Commodity recommendation method and system and storage medium
The method constructs a high-quality product pool using multi-dimensional evaluation and user behavior data to address the inefficiencies in existing recommendations, ensuring accurate and timely suggestions.
Patent Information
- Application Number
- CN202411202327.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-29
- Publication Date
- 2024-12-17
AI Technical Summary
Current product recommendation methods on shopping websites fail to account for user preference changes and do not adequately judge product quality, leading to inefficient and low-quality recommendations that reduce user satisfaction and trust.
A product recommendation method that constructs a selected product pool based on multi-dimensional evaluation values derived from transaction data and user behavior data, using algorithms like collaborative filtering, deep learning, and reinforcement learning to screen high-quality products for accurate and efficient recommendations.
Improves the accuracy and efficiency of product recommendations by ensuring only high-quality products are suggested, dynamically updating the selected product pool based on real-time transaction data, thus enhancing user satisfaction and trust in shopping websites.
Smart Images

Figure CN119151634A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of information processing technology, and in particular to a commodity recommendation method, system and storage medium. Background Art
[0002] In the prior art, when a user views product information displayed on a shopping website through the Internet and purchases products online, the shopping website often recommends products to the user based on the user's browsed products or purchase history information. However, the recommendation methods in the prior art often ignore the user's possible diverse needs and interest changes, and therefore cannot meet the user's needs for recommendations.
[0003] On the other hand, the recommendation methods in the prior art do not judge and screen the comprehensive quality of the recommended products when making product recommendations, resulting in low-quality products being mixed in with the recommended products. Users need to identify low-quality products when browsing the recommended products, which increases the time it takes for users to select products. In addition, shopping websites recommending low-quality products will also reduce user satisfaction with the shopping websites, thereby affecting user trust in the shopping websites.
[0004] Therefore, the prior art needs to be further improved. Summary of the invention
[0005] In view of the above-mentioned deficiencies in the prior art, an object of the present invention is to provide a product recommendation method, system and storage medium to solve the defects of inaccurate recommendation information and low efficiency in selecting products in the prior art.
[0006] In a first aspect, this embodiment discloses a product recommendation method, which includes:
[0007] Get product search information;
[0008] Acquire user behavior data based on product search information; wherein the user behavior data includes: user purchase product information, user browsing product information and user preference information;
[0009] Based on the product search information and the user behavior data, products are screened from a pre-established selected product pool to obtain product screening results; wherein the selected product pool is a set of products selected from the full amount of product data;
[0010] Product recommendations are displayed based on the product screening results.
[0011] Optionally, the method for constructing the selected commodity pool includes:
[0012] Obtaining transaction data of each commodity in the full commodity data; wherein the transaction data includes: commodity sales data, commodity logistics data, commodity information accuracy, and merchant service information;
[0013] Determine the multi-dimensional evaluation value of each product based on transaction data;
[0014] Filter out target products according to the multi-dimensional evaluation values corresponding to each product according to the preset screening conditions; the preset screening conditions are: filter out products with multi-dimensional evaluation values exceeding the standard or filter out products with multi-dimensional evaluation values in the front of the preset number;
[0015] A pool of selected products is constructed using the screened target products.
[0016] Optionally, the step of determining the multi-dimensional evaluation value of each commodity according to the transaction data includes:
[0017] Get the total transaction amount of each commodity in the preset time period;
[0018] Optimizing the parameters of the trained multi-dimensional evaluation model according to the total commodity transaction amount to obtain a multi-dimensional evaluation model after parameter optimization; wherein the multi-dimensional evaluation model is obtained by training based on historical transaction data of multiple commodities as training samples;
[0019] The transaction data corresponding to each commodity is input into the multi-dimensional evaluation model after parameter optimization, and the multi-dimensional evaluation value of each commodity output by the multi-dimensional evaluation model after parameter optimization is obtained.
[0020] Optionally, the step of determining the multi-dimensional evaluation value of each commodity according to the transaction data includes:
[0021] The multi-dimensional evaluation value is calculated using the preset weights corresponding to product sales data, product logistics data, product information accuracy and merchant service information.
