Information Recommendation and Acquisition Method, Device, and Storage Medium

Abstract

Embodiments of the present application provide a method, device, and storage medium for information recommendation and acquisition. In the embodiments of the present application, for inventory resources that can be traded online, the front-end recommendation is combined with the inventory information at the supply chain end, and at the same time, multi-dimensional information such as the shortage degree information of inventory resources, the current inventory information, and the preference information of target users for inventory resources is integrated to implement an information recommendation method based on inventory balance. This method can incorporate inventory information from the perspective of supply chain management with almost no increase in computational burden, and can perform personalized recommendations for users from a global perspective, which is beneficial to more accurately recommend inventory resources to users, reduce the probability of recommending invalid resources, improve the user experience, and at the same time increase the overall retail revenue.

Description

Technical Field

[0001] The present application relates to the field of Internet technology, and in particular to an information recommendation and acquisition method, device and storage medium. Background Art

[0002] The rapid development of the retail industry has given rise to a new retail model that integrates all channels and multiple scenarios. Under the new retail model, machine learning-based methods are increasingly used in the estimation of product click-through rates (CTR), click-through conversion rates (CVR), or gross merchandise volume (GMV), continuously improving the estimation accuracy; after the estimation is completed, personalized product recommendations can be made for users based on multi-objective requirements, attempting to maximize indicators such as CTR, CVR, or GMV.

[0003] At present, the common practice is to estimate the click probability of products based on historical data, such as CTR or CVR, sort the products according to the click probability, and select the products with the highest click probability for recommendation to users. However, due to the randomness of users' access to apps or web pages, it may not be possible to accurately recommend products to users based only on click probability. For example, invalid products may be recommended to users, resulting in users being unable to purchase the recommended products, which reduces the user experience. Summary of the invention

[0004] Multiple aspects of the present application provide an information recommendation and acquisition method, device and storage medium for more accurately recommending inventory resources to users, reducing the probability of recommending invalid resources, and improving user experience.

[0005] An embodiment of the present application provides an information recommendation method, comprising: receiving a page request sent by a terminal device, the page request including a user identifier, the user identifier being used to identify a target user who initiates the page request operation; obtaining scarcity information and current inventory information of at least one inventory resource that can be traded online, and predicting preference information of the target user for the at least one inventory resource; selecting a target inventory resource from the at least one inventory resource according to the scarcity information, the current inventory information of the at least one inventory resource, and the preference information of the target user for the at least one inventory resource; and sending information of the target inventory resource to the terminal device, so that the terminal device displays the information of the target inventory resource on a page requested by the target user.

[0006] The embodiment of the present application also provides an information acquisition method, comprising: responding to a page request operation, sending a page request to a server device, the page request including a user identifier, the user identifier being used to identify a target user who initiates the page request operation; receiving information about a target inventory resource returned by the server device, and displaying the information about the target inventory resource on a page requested by the target user; wherein the target inventory resource is selected by the server device based on scarcity information of at least one inventory resource that can be traded online, current inventory information, and preference information of the target user for the at least one inventory resource.

[0007] The embodiment of the present application also provides a server-side device, comprising: a memory and a processor; the memory is used to store a computer program or instruction; the processor is coupled to the memory and is used to execute the computer program or instruction, so as to: receive a page request sent by a terminal device, the page request includes a user identifier, and the user identifier is used to identify a target user who initiates the page request operation; obtain scarcity information and current inventory information of at least one inventory resource that can be traded online, and predict the target user's preference information for the at least one inventory resource; select a target inventory resource from the at least one inventory resource according to the scarcity information of the at least one inventory resource, the current inventory information and the target user's preference information for the at least one inventory resource; and send the information of the target inventory resource to the terminal device, so that the terminal device displays the information of the target inventory resource on the page requested by the target user.

[0008] An embodiment of the present application also provides a terminal device, comprising: a memory, a processor and a display; the memory is used to store computer programs or instructions; the processor is coupled to the memory and is used to execute the computer programs or instructions, so as to: respond to a page request operation, send a page request to a server device, the page request includes a user identifier, and the user identifier is used to identify a target user who initiates the page request operation; receive information about a target inventory resource returned by the server device, and display the information about the target inventory resource on a page requested by the target user; the display is used to display the page requested by the target user; wherein the target inventory resource is selected by the server device according to scarcity information of at least one inventory resource that can be traded online, current inventory information, and preference information of the target user for the at least one inventory resource.

[0009] The embodiment of the present application also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it can cause the processor to implement the steps in the information recommendation or acquisition method provided in the embodiment of the present application.

[0010] An embodiment of this application also provides a computer program product, including computer programs / instructions, which, when executed by a processor, cause the processor to implement the steps in the information recommendation or acquisition method provided by the embodiment of this application.

[0011] In the embodiment of this application, for inventory resources that can be traded online, the front-end recommendation is combined with the inventory information on the supply chain side, and at the same time, multi-dimensional information such as the shortage degree information of inventory resources, the current inventory information, and the preference information of target users for inventory resources is integrated to implement an information recommendation method based on inventory balance. This method can incorporate inventory information from the perspective of supply chain management with almost no increase in computational burden, and can perform personalized recommendations for users from a global perspective, which is beneficial to more accurately recommend inventory resources to users, reduce the probability of recommending invalid resources, improve the user experience, and at the same time increase the overall retail revenue. Description of the Drawings

[0012] The drawings described herein are used to provide a further understanding of this application, and constitute a part of this application. The illustrative embodiments of this application and their descriptions are used to explain this application, and do not constitute an improper limitation of this application. In the drawings:

[0013] Figure 1 It is a schematic structural diagram of a transaction data processing system provided by an exemplary embodiment of this application;

[0014] Figure 2 It is a schematic flow diagram of estimating the shadow price of a commodity provided by an exemplary embodiment of this application;

[0015] Figure 3 It is a schematic flow diagram of an information recommendation method provided by an exemplary embodiment of this application;

[0016] Figure 4 It is a schematic flow diagram of an information acquisition method provided by an exemplary embodiment of this application;

[0017] Figure 5 It is a schematic structural diagram of a server device provided by an exemplary embodiment of this application;

[0018] Figure 6 It is a schematic structural diagram of a terminal device provided by an exemplary embodiment of this application. Detailed Embodiments

[0019] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments of this application and the corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in this application without creative efforts belong to the scope of protection of this application.

