical feature, and use a multi-layer perceptron to splice and input multiple numerical features into different initial large models; Divide the training set and test set by cross-validation method; According to the training set, the different initial large models are trained, and according to the test set, the training effect of the large model is evaluated through A / B testing to select the optimal large model.

6. The method for recommending exchange products based on Shopify according to claim 1, characterized in that: The obtaining the priority of the exchange commodity according to the recommendation weight index, sorting the exchange commodity according to the priority, and recommending the exchange commodity according to the sorting includes: For products with clear reasons for return, select similar products corresponding to the returned products according to the categories of the returned products; By extracting and splitting keywords from the titles of the similar products, key features of the similar products are obtained, and based on the key features, a list of alternative products with the highest relevance is queried through fuzzy matching search; Obtaining the priority of the replacement product according to the replacement product list and the recommendation weight index; The exchange commodities are sorted according to the priorities, and the exchange commodities are recommended according to the sorting.

7. The method for recommending exchange products based on Shopify according to claim 6, characterized in that: After the exchange commodities are sorted according to the priorities and the exchange merchants are recommended according to the sorting, the method further includes: Obtain user feedback data and market change data; According to the user feedback data and the market change data, the indicator weights are adjusted and the parameters of the large model are adjusted and updated.

8. A product exchange recommendation device based on Shopify, characterized in that: The Shopify-based exchange recommendation device includes: A request receiving module, used to receive a product exchange application request, wherein the product exchange application request carries a product exchange application identifier; A basic data acquisition module, used to acquire basic data associated with the exchange application identifier through Shopify's API according to the exchange application identifier; A weight index calculation module, used to obtain the scores of the exchange goods on various preset key indicators based on the basic data through the trained large model, and calculate the recommended weight index of the exchange goods based on the scores and preset indicator weights; The exchange commodity recommendation module is used to obtain the priority of the exchange commodity according to the recommendation weight index, and sort the exchange commodities according to the priority, and recommend the exchange commodities according to the sorting.

9. A product exchange recommendation device based on Shopify, characterized in that: The Shopify-based exchange recommendation device includes: a memory and at least one processor, the memory stores instructions, and the memory and the at least one processor are interconnected via a line; The at least one processor calls the instructions in the memory to enable the Shopify-based exchange recommendation device to execute the Shopify-based exchange recommendation method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the Shopify-based exchange recommendation method according to any one of claims 1 to 7 is implemented.

Images