[0022] Optionally, the step of determining the multi-dimensional evaluation value of each commodity according to the transaction data further includes:
[0023] Get updated transaction information of goods at preset intervals;
[0024] The multi-dimensional evaluation values of the products are updated according to the updated transaction information.
[0025] Optionally, the step of constructing the selected commodity pool further includes:
[0026] It is determined whether the number of commodities exceeding the standard multi-dimensional evaluation value reaches the target number. If the target number is not reached, the standard multi-dimensional evaluation value is reset or all commodities are added to the selected commodity pool.
[0027] Optionally, the step of acquiring user behavior data according to product search information includes:
[0028] Determine the user's account information based on the product search information;
[0029] Obtain information about the products purchased by the user, the products browsed by the user, and the user's preference information from the user's account information.
[0030] Optionally, the step of screening products from a pre-established selected product pool based on the product search information and the user behavior data to obtain product screening results includes:
[0031] The product search information and user behavior data are synchronously input into the recommendation algorithm model to obtain the product screening results output by the recommendation algorithm model; the recommendation algorithm model is trained on the collected historical product transaction data and historical user behavior data based on collaborative filtering algorithm, deep learning algorithm or reinforcement learning algorithm.
[0032] In a second aspect, this embodiment discloses a product recommendation system, which includes:
[0033] An information receiving module is used to obtain product search information;
[0034] A user information acquisition module is used to acquire user behavior data based on the product search information; wherein the user behavior data includes: user purchase product information, user browsing product information and user preference information;
[0035] An information search module, used to screen products from a pre-established selected product pool based on the product search information and the user behavior data, and obtain product screening results; wherein the selected product pool is a set of products selected from the full amount of product data;
[0036] The recommendation display module is used to display product recommendations based on the product screening results.
[0037] In a third aspect, this embodiment discloses a computer storage medium, wherein a product recommendation program is stored on the computer-readable storage medium, and when the product recommendation program is executed by a processor, the steps of the product recommendation method are implemented.
[0038] Beneficial effects: The present invention provides a product recommendation method, system and storage medium, which obtains product search information; obtains user behavior data based on the product search information; wherein the user behavior data includes: user purchase product information, user browsing product information and user preference information; based on the product search information and the user behavior data, screens products from a pre-established selected product pool to obtain product screening results, and displays product recommendations based on the product screening results. The method and system disclosed in this embodiment screen products based on a pre-established selected product pool. Since the selected product pool is a collection of products that meet user selection criteria and have higher quality selected from the full amount of product data, product searches based on the selected product pool can improve the accuracy of product recommendations, and the selected product pool is dynamically updated as the product transaction data changes, thereby ensuring the real-time nature of product recommendations and providing convenience for users to purchase products efficiently. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] Figure 1 It is a flowchart of the steps of a commodity recommendation method provided by the present invention;
[0040] Figure 2 It is a schematic diagram of the principle of the commodity recommendation method provided by the present invention;
[0041] Figure 3 It is a principle structural block diagram of the commodity recommendation system of the present invention. DETAILED DESCRIPTION
[0042] In order to make the purpose, technical solution and advantages of the present invention clearer and more specific, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0043] Currently, shopping websites often recommend products to users to facilitate users to quickly purchase the products they are interested in, while also increasing the shopping website's product sales. This not only increases users' shopping time, but also provides convenience for shopping malls.
[0044] However, current product recommendation methods are often based only on the user's historical shopping list, or often only on the user's browsing history, so they do not take into account the changes in user preferences. Moreover, when shopping websites recommend products, they do not judge the quality of the products. Good quality products and poorly rated products are recommended to users at the same time. Therefore, users not only need time to further screen the recommended products, but also reduce the user's credibility in the shopping website, which cannot provide users with a good shopping experience.