[0020] Under the existing new retail model, there is a problem that it is impossible to accurately recommend products to users when recommending products to users. To address the problems of the prior art, an information recommendation method is provided in an embodiment of this application. This method can not only perform personalized recommendations for products that can be traded online and have inventory information in the e-commerce field, but also be used for personalized recommendations for other resource objects that can be traded online and have inventory information. In the embodiments of this application, products that can be traded online and have inventory information in the e-commerce field, as well as other resource objects that can be traded online and have inventory information, are collectively referred to as inventory resources. For inventory resources that can be traded online, the front-end recommendation is combined with the inventory information at the supply chain end, and at the same time, multi-dimensional information such as the shortage degree information of inventory resources, the current inventory information, and the preference information of target users for inventory resources is integrated to implement an information recommendation method based on inventory balance. This method can incorporate inventory information from the perspective of supply chain management with almost no increase in computational burden, and can perform personalized recommendations for users from a global perspective, which is beneficial to more accurately recommend inventory resources to users, reduce the probability of recommending invalid resources, and improve the user experience.

[0021] The following will detail the technical solutions provided by each embodiment of this application in conjunction with the drawings.

[0022] Figure 1 It is a schematic structural diagram of a transaction data processing system provided for an exemplary embodiment of this application. As Figure 1 shown, the transaction data processing system 100 includes: a terminal device 101, a server device 102, and an inventory management device 103; the server device 102 is communicatively connected to the terminal device 101 and the inventory management device 103.

[0023] Among them, the communication connection between the server device 102 and the terminal device 101 and the inventory management device 103 can be a wireless connection method or a wired connection method. If the server device 102 communicates with the terminal device 101 or the inventory management device 103 through a mobile network, the network mode of this mobile network can be any one of 2G (GSM), 2.5G (GPRS), 3G (WCDMA, TD-SCDMA, CDMA2000, UTMS), 4G (LTE), 4G+ (LTE+), 5G, WiMax, etc.

[0024] In this embodiment, the transaction data processing system 100 can provide some tradable resource objects for users. The transactions include, but are not limited to: purchase, procurement, exchange, or redemption, etc. The resource objects can be various commodities in the e-commerce field, or other goods, raw materials, or labor resources, electronic video resources, etc. that can be traded online. These resource objects all have a quantity attribute, and the resource warehouse 104 can store and manage these resource objects. Therefore, these resource objects can be called inventory resources. If the resource object is a physical object such as various tangible commodities, goods, or raw materials, the resource warehouse can be a physical warehouse or accommodation space set at certain locations; if the resource object is a virtual object such as an electronic video resource, the resource warehouse can be a storage space with certain information storage or recording functions, such as a disk, hard disk, or database, etc. Among them, the inventory management device 103 corresponds to the resource warehouse 104 and is mainly responsible for maintaining and managing the inventory information of the inventory resources stored in the resource warehouse 104. For example, it can dynamically update the current inventory information according to the transaction process of the inventory resources, and is responsible for managing replenishment-related matters and information, such as determining the replenishment time, replenishment quantity, and issuing replenishment notifications, etc. The inventory information of the inventory resources mainly refers to the inventory quantity of the inventory resources.

[0025] In this embodiment, with the mutual cooperation of the terminal device 101 and the server device 102, users can perform transaction operations on the inventory resources provided by the transaction data processing system 100. Here, the transaction operations include online transaction operations or online-offline combined transaction operations. Among them, the terminal device used by the user can be, for example, a smart phone, a tablet computer, a personal computer, a wearable device, etc. An application software for users to perform online and / or offline transactions is installed on the terminal device 101. The application software can be an application program (APP), a client, a mini-program, or an SDK, etc. The terminal device 101 running the application software can provide users with online and / or offline transaction functions. For example, taking the commodities in the e-commerce field as an example, users can initiate online transaction operations through the APP on the terminal device 101. For example, users can select commodities online, add them to the shopping cart, place an order online, and make an online payment, etc., and can also post evaluations online. Or, users can select offline commodities in an offline store and scan the QR code or barcode information of the selected offline commodities through the APP scanning function on the terminal device 101 to obtain attribute information such as the price of the offline commodities, and form an electronic order for online payment to achieve an online-offline combined transaction operation. Or, users can select commodities sold in an offline store online, place an order online, complete the payment, and then the store staff is responsible for picking up the goods in the offline store and delivering them to the users to achieve an online-offline combined transaction operation.

[0026] The server device 102 may be a server that processes resource transactions in a network virtual environment, and generally refers to a server that uses the network to conduct online resource transactions. Users generally need to register their identity information on the server to use the registered account to purchase and redeem inventory resources and other transactions. For example, it may be a transaction server of various e-commerce platforms or online transaction websites, or a third-party server. In terms of physical implementation, the server device 102 may be any device that can provide computing services, respond to service requests, and perform processing, such as a conventional server, cloud server, cloud host, virtual center, server array, etc. The server is mainly composed of a processor, a hard disk, a memory, a system bus, etc., which is similar to a general computer architecture.