[0045] In order to overcome the above problems, the present invention discloses a product recommendation method, which first selects products with high comprehensive evaluations based on user evaluation information on products, sales information of products, or other rating information of products, and forms a selected product pool with products with high evaluations. When products need to be recommended to users, products are searched only from the selected product pool, thereby avoiding recommending products with poor evaluations to customers. Furthermore, when searching for products from the selected product pool, the product search is not only performed based on the product search information input by the user, but also a joint search is performed based on user behavior data to improve the accuracy of product search and recommendation, so as to provide users with more efficient product recommendations.
[0046] The method, system and storage medium disclosed in the present invention are further described in detail below with reference to the accompanying drawings.
[0047] This embodiment discloses a product recommendation method. Figure 1 As shown, including:
[0048] Step S1, obtaining product search information.
[0049] The commodity search information described in this embodiment is the commodity demand information input by the user, and the commodity search information may be commodity name, commodity category, commodity abbreviation or commodity merchant name and other commodity-related information, through which the commodity-related information can be searched for the commodity required by the user. For example, the commodity search information may be only the commodity name (e.g., notebook, pen, water cup), or the commodity category (e.g., milk, clothes), or the commodity abbreviation (e.g., MP3, CPU), or the commodity merchant name (e.g., a certain company).
[0050] Furthermore, this step is applied to a smart device with a product selling function or a smart device that can display shopping websites based on the Internet and can make product recommendations. The method for the smart device to obtain product search information is not only that the user directly inputs it through a keyboard or touch screen, but also that the user forwards it through other smart terminals, or the user issues a product search instruction through voice, and the smart device receives the voice product search instruction issued by the user, and identifies the product information contained in the product search instruction to obtain the product search information.
[0051] Step S2: Obtain user behavior data based on the product search information; wherein the user behavior data includes: user purchase product information, user browsing product information and user preference information.
[0052] After the product search information is obtained in the above step S1, the user's behavior data is obtained synchronously. Specifically, the user behavior data is the information of the user when purchasing products, including: purchase product information, browsing product information or preference information. The above information can be historical data for a period of time before the current time. In one implementation, after receiving the product search information input by the user, the order information in the user's shopping account is obtained synchronously, the items mainly included in the recent shopping list are obtained, and the product information browsed by the user is obtained from the browsing history of the account, and the preference information is analyzed based on the product information in the shopping list and the browsing history.
[0053] User preference information can include multiple aspects. First, the categories and types of goods. Users may have specific purchasing preferences for different types of goods. For example, some users may prefer to buy electronic products and clothing. Second, consumer behavior analysis. Through the goods purchased by users, their needs and preferences can be analyzed, such as preference for a certain brand and product replacement cycle. Third, user attributes, including the user's age, gender, preferred payment method, purchase frequency, etc.
[0054] Furthermore, the step of obtaining user behavior data based on product search information includes:
[0055] Determine the user's account information based on the product search information; obtain the user's purchase product information, user browsing product information and user preference information from the user's account information.
[0056] After receiving the product search information input by the user, the user's account information can be determined based on the received product search information, and the above-mentioned user behavior data can be obtained based on the user's account information.
[0057] Step S3: Screen products from a pre-established selected product pool based on the product search information and the user behavior data to obtain product screening results; wherein the selected product pool is a set of products selected from the full amount of product data.
[0058] After obtaining the user behavior data, the product search information and the user behavior data can be used together as the basis for product screening, thereby improving the accuracy of product search.
[0059] Furthermore, the product screening based on product search information and user behavior data in this step is based on screening of products in a pre-built selected product pool, rather than screening based on the full amount of product data.
[0060] The selected product pool provided in this embodiment is constructed based on the selection of full product data. When selecting high-quality products from the full product data, multi-dimensional information of different aspects is referenced to screen out highly rated products to improve the accuracy of product search. The full product data can be all product information published on the shopping website, or all product data published on the entire network.
[0061] Specifically, the steps to build a selected commodity pool include:
[0062] Step S31, obtaining transaction data of each commodity in the full commodity data; wherein the transaction data includes: commodity sales data, commodity logistics data, commodity information accuracy, and merchant service information.
[0063] When selecting the full amount of product data, the transaction data of each product in the full amount of product data is first obtained. The full amount of product data is all the products sold on the shopping website. It can be imagined that since the products sold on the shopping website will change dynamically with the inventory of the products, new products on the shelves and other information, the full amount of product data is a dynamically changing data, and the selected product pool constructed from the full amount of product data can also be a dynamically changing product set. With the changes in the transaction data of each product, its evaluation may change. As the product is out of season and removed from the shelves, the product will also be deleted from the selected product pool, so the selected product pool is also updated synchronously with the update of the data.
[0064] Specifically, the transaction data of each commodity includes: commodity sales data, commodity logistics data, commodity information accuracy, and merchant service information. The above data are all reference information for users to select commodities. Commodity sales data refers to the total sales volume or recent sales volume of the commodity, and commodity logistics data refers to the relevant information of the commodity during transportation, which can be the express information used or the commodity receipt information. The accuracy of commodity information refers to the relevant information when the shopping website introduces the commodity in detail. By obtaining the user's after-sales evaluation or shopping description of the commodity, it is judged whether the merchant's introduction of the commodity is wrong, so as to determine the accuracy of the commodity information. Furthermore, merchant service information refers to the merchant's behavior information on the commodity during the sales process, whether the merchant's pre-sales, in-sales and after-sales services are properly handled, whether service disputes arise, and other information.
[0065] Specifically, when acquiring the above-mentioned transaction data, product sales data, product logistics data, product information accuracy, merchant service information and other contents contain a large amount of text information. When processing text information, advanced NLP technology is used to understand the emotional tendencies and key information in the text. By deeply analyzing the semantic content of the comments contained in the text information, the true feelings of consumers about the products can be captured more accurately, thereby improving the quality and accuracy of the selected pool.
[0066] Step S32: Determine the multi-dimensional evaluation value of each commodity based on the transaction data.
[0067] After the transaction data of each commodity is obtained in the above step S31, a multi-dimensional evaluation value of each commodity is determined based on the transaction data. Specifically, the multi-dimensional evaluation value is a comprehensive evaluation value for the commodity. Based on the comprehensive evaluation value, the quality of the commodity, user experience satisfaction, and the functionality of the commodity can be roughly obtained.
[0068] In detail, the step of determining the multi-dimensional evaluation value of each commodity according to the transaction data includes:
[0069] Get the total transaction amount of each commodity in the preset time period;
[0070] Optimizing the parameters of the trained multi-dimensional evaluation model according to the total commodity transaction amount to obtain a multi-dimensional evaluation model after parameter optimization; wherein the multi-dimensional evaluation model is obtained by training based on historical transaction data of multiple commodities as training samples;
[0071] The transaction data corresponding to each commodity is input into the multi-dimensional evaluation model after parameter optimization, and the multi-dimensional evaluation value of each commodity output by the multi-dimensional evaluation model after parameter optimization is obtained.
[0072] In the step of determining the multi-dimensional evaluation value of each commodity based on the transaction data, the transaction data of each commodity can be directly input into the trained multi-dimensional evaluation model, and the multi-dimensional evaluation value of the commodity can be obtained based on the multi-dimensional evaluation model. In this step, in order to obtain a more accurate multi-dimensional evaluation value, the total transaction amount of each commodity in a preset time period is used to adjust and optimize the parameters of the multi-dimensional evaluation model to obtain a more accurate multi-dimensional evaluation value.
[0073] Specifically, the steps of optimizing the parameters of the trained multi-dimensional evaluation model using the total commodity transaction amount include: optimizing the model parameters using a parameter optimization algorithm or using the total commodity transaction amount as a proportional coefficient of an adjustment parameter to optimize the model parameters.
[0074] The multidimensional evaluation model is trained based on a sample set of historical commodity data. Different types of commodity information can be used as parameters that affect commodity evaluation and are given different weights. Different types of commodity information are input into the preset network model as training sample data. After the preset network model learns and evaluates each different type of information, a trained multidimensional evaluation model is obtained.