[0027] In this embodiment, for any type of inventory resources, the process in which the user conducts online transactions or online and offline combined transactions through the terminal device 101 and the server device 102 is the same or similar. For ease of understanding and description, in this system embodiment, the working principle of the system 100 of this embodiment is explained by taking the inventory resources as commodities in the e-commerce field as an example. In the e-commerce field, based on the system 100 of this embodiment, the user can browse the commodity information provided by the merchant online through the terminal device 101, and determine which commodities to purchase online. Alternatively, in the case of offline shopping, the user can also browse the commodity information provided by the merchant online through the terminal device 101, and then purchase the desired commodities in the offline store. Regardless of the shopping method, in order to enable users to obtain the product information they need from a large amount of product information through the terminal device 101 more conveniently and efficiently, the server device 102 also has an information recommendation function, which can recommend product information that better meets the user's needs to the user through the terminal device 101, thereby reducing the amount of product information that the user needs to browse. In this way, the user does not need to browse and select one by one from a large number of library product information, but can directly select from the small amount of recommended product information, thereby spending less time and more efficiently selecting the required product information, which is conducive to improving user experience.

[0028] Specifically, if Figure 1 As shown in ①, the user initiates a page request operation through the terminal device 101. In response to the page request operation, the terminal device 101 sends a page request to the server device 102, and carries a user identifier in the page request. The user identifier is used to identify the user who initiated the page request operation, so that the server device 102 can recommend personalized information to the user. Figure 1 As shown in ②-⑤ in FIG, the server device 102 makes personalized product recommendations for the user and returns the information of the recommended target product to the terminal device 101. The personalized recommendation process will be described later. Figure 1As shown in ⑥ in [reference], after the terminal device 101 receives the information of the target commodity returned by the server device 102, it displays the requested page to the user. At the same time, the information of the target commodity is displayed on the page requested by the user. The probability of the target commodity containing the commodities liked or needed by the user is higher, which is convenient for the user to quickly select and purchase the required commodities from the target commodities, saving the time for browsing commodities and improving the shopping efficiency and experience.

[0029] Further optionally, when the terminal device 101 displays the information of the target commodity on the page requested by the user, it can adopt a random display method to display the information of the target commodity on the page requested by the user; or, it can also display the information of the target commodity on the page requested by the user in order from high to low according to the user's preference degree for the target commodity; or, it can also display the information of the target commodity on the page requested by the user in order from low to high according to the shortage degree of the target commodity; or, it can also display the information of the target commodity on the page requested by the user in order from high to low according to the current inventory information of the target commodity; or, it can also combine at least two of the current inventory information, shortage degree information, and user preference information of the target commodity to calculate the priority of the target commodity, and display the information of the target commodity on the page requested by the user in order from high to low according to the priority of the target commodity.

[0030] In this embodiment, a shopping application software is installed on the terminal device 101. By running this application software, the terminal device 101 can provide an application page to the user. According to the type of the application software, the application page can be an APP page, a mini-program page, or a browser page. In an optional embodiment, for any type of application software, the recommendation function can be embedded in each page provided by the application software, that is, no matter which page the user requests, the terminal device 101 and the server device 102 cooperate with each other to perform personalized commodity recommendation for the user, and the terminal device 101 displays the information of the recommended target commodity on the page requested by the user. In addition, the recommendation function can also be embedded in some pages, so that only when the user requests a specific page embedded with the recommendation function, the terminal device 101 and the server device 102 will cooperate with each other to perform personalized commodity recommendation for the user, and the terminal device 101 displays the information of the recommended target commodity on the specific page requested by the user. In this embodiment, the specific page embedded with the recommendation function is not limited. For example, it can be the home page, shopping cart page, group page, or user details page provided by the application software. The following is an example:

[0031] For example, a recommendation function can be embedded on the application's home page. When the user requests the application's home page, the terminal device 101 can respond to the operation of the user initiating the request for the application's home page, send a page request to the server device 102 to request the server device 102 to perform personalized information recommendation for the user and return the information of the recommended target products; the terminal device 101 renders the application's home page, and while displaying the application's home page to the user, displays the information of the target products returned by the server device 102 on the application's home page to guide the user to purchase the required products from the target products. Among them, an example of the application's home page displaying the information of the target products is as follows Figure 1 shown. It should be noted that the user can click on the icon of the application software on the terminal device 101 to request to start the application software. If the application software defaults to enter the application's home page when starting through the icon of the application software, then the user clicking on the icon of the application software is equivalent to initiating the operation of requesting the application's home page. Or, in the case where the application software provides a home page jump function, the user can also click on the navigation label pointing to the home page on the current application page to initiate the operation of jumping to the application's home page, and this operation is also equivalent to initiating the operation of requesting the application's home page.

[0032] Another example is that a recommendation function can be embedded on the shopping cart page. When the user requests the shopping cart page, the terminal device 101 can respond to the operation of the user initiating the request for the shopping cart page, send a page request to the server device 102 to request the server device 102 to perform personalized information recommendation for the user and return the information of the recommended target products; the terminal device 101 renders the shopping cart page, and while displaying the shopping cart page to the user, displays the information of the target products returned by the server device 102 on the shopping cart page to guide the user to purchase the required products from the target products. Among them, an example of the shopping cart page displaying the information of the target products is as follows Figure 1 shown. Optionally, in the case where the application software provides a shopping cart page jump function, the user can also click on the navigation label pointing to the shopping cart page on the current application page to initiate the operation of jumping to the shopping cart page, and this operation is equivalent to initiating the operation of requesting the shopping cart page. Or, a shopping cart icon or floating window can also be displayed on each page, and the user clicking on the shopping cart icon or floating window can also initiate the operation of requesting the shopping cart page.

[0033] For another example, a recommendation function can be embedded on the group page of a shopping software. When the user requests the group page, the terminal device 101 can respond to the operation of the user initiating the request for the group page, send a page request to the server device 102, so as to request the server device 102 to perform personalized information recommendation for the user and return the information of the recommended target products; the terminal device 101 renders the group page, and while displaying the group page to the user, displays the information of the target products returned by the server device 102 on the group page, so as to guide the user to purchase the required products from the target products. Optionally, in the case where the application software provides a group page jump function, the user can also click on the navigation label pointing to the group page on the current application page, for example, the common "My" label in each application, to initiate the operation of jumping to the group page, and this operation is equivalent to initiating the operation of requesting the group page.