[0075] After the trained multidimensional evaluation model is obtained, the parameters of the multidimensional evaluation model are optimized based on the total transaction amount of each commodity, and then the multidimensional evaluation model with optimized parameters is used to process the transaction data of each commodity to obtain the multidimensional evaluation value corresponding to each commodity.
[0076] Step S33, according to the preset screening conditions and the multi-dimensional evaluation values corresponding to each product, the target products are screened out; the preset screening conditions are: screening out products that exceed the standard multi-dimensional evaluation value or screening out products whose multi-dimensional evaluation values are ranked in the front of a preset number.
[0077] After obtaining the multi-dimensional evaluation values corresponding to each product, the products that meet the preset screening conditions are screened out from the full amount of product data based on the multi-dimensional evaluation values of the products. The preset screening conditions can be a multi-dimensional evaluation threshold or a preset number. That is, products with multi-dimensional evaluation values higher than the preset standard multi-dimensional evaluation value are screened into the selected product pool, or products with high multi-dimensional evaluation values are screened into the selected product pool.
[0078] In addition, this step can also calculate the multi-dimensional evaluation value in another way, that is, the step of determining the multi-dimensional evaluation value of each commodity based on the transaction data can also include: calculating the multi-dimensional evaluation value using the preset weights corresponding to the commodity sales data, commodity logistics data, commodity information accuracy and merchant service information.
[0079] By using different information of commodity transaction data as evaluation criteria and assigning different weights to different information, a multi-dimensional evaluation value is obtained. Specifically, different scores are first given based on the above commodity sales data, commodity logistics data, commodity information accuracy and merchant service information, and then the multi-dimensional evaluation value is calculated based on the preset weights of the above information.
[0080] Furthermore, since the above information corresponding to the product is in a dynamic change process, the multi-dimensional evaluation value calculated based on the above transaction data can change dynamically, so the latest multi-dimensional evaluation value corresponding to each product can be calculated regularly or irregularly, and then the products in the selected product pool can be updated. Therefore, this step also includes: obtaining updated transaction information of the product at a preset time interval; updating the multi-dimensional evaluation value of the product according to the updated transaction information.
[0081] Step S34: construct a selected commodity pool with the screened target commodities.
[0082] The products screened out in the above step S33 are taken as selected products and saved into the selected product pool to construct a selected product pool. It is conceivable that when constructing the selected product pool, if the number of screened products is small, it may not be possible to make good product recommendations for users. Therefore, the step of constructing the selected product pool also includes:
[0083] It is determined whether the number of commodities exceeding the standard multi-dimensional evaluation value reaches the target number. If the target number is not reached, the standard multi-dimensional evaluation value is reset or all commodities are added to the selected commodity pool.
[0084] If the number of filtered products is too small to meet your needs, you can add all products to the selected product pool, or directly search for products in the full range of products.
[0085] Furthermore, after the selected commodity pool is constructed, the step of screening commodities from the pre-established selected commodity pool based on the commodity search information and the user behavior data to obtain the commodity screening result includes:
[0086] The product search information and user behavior data are synchronously input into the recommendation algorithm model to obtain the product screening results output by the recommendation algorithm model; the recommendation algorithm model is trained on the collected historical product transaction data and historical user behavior data based on collaborative filtering algorithm, deep learning algorithm or reinforcement learning algorithm.
[0087] In this step, the pre-trained recommendation algorithm model is used to implement the selection of products. The recommendation algorithm model evaluates and selects each product in the selected product pool in turn based on the received product search information and user behavior data, thereby obtaining a product selection result.
[0088] Specifically, the recommendation algorithm model is obtained by training based on a collaborative filtering algorithm, a deep learning algorithm or a reinforcement learning algorithm.
[0089] Step S4: display product recommendations based on the product screening results.
[0090] After the product screening results are obtained in the above step S3, the recommended products are displayed to the user. When displaying the recommended products, the products with the highest matching degree with the product search information and the closest to the favorite preference are displayed first, and then other products are displayed separately, so that the user can quickly view the purchase recommendations of the required products, thereby improving the efficiency of the user's purchase of products.
[0091] Combine the following Figure 2 As shown, the method provided in this embodiment is further described.
[0092] When the user opens the trading platform of the client, there will be a recommended product section on the trading platform. The product information displayed in the recommended product section is recommended to the user after screening the products in the selected product pool based on the recommendation algorithm model.
[0093] The recommendation algorithm model can be based on collaborative filtering algorithm, deep learning algorithm or intensity learning algorithm, and is trained with user historical purchase information as sample information. The recommendation algorithm model also includes user preference measurement management, that is, combining user behavior data, and taking user long-term shopping preferences, short-term shopping preferences, browsing preferences and transaction preferences as user behavior data. Since the user behavior data contains user preferences, more accurate products can be recommended to users.
[0094] The recommendation algorithm model recommends products based on the selected product pool, and the products in the selected product pool are selected based on multi-dimensional evaluation values obtained based on comprehensive evaluations of different aspects, and are screened based on multi-dimensional evaluation values, such as: DSR score (store dynamic score), product sales and comments, etc., and the multi-dimensional evaluation values will be updated according to whether it is a hot product or a new product to obtain the final screening result. In one implementation, the calculation formula of the DSR score mainly involves four dimensions: on-time delivery, accuracy of item description, seller service attitude and delivery speed, and the weight of each dimension is different. For example: the weight of the on-time delivery dimension is 20%. The weight of the item description accuracy is 50%. The weight of the seller's service attitude is 20%. The weight of the delivery speed is 10%. In order to ensure more accuracy, the DSR score of the product is the DSR score received by all products within 90 days ÷ 90 days.
[0095] The method provided in this embodiment can also be used for e-commerce or other trading formats, where there will be paid promoted products or services for advertisers, to build a whitelist strategy for advertisers, support the selection of whitelisted products or services from the full product pool or premium product pool, and push the whitelisted products directly into the selected pool through the whitelist setting.
[0096] In order to facilitate the screening of goods and obtain a pool of selected goods, in the specific implementation, the sources of information can be divided into a product library and a content library. The product library is used to obtain direct information related to each product, and the content library is used to obtain user behavior data. Then, evaluations can be made separately for the information obtained from different information libraries, and a comprehensive analysis can be conducted to obtain the final product screening results.
[0097] The method of this embodiment expands and optimizes the traditional DSR scoring through advanced data processing and machine learning technologies, such as including innovative algorithm designs, such as adaptive learning mechanisms, and automatically adjusting the scoring weights to reflect the latest user preferences and market trends. In addition, the method of this embodiment also integrates multi-dimensional data, not just simple numerical scores, but also combines user behavior data, and performs sentiment analysis on the evaluation text of the product based on user behavior data, providing a more comprehensive and in-depth product evaluation. Moreover, the method of this embodiment also adds a dynamic update and self-optimization mechanism for product evaluation, which can automatically adjust the scoring criteria and the products in the selected product pool according to real-time data and feedback, so it has strong real-time performance.
[0098] In a second aspect, this embodiment discloses a product recommendation system, such as Figure 3 As shown, including:
[0099] The information receiving module 310 is used to obtain commodity search information; its function is as described in step S1.
[0100] The user information acquisition module 320 is used to acquire user behavior data according to the product search information; wherein the user behavior data includes: user purchase product information, user browsing product information and user preference information; its function is as described in step S2.
[0101] The information search module 330 is used to screen products from a pre-established selected product pool based on the product search information and the user behavior data to obtain product screening results; wherein the selected product pool is a set of products selected from the full amount of product data; its function is as described in step S3.
[0102] The recommendation display module 340 is used to display product recommendations based on the product screening results, and its function is as described in step S4.
[0103] The system also includes: a selected commodity pool construction module.