[0034] For another example, a recommendation function can be embedded on the user's personal details page. When the user requests the personal details page, the terminal device 101 can respond to the operation of the user initiating the request for the personal details page, send a page request to the server device 102, so as to request the server device 102 to perform personalized information recommendation for the user and return the information of the recommended target products; the terminal device 101 renders the personal details page, and while displaying the personal details page to the user, displays the information of the target products returned by the server device 102 on the personal details page, so as to guide the user to purchase the required products from the target products. Optionally, in the case where the application software provides a personal details page jump function, the user can also click on the navigation label pointing to the personal details page on the current application page, such as the common "My" label in each application, to initiate the operation of jumping to the personal details page, and this operation is equivalent to initiating the operation of requesting the personal details page.

[0035] Regardless of which of the above methods, when the user requests a page embedded with a recommendation function, the terminal device 101 can respond to the page request operation, send a page request to the server device 102, and carry the user identifier in the page request; for the server device 102, it can receive the page request sent by the terminal device 101, obtain the user identifier from the page request, and determine the user who initiated the page request according to the user identifier. For the convenience of distinction and description, in the following embodiments of the present application, the user who initiates the page request operation is referred to as the target user. After determining the target user, the server device 102 can perform personalized product recommendation for the target user.

[0036] In this embodiment, the server device 102 determines at least one type of commodity that can participate in online transactions. These commodities can be all the commodities provided by the merchant or some of the commodities provided by the merchant, and there is no limitation in this regard. These commodities come from the resource warehouse of the merchant. The resource warehouse can be one or multiple, and can be the warehouse of the store or the regional-level warehouse, and there is no limitation in this regard. Further, in order to provide personalized commodity recommendations for the target user, the personalized needs of the target user can be considered, and targeted commodity recommendations can be made to the target user. In addition, in this embodiment, considering the application requirements of the integration of offline and online transactions, in the process of making commodity recommendations to the target user, in addition to considering the personalized needs of the target user, the inventory information of the marketing end is also combined, and the marketing strategy at the front end of the merchant is combined with the inventory decision at the back end, and joint optimization recommendations for the front and back ends can be made from the perspective of global revenue management.

[0037] Specifically, as Figure 1 shown in ② and ③ in, after the server device 102 receives the page request, on the one hand, it obtains the shortage degree information and the current inventory information of at least one type of commodity; on the other hand, it predicts the preference information of the target user for at least one type of commodity. Among them, the current inventory information can be obtained in real time from the inventory management device 103. The shortage degree information of at least one type of commodity can be obtained in real time, or it can also be obtained and saved offline in advance. For example, the shortage degree information of at least one type of commodity can be obtained and saved periodically, so that the latest saved shortage degree information of at least one type of commodity can be directly obtained each time it is used. Correspondingly, the preference information of the target user for at least one type of commodity can be predicted in real time, or it can also be predicted and saved offline in advance, so that the latest saved preference information of the target user for at least one type of commodity can be directly obtained each time it is used. In 1, the case of predicting the preference information of the target user for at least one type of commodity in real time is taken as an example for illustration, but it is not limited to this. After that, as Figure 1 shown in ④ in, the server device 102 selects target commodities from at least one type of commodity according to the shortage degree information, the current inventory information of at least one type of commodity, and the preference information of the target user for at least one type of commodity; further, as Figure 1 shown in ⑤ in, the server device 102 returns the information of the selected target commodities to the terminal device 101 for the terminal device 101 to display the information of the target commodities on the page requested by the target user, as Figure 1 shown in ⑥ in.

[0038] When recommending products to users, if solely based on users' preferences for products, then in the case where most users prefer the same product, this product will be recommended to most users. In the situation where the inventory of this product is insufficient, ineffective product recommendations will occur, that is, some users are recommended this product but cannot purchase the recommended product due to insufficient inventory. This will seriously affect the users' shopping experience and also reduce the users' trust in the recommendation system, and the advantages of the recommendation system cannot be fully utilized. In the embodiments of the present application, by combining the preferences of the target user, the shortage degree of the product, and the inventory information at the same time, rather than solely relying on users' preferences for products, when the product is in short supply and the inventory is insufficient, other products can be recommended to users as a compromise, which is conducive to more reasonably recommending products to users, reducing the occurrence of ineffective product recommendations, and ensuring the users' shopping experience while improving the users' shopping efficiency.

[0039] In the above or below embodiments of the present application, when the service end device 102 predicts the preference information of the target user for at least one inventory resource in real time or in advance, it can obtain the portrait data of the target user according to the user identifier; the portrait data of the target user at least includes: the basic attribute information of the target user and the historical transaction behavior data of the target user; among them, the basic attribute information of the target user includes but is not limited to: the consumption ability, educational background, etc. of the target user; the historical transaction behavior data of the target user includes but is not limited to: the types of historical transaction behaviors such as purchase, adding to the shopping cart, payment, comment, etc., the product attributes involved in the historical transaction behavior, the time and occurrence frequency of the historical transaction behavior, etc. Further, the service end device 102 predicts the preference information of the target user for at least one product in real time or in advance according to the portrait data of the target user. Furthermore, the service end device 102 can simultaneously predict the preference information of the target user for at least one product according to the portrait data of the target user and the basic attribute information of at least one product. The basic attribute information of each product includes but is not limited to: category, brand, specification, price, brand tone, sales situation of the product, etc. Preferably, considering that the historical transaction behavior data in the target user portrait data will change dynamically over time, the latest portrait data of the target user can be obtained each time a product recommendation needs to be made to the target user, and according to the latest portrait data, or simultaneously according to the latest portrait data and the basic attribute information of at least one product, the preference information of the target user for at least one product is predicted in real time, which can improve the accuracy of the prediction results.

[0040] In an optional embodiment, a preference prediction model can be pre-trained, and then the server device 102 can call the preference prediction model to predict in real time or in advance the preference information of the target user for at least one commodity. Specifically, the server device 102 can input the portrait data of the target user and the basic attribute information of at least one commodity into the preference prediction model, and the preference prediction model outputs the preference information of the target user for at least one commodity.