[0104] The selected commodity pool construction module is used to realize the construction of the selected commodity pool, which is specifically used to obtain the transaction data of each commodity in the full commodity data; wherein the transaction data includes: commodity sales data, commodity logistics data, commodity information accuracy, merchant service information; determine the multi-dimensional evaluation value of each commodity based on the transaction data; filter out the target commodities according to the multi-dimensional evaluation values corresponding to each commodity according to the preset filtering conditions; the preset filtering conditions are: filter out commodities that exceed the standard multi-dimensional evaluation value or filter out commodities whose multi-dimensional evaluation values are in the front preset number; construct a selected commodity pool with the filtered target commodities.
[0105] Furthermore, the selected commodity pool construction module also includes: a multi-dimensional evaluation value calculation unit.
[0106] The multi-dimensional evaluation value calculation unit is used to obtain the total transaction amount of each commodity in a preset time period; optimize the parameters of the trained multi-dimensional evaluation model according to the total transaction amount of the commodities to obtain the multi-dimensional evaluation model after parameter optimization; wherein the multi-dimensional evaluation model is trained based on the historical transaction data of multiple commodities as training samples; the transaction data corresponding to each commodity is input into the multi-dimensional evaluation model after parameter optimization to obtain the multi-dimensional evaluation value of each commodity output by the multi-dimensional evaluation model after parameter optimization.
[0107] In addition, the multi-dimensional evaluation value calculation unit is also used to calculate the multi-dimensional evaluation value based on the preset weights corresponding to the commodity sales data, commodity logistics data, commodity information accuracy and merchant service information.
[0108] The selected commodity pool construction module further includes: an updating unit and a quantity judging unit.
[0109] The updating unit is used to obtain updated transaction information of the commodity at preset intervals; and update the multi-dimensional evaluation value of the commodity according to the updated transaction information.
[0110] The number judgment unit is used to judge whether the number of commodities exceeding the standard multi-dimensional evaluation value reaches the target number. If the target number is not reached, the standard multi-dimensional evaluation value is reset or all commodities are added to the selected commodity pool.
[0111] The step of acquiring user behavior data according to the product search information in the information acquisition module specifically includes: determining the user's account information according to the product search information; and acquiring the user's product purchase information, user's product browsing information and user preference information from the user's account information.
[0112] The information search module includes: a commodity screening unit;
[0113] The commodity screening unit is used to synchronously input commodity search information and user behavior data into the recommendation algorithm model to obtain the commodity screening results output by the recommendation algorithm model; the recommendation algorithm model is obtained by training the collected historical commodity transaction data and historical user behavior data based on collaborative filtering algorithm, deep learning algorithm or reinforcement learning algorithm.
[0114] In a third aspect, this embodiment discloses a computer storage medium, wherein a product recommendation program is stored on the computer-readable storage medium, and when the product recommendation program is executed by a processor, the steps of the product recommendation method are implemented.
[0115] The present invention provides a product recommendation method, system and storage medium, which obtains product search information and obtains user behavior data based on the product search information; based on the product search information and the user behavior data, products are screened from a pre-established selected product pool to obtain product screening results, and product recommendations are displayed based on the product screening results. The method and system disclosed in this embodiment screen products based on a pre-established selected product pool. Since the selected product pool is a collection of products that meet user selection criteria and have higher quality selected from the full amount of product data, product searches based on the selected product pool can improve the accuracy of product recommendations, and the selected product pool is dynamically updated as the product transaction data changes, thereby ensuring the real-time nature of product recommendations and providing convenience for users to purchase products efficiently.
[0116] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.
[0117] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise clearly and specifically defined.
[0118] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or N executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes additional implementations, in which the order shown or discussed may not be followed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.
[0119] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can read instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or N wirings (electronic devices), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically by optically scanning the paper or other medium and then editing, interpreting or otherwise processing in a suitable manner if necessary and then storing it in a computer memory.
[0120] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above embodiment, the N steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.
[0121] A person skilled in the art may understand that all or part of the steps in the above-mentioned embodiment method may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.
[0122] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.
[0123] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.
Claims
1. A product recommendation method, characterized in that: include: Get product search information; Acquire user behavior data based on product search information; wherein the user behavior data includes: user purchase product information, user browsing product information and user preference information; Based on the product search information and the user behavior data, products are screened from a pre-established selected product pool to obtain product screening results; wherein the products contained in the selected product pool are products selected from the full amount of product data; Product recommendations are displayed based on the product screening results.
2. The commodity recommendation method according to claim 1, characterized in that: The method for constructing the selected commodity pool includes: Obtaining transaction data of each commodity in the full commodity data; wherein the transaction data includes: commodity sales data, commodity logistics data, commodity information accuracy, and merchant service information; Determine the multi-dimensional evaluation value of each product based on transaction data; Filter out target products according to the multi-dimensional evaluation values corresponding to each product according to the preset screening conditions; the preset screening conditions are: filter out products with multi-dimensional evaluation values exceeding the standard or filter out products with multi-dimensional evaluation values in the front of the preset number; A pool of selected products is constructed using the screened target products.
3. The commodity recommendation method according to claim 2, characterized in that: The step of determining the multi-dimensional evaluation value of each commodity according to the transaction data comprises: Get the total transaction amount of each commodity in the preset time period; Optimizing the parameters of the trained multi-dimensional evaluation model according to the total commodity transaction amount to obtain a multi-dimensional evaluation model after parameter optimization; wherein the multi-dimensional evaluation model is obtained by training based on historical transaction data of multiple commodities as training samples; The transaction data corresponding to each commodity is input into the multi-dimensional evaluation model after parameter optimization, and the multi-dimensional evaluation value of each commodity output by the multi-dimensional evaluation model after parameter optimization is obtained.
4. The commodity recommendation method according to claim 2, characterized in that: The step of determining the multi-dimensional evaluation value of each commodity according to the transaction data comprises: The multi-dimensional evaluation value is calculated using the preset weights corresponding to product sales data, product logistics data, product information accuracy and merchant service information.
5. The commodity recommendation method according to claim 3 or 4, characterized in that: The step of determining the multi-dimensional evaluation value of each commodity according to the transaction data also includes: Get updated transaction information of goods at preset intervals; The multi-dimensional evaluation values of the products are updated according to the updated transaction information.
6. The commodity recommendation method according to claim 2, characterized in that: The method for constructing the selected commodity pool further includes the steps of: It is determined whether the number of commodities exceeding the standard multi-dimensional evaluation value reaches the target number. If the target number is not reached, the standard multi-dimensional evaluation value is reset or all commodities are added to the selected commodity pool.
7. The commodity recommendation method according to claim 1, characterized in that: The step of obtaining user behavior data based on product search information includes: Determine the user's account information based on the product search information; Obtain information about the products purchased by the user, the products browsed by the user, and the user's preference information from the user's account information.
8. The commodity recommendation method according to claim 1, characterized in that: The step of screening products from a pre-established selected product pool based on the product search information and the user behavior data to obtain product screening results includes: The product search information and user behavior data are synchronously input into the recommendation algorithm model to obtain the product screening results output by the recommendation algorithm model; the recommendation algorithm model is trained on the collected historical product transaction data and historical user behavior data based on collaborative filtering algorithm, deep learning algorithm or reinforcement learning algorithm.
9. A product recommendation system, characterized in that: include: An information receiving module is used to obtain product search information; A user information acquisition module is used to acquire user behavior data based on the product search information; wherein the user behavior data includes: user purchase product information, user browsing product information and user preference information; An information search module, used to screen products from a pre-established selected product pool based on the product search information and the user behavior data, and obtain product screening results; wherein the selected product pool is a set of products selected from the full amount of product data; The recommendation display module is used to display product recommendations based on the product screening results.
10. A computer storage medium, characterized in that: The computer-readable storage medium stores a product recommendation program, and when the product recommendation program is executed by a processor, the steps of the product recommendation method according to any one of claims 1 to 8 are implemented.
Citation Information
Cited By
Product screening display method and platform based on user portrait
CN119477486A
Personalized commodity recommendation method and system based on multi-modal data fusion
CN120387875A
Personalized commodity recommendation method and system based on multi-modal data fusion
CN120387875B