[0041] Further optionally, the CVR and CTR of the target user for at least one commodity can be used to represent the preference information of the target user for at least one commodity, but it is not limited thereto. For example, the CVR or CTR of the target user for at least one commodity can also be used to represent the preference information of the target user for at least one commodity. Of course, other parameters other than CVR and CTR can also be used, such as the preference degree of the target user for the commodity, or the matching degree between the target user and the commodity, etc., to represent the preference information of the target user for each commodity. In the case of using the CVR and CTR of the target user for at least one commodity to represent the preference information of the target user for at least one commodity, the server device 102 can predict the CVR and CTR of the target user for at least one commodity according to the portrait data of the target user. Further optionally, in the case of using a preference prediction model, the preference prediction model can be a model capable of predicting CVR and CTR simultaneously, or can also include a CVR prediction model and a CTR prediction model.

[0042] After obtaining the CVR and CTR of the target user for at least one commodity, the server device 102 can select a target commodity from at least one commodity according to the shortage information of at least one inventory resource, the current inventory information, and the CVR and CTR of the target user for at least one commodity, and return the information of the target commodity to the terminal device 101 for the terminal device 101 to display the information of the target commodity on the page requested by the target user.

[0043] In the above or below embodiments of the present application, one way to select a target inventory resource from at least one commodity according to the shortage information of at least one commodity, the current inventory information, and the preference information of the target user for at least one commodity includes:

[0044] First, according to the shortage information of at least one commodity and the preference information of the target user for at least one commodity, determine the expected revenue of at least one commodity.

[0045] Optionally, in the case where the CVT and CTR of the goods by the target user are used to represent the preference information of the target user for the goods, the above method for determining the expected revenue of at least one good according to the shortage degree information of at least one good and the preference information of the target user for at least one good includes: correcting the price attribute of at least one good according to the shortage degree information of at least one good to obtain the corrected price of at least one good; determining the expected revenue of at least one good according to the CVR and CTR of the target user for at least one good and the corrected price of at least one good.

[0046] Next, according to the current inventory information of at least one good, the expected revenue of at least one good is corrected.

[0047] Optionally, the above correction of the expected revenue of at least one good according to the current inventory information of at least one good includes: generating an inventory penalty factor for at least one good according to the current inventory information and the initial inventory information of at least one good; using the inventory penalty factor of at least one good to correct the expected revenue of at least one good. For example, for each good, the inventory penalty factor of the good can be calculated according to the ratio of the current inventory information to the initial inventory information of the good; furthermore, the corrected expected revenue of the good can be obtained by multiplying the inventory penalty factor of the good by the expected revenue of the good.

[0048] Finally, according to the corrected expected revenue of at least one good, target goods are selected from at least one good.

[0049] Optionally, at least one good can be sorted according to the order of the corrected expected revenue of at least one good from large to small, and then, several goods ranked at the top are selected as target goods. Alternatively, goods with corrected expected revenue within a set revenue range can also be selected as target goods. The specific implementation manner of selecting target goods according to the corrected expected revenue of at least one good in the embodiments of the present application is not limited.

[0050] After obtaining the target goods, the server device 102 can return the information of the target goods to the terminal device 101 for the terminal device 101 to display the information of the target goods on the page requested by the target user, so that the user can purchase the required goods from the target goods, which is time-saving and efficient, and the merchant can also obtain the maximum revenue. Optionally, when the terminal device 101 displays the information of the target goods, it can display the information of the target goods on the page requested by the user in turn according to the corrected expected revenue of the target goods from high to low. The information of the target goods includes but is not limited to: pictures, names, prices, weights or quantities of the target goods, and relevant preferential information or preferential strategies, etc.

[0051] In the above or below embodiments of the present application, the shadow price of a commodity can be used to reflect the shortage information of the commodity. The shadow price of a commodity can reflect the true situation of the commodity shortage, and the shadow price of a shortage commodity is relatively high. That is to say, the higher the shadow price of a certain commodity, the greater the shortage degree of this commodity. Further, in the embodiments of the present application, the server device 102 can use the historical preference information of historical users for at least one commodity as the data basis, and use the linear programming method to estimate the shadow price of at least one commodity. Further optionally, when the number of historical users is large, the data basis of the historical preference information of historical users for at least one commodity will be very large. In view of this, the historical preference information of historical users for at least one commodity can be sampled to obtain the historical preference information of sampled historical users for at least one commodity; based on the historical preference information of sampled historical users for at least one commodity, estimate the preference of future arriving users for at least one commodity, so as to obtain the shadow price of at least one commodity, which can reduce the calculation amount and improve the calculation efficiency.

[0052] Further, considering that the shortage degree of at least one commodity will be affected by various factors and may be dynamically changing, therefore, when receiving a page request each time, the shadow price of at least one commodity can be estimated in real time based on the historical preference information of sampled historical users for at least one commodity, or the shadow price of at least one commodity can be estimated periodically based on the historical preference information of sampled historical users for at least one commodity. This can improve the accuracy of commodity recommendation based on the shadow price of the commodity and avoid the deficiency that only calculating the shadow price once cannot accurately depict the commodity recommendation effect. Of course, considering the calculation efficiency and calculation resources, preferably, the shadow price of at least one commodity can be estimated periodically. In view of this, a detailed implementation manner of the server device 102 using the linear programming method to estimate the shadow price of at least one commodity periodically is as Figure 2 shown, including the following operations:

[0053] Sampling operation: Sample the historical preference information of historical users for at least one commodity to obtain the historical preference information of sampled historical users for at least one commodity.

[0054] In the sampling operation, a part of the historical preference information of historical users for at least one commodity can be periodically extracted from the historical preference information of historical users for at least one commodity. Since the extracted part of the historical preference information of historical users for at least one commodity will be used to guide the subsequent recommendation process, it should be representative and can reflect as much as possible the preference information of future arriving users for at least one commodity. Future arriving users refer to users who will initiate a page request through the terminal device 101 in the future.

[0055] In an alternative embodiment, the historical CVR and historical CTR of at least one commodity by historical users can be used to represent the historical preference information of historical users for at least one commodity. Based on this, the CTR and CVR of some historical users can be periodically extracted from historical data, denoted as and The superscript h indicates historical data, the subscript i indicates the i-th commodity, and the subscript u indicates the sampled historical user. Then represents the CTR of the sampled historical user u for the i-th commodity, represents the CVR of the sampled historical user u for the i-th commodity.

[0056] Further optionally, for the current time window, the number of users likely to arrive in the current time window can be predicted according to the number of historical users who appeared in the historical same-period time window, denoted as k cur ; according to the number k cur of users likely to arrive in the current time window, sampling is performed from the historical users who appeared in the historical same-period time window to obtain sampled historical users. For example, if the current time window is from 7:00 to 8:00 in the morning, sampling can be performed from the historical users who appeared during the period from 7:00 to 8:00 every morning in the recent week, or sampling can also be performed from the historical users who appeared during the period from 7:00 to 8:00 every morning in the recent 10 days. Optionally, the number of sampled historical users is k cur , but not limited thereto. Further, from the historical preference information of historical users for at least one commodity, the historical preference information of sampled historical users for at least one commodity, such as CVR and CTR, is obtained.

[0057] Linear programming model construction operation: Based on the historical preference information of sampled historical users for at least one commodity and the price attributes of at least one commodity, a linear programming model is constructed with the recommendation probability of at least one commodity as the decision variable and the maximum expected benefit of sampled historical users for at least one commodity in the current time window as the objective.

[0058] In an alternative embodiment, if the historical CVR and historical CTR of users for commodities are used to represent the historical preference information of historical users for commodities, the objective function of the constructed linear programming model can be expressed by the following formula:

[0059]

[0060] Further, in the process of constructing the linear programming model, the average distribution of commodity inventory and the constraint of the maximum number of commodities that can be recommended each time can also be considered to construct the constraint conditions of the linear programming model. Specifically: combined with the number k curand the current inventory information of at least one commodity, determine the allocation quantity of at least one commodity within the current time window According to the allocation quantity of at least one commodity within the current time window and the maximum number K of inventory resource types that can be recommended each time, construct the constraint conditions of the linear programming model. These constraint conditions can be expressed as the following formulas (2) and (3):

[0061]

[0062]

[0063] The above formula (1) represents the maximization of the expected revenue of sampled historical users for N commodities within the current time window; where represents the expected revenue of sampled historical user u for the i-th commodity within the current time window, represents the CTR of sampled historical user u for the i-th commodity, represents the CVR of sampled historical user u for the i-th commodity, r i represents the price attribute of the i-th commodity, x iu represents the recommendation probability of the i-th commodity for sampled historical user u, Ts represents the total number of sampled historical users, optionally, Ts = k cur , N represents the total number of at least one commodity.

[0064] In the above formula (1), x iu is a decision variable and satisfies In the linear programming model of this embodiment, the decision variable x iu is not a binary decision variable of 0-1, but is relaxed to a continuous variable in the range of [0,1], but the solution of the corresponding integer programming model is still a 0-1 variable.

[0065] The above formula (2) is an embodiment of the requirement for commodity inventory balance. After appropriate adjustment of the initial inventory at the start of the current time window, it is allocated to each time window, and it is required that the total quantity of the i-th commodity sold to Ts users within the current time window should be less than or equal to the allocation quantity of this commodity within the current time window, represents the allocation quantity of the i-th commodity within the current time window.

[0066] The above formula (3) is an embodiment of the requirement that the number of commodities that can be recommended to users each time does not exceed the maximum number, where K represents the maximum number of commodities that can be recommended to users each time.

[0067] Linear scale model solving operation: Based on the duality theory, the above linear programming model is solved to obtain the shadow price of at least one commodity. The shadow price of each commodity is the dual value of the recommendation probability of the commodity, reflecting the scarcity of the commodity. Among them, the scarcity of the commodity also reflects the popularity of the commodity. The process of solving the above linear programming model based on the duality theory is not described in detail in the embodiment of the present application.

[0068] After obtaining the shadow price of each product in the above manner, within the current time window, when a target user initiates a page request for a page with an embedded recommendation function, the server device 102 predicts the CVR and CTR of the target user for at least one product based on the user ID carried in the page request and the portrait data corresponding to the user ID and the basic attribute information of at least one product, which is recorded as and The superscript cur represents the current time window, the subscript i represents the i-th product, and the subscript o represents the target user. represents the CVR of the target user for the i-th product, represents the CTR of the target user for the i-th product. On the other hand, the server device 102 can obtain the shadow price of at least one product, denoted as α i , represents the shadow price of the i-th commodity; according to the shadow price of at least one commodity, the price attribute of at least one commodity is modified to obtain the modified price of at least one commodity. The modified price of the i-th commodity can be expressed as r i -α i ; Further, according to the CVR and CTR of the target user for at least one commodity and the revised price of at least one commodity, the expected profit of at least one commodity is determined; the expected profit of the i-th commodity can be expressed as According to the current inventory information of at least one commodity and the initial inventory information of the current time window, the inventory penalty factor of at least one commodity is generated. The inventory penalty factor of the i-th commodity can be expressed as Where f(x) represents the inventory penalty factor function, represents the current inventory information of the ith product, and the superscript t represents the current time. represents the initial inventory information of the i-th commodity in the current time window; the expected return of at least one commodity is corrected according to the inventory penalty factor of at least one commodity to obtain the corrected expected return of at least one commodity. The corrected expected return of the i-th commodity can be expressed as At least one commodity is sorted in descending order based on the corrected expected return of the at least one commodity, and one or more commodities with the highest ranking are recommended as target commodities.

[0069] In the above embodiments, two factors, namely the shadow price and inventory balance, are organically combined, integrating the modeling ability of linear programming, the theoretical basis of duality theory, and the dynamics of the inventory balance algorithm. When constructing a linear programming model for calculating the shadow price, the limited commodity inventory is reasonably scaled. By periodically solving the linear programming problem based on historical data, the shadow price of the commodity is dynamically updated, and an inventory penalty factor based on real-time inventory is introduced to adjust the expected revenue index of each commodity in real time. This can not only appropriately reduce the conservatism caused by only considering inventory balance, but also overcome the defect that the shadow price remains static within a single period and cannot reflect user differences, which is beneficial to more accurately recommend commodities and ensure that the merchant's revenue is not damaged. In addition, due to the simplicity of the inventory balance algorithm, inventory information from the perspective of supply chain management can be incorporated almost without increasing the additional computational burden, which helps to consider the personalized recommendation problem from a global perspective. Furthermore, in the embodiments of the present application, the characteristics of the supply chain end and logistics operations are fully considered, enabling the combination of front-end commodity recommendation and commodity inventory management. From the perspective of the merchant, commodity recommendation is no longer simply based on the purchase probability or the expected profit of a single product, but rather incorporates the commodity inventory at the supply chain end. By explicitly introducing real-time inventory level information for personalized commodity recommendation, the merchant's GMV can be maximized.

[0070] In practical applications, most commodities have a validity period. Some commodities have a longer validity period, while some have a shorter validity period, and commodities need to be sold within the validity period. Especially for some fresh products, such as vegetables, fresh milk, meat, etc., these commodities have a shorter validity period and cannot be sold after spoiling. The validity period of a commodity has a certain impact on its shadow price. Based on this, in some alternative embodiments of the present application, during the process of calculating the shadow price of a commodity, the salvage value information of the commodity can also be introduced, and this salvage value information is determined by the validity period of the commodity. Among them, each commodity has a salvage value information, and the salvage value information of different commodities is different. The longer the validity period of a commodity, the greater its salvage value information. On the contrary, the shorter the validity period of a commodity, the smaller its salvage value information. Even for some commodities, such as fish, fresh milk, meat, etc., their salvage value information may even be negative.

[0071] Optionally, the salvage value information of each commodity can be determined in advance according to the validity period of the commodity by using the corresponding salvage value determination rule. For example, assume that the morning price of a commodity is 10 yuan, then its evening price drops to 6 yuan, and the salvage value information of this commodity is 6 yuan. Or, the morning price of a commodity is 15 yuan, and in the evening, this commodity needs to be processed into other foods, and the price of the processed food is 5 yuan, then the salvage value information of this commodity is 5 yuan. The way of determining the salvage value information of the commodity here is only an example and is not limited thereto.

[0072] Among them, the process of estimating the shadow price of a commodity when introducing the residual value information of the commodity is similar to Figure 2 the process of estimating the shadow price of a commodity without introducing the residual value information of the commodity shown in

[0073] The construction of the objective function: According to the historical preference information of at least one commodity of sampled historical users, the price attributes of at least one commodity, and the recommendation probabilities of at least one commodity, generate the basic expected revenue function of at least one commodity for historical users within the current time window; According to the residual value information of at least one commodity and the current inventory information of at least one commodity, generate the loss expected revenue function of at least one commodity within the current time window; Take maximizing the sum of the basic expected revenue function and the loss expected revenue function as the objective function of the linear programming model. This objective function can be expressed as the following formula (4):

[0074]

[0075] The above formula (1) represents maximizing the sum of the basic expected revenue function and the loss expected revenue function; among them, represents the basic expected revenue function, represents the loss expected revenue function; w i represents the residual value information of the i-th commodity, represents the remaining inventory information of the i-th commodity. For the description of other parameters, reference can be made to the foregoing embodiments, which will not be elaborated here.

[0076] The construction of the constraint conditions: Considering the constraints of the average distribution of commodity inventory, the remaining inventory of commodities, and the maximum number of commodities that can be recommended each time, construct the constraint conditions of the linear programming model. Specifically: Combining the possible number of users k cur arriving within the current time window and the current inventory information of at least one commodity, determine the distribution quantity of at least one commodity within the current time window According to the distribution quantity of at least one commodity within the current time window the remaining inventory information of at least one commodity and the maximum inventory resource quantity K that can be recommended each time, construct the constraint conditions of the linear programming model. These constraint conditions can be expressed as the following formulas (5) and (6):

[0077]

[0078]

[0079] The above formula (5) reflects the requirements for the balance of commodity inventory. After appropriately adjusting the initial inventory at the start of the current time window, it is allocated to subsequent time windows. And it is required that the sum of the total quantity of the i-th commodity sold to Ts users within the current time window and the remaining inventory information of the i-th commodity should be less than or equal to the allocation quantity of the commodity within the current time window. represents the allocation quantity of the i-th commodity within the current time window. Among them, it is necessary to satisfy Formula (6) is the same as formula (3), and will not be elaborated here.

[0080] Similarly, based on the duality theory, the above linear programming model can be solved to obtain the shadow price of at least one commodity. After obtaining the shadow price of at least one commodity, the process of commodity recommendation based on the shadow price of at least one commodity, the current inventory information, and the CVR and CTR of the target users for at least one commodity is the same as that of the foregoing embodiments, and will not be elaborated here.

[0081] In the above or following embodiments of the present application, the calculation function f(x) used to calculate the inventory penalty factor is not limited. f(x) can be any non-decreasing concave function and satisfies f(0)=0 and f(1)=1. In an alternative embodiment, This calculation function can ensure that the competitive ratio of the algorithm can reach at least In the above embodiments of the present application, in the case of In another alternative embodiment, f(x)=x can be adopted.

[0082] In the following alternative embodiments of the present application, in the process of calculating the inventory penalty factor, in addition to considering the current inventory information and the initial inventory information of the commodity, other information that affects the inventory information can also be considered, and multi-source information is integrated to calculate the inventory penalty factor.

[0083] In alternative embodiment S1, consider the initial inventory Current inventory information and predicted transaction information

[0084] Considering that at least one commodity may support multiple trading channels, a trading channel refers to a channel through which users can purchase commodities online, including but not limited to: mini programs and APPs developed by merchants themselves, as well as APPs and mini programs of third parties collaborating with merchants. The mini programs and APPs developed by merchants themselves, third-party mini programs, and third-party APPs belong to different trading channels. For the convenience of distinction and description, the trading channel used by the target user to initiate a page request is called the target trading channel, and the trading channels different from the target trading channel are called other trading channels.

[0085] In this case, when calculating the inventory penalty factor, the initial inventory of the current time window can be considered simultaneously Current inventory information and the predicted transaction information of the product currently on other trading channels, denoted as The superscript j represents the target trading channel, and its values can be 1, 2, 3, 4, etc. The -j represents other trading channels. The subscript i represents the i-th product, and the subscript t represents the current time represents the predicted transaction information of the i-th product currently on other trading channels. That is, the parameter x in the inventory penalty function f(x) is related to and In this embodiment, the relationship between the parameter x and and is not limited. For example, a relationship between the parameter x and and can be expressed as: Then the inventory penalty factor can be expressed as but not limited to this. Among them, y + = max{0, y},

[0086] Before using the predicted transaction information of at least one product currently on other trading channels, based on the historical transaction information of at least one product on other trading channels, the predicted transaction information of at least one product currently on other trading channels can be predicted. For any product, its historical transaction information on other trading channels includes but is not limited to: transaction time (such as weekdays, weekends, holidays), transaction volume, characteristics of the traded product, traffic information of the trading channel at that time, marketing information adopted by the trading channel at that time, etc.

[0087] In the alternative embodiment S2, the initial inventory Current inventory information and the predicted transaction information are inaccurate.

[0088] On the basis of the alternative embodiment S1, further consider the situation where the predicted transaction information of at least one product currently on other trading channels cannot be accurately predicted. In this case, a relatively robust processing method can be adopted. For example, the predicted transaction information with a small value but a high probability of occurrence can be estimated. Optionally, the predicted transaction information of at least one product currently on other trading channels can be predicted Calculate the standard deviation of the predicted transaction information of at least one product currently on other trading channels, and according to this standard deviation, the predicted transaction information of at least one product on other trading channels ​ Make corrections, and based on the corrected predicted transaction information Calculate the inventory penalty factor. Optionally, the predicted transaction information can be corrected in the following manner to obtain wherein represents the standard deviation.

[0089] Based on the above, in this embodiment, the parameter x in the inventory penalty function f(x) is related to and In this embodiment, the relationship between the parameter x and and is not limited. For example, the relationship between the parameter x and and can be expressed as: Then the inventory penalty factor can be expressed as but not limited thereto. Wherein

[0090] In the alternative embodiment S3, consider the initial inventory the current inventory information the predicted transaction information is inaccurate and the synchronization of inventory information between offline and online is lagging.

[0091] Based on the alternative embodiment S2, further consider the problem that there is a certain lag in the synchronization of offline inventory information to online. For example, when the inventory manager updates the inventory information at 11:00 and synchronizes the updated inventory information to the server device 102 at 11:20, a certain number of goods may be sold through the offline trading channel during the period from 11:00 to 11:20. If the inventory information at 11:00 is still used for processing at 11:20, there will be an error. In view of this, further consider the "protection level" during the period when the offline inventory information is synchronized to online in the process of calculating the inventory penalty factor, so as to prevent the situation of out-of-stock when the user purchases the recommended goods after the goods are recommended according to the lagged inventory information, and reduce the user's shopping experience. Taking the period from 11:00 to 11:20 as an example, this period is the period when the offline inventory information is synchronized to online. In this embodiment, the "protection level" of the goods during the period when the offline inventory information is synchronized to online is called the predicted transaction information of the goods during the period when the offline inventory information is synchronized to online. For the i-th good, the predicted transaction information of the good during the current period when the offline inventory information is synchronized to online is denoted as PL i .

[0092] Optionally, the predicted transaction information of each commodity during the current online synchronization period of offline inventory information can be predicted according to the historical transaction information of each commodity during the historical online synchronization period of offline inventory information. For example, for the i-th commodity, the sales volume of the commodity during the online synchronization period of offline inventory information in the recent one month can be counted, and the sales volume of the commodity during the current online synchronization period of offline inventory information can be predicted according to the sales volume during the online synchronization period of offline inventory information in the recent one month.

[0093] For each commodity, after obtaining the predicted transaction information of the commodity during the current online synchronization period of offline inventory information, when calculating the inventory penalty factor for the commodity, the quantity of the commodity that has been sold during the online synchronization period of offline inventory information (i.e., the predicted transaction information) can be removed. In this way, when recommending commodities to online users based on the inventory penalty factor, corresponding commodities will not be pushed when the inventory level is relatively low, avoiding the situation of out-of-stock when users purchase the recommended commodities.

[0094] Based on the above, in this embodiment, the parameter x in the inventory penalty function f(x) is related to and PL i In this embodiment, the relationship between the parameter x and and PL i is not limited. For example, a relationship between the parameter x and and PL i can be expressed as: Then the inventory penalty factor can be expressed as However, it is not limited thereto. Wherein, y + = max{0, y},

[0095] In the above optional embodiment S3, both the predicted transaction information and the predicted transaction information PL i are considered, but it is not limited thereto. In another optional embodiment of the present application, only the initial inventory the current inventory information and the predicted transaction information PL i can be considered. Regarding only considering the initial inventory the current inventory information and the predicted transaction information PL iThe method is similar to the above and will not be described in detail. That is, in the process of generating the inventory penalty factor of at least one commodity, at least one predicted transaction information corresponding to at least one commodity can be obtained, wherein the at least one predicted transaction information includes the predicted transaction information of at least one commodity currently on other transaction channels and / or the predicted transaction information of at least one commodity during the period of synchronization of the current offline inventory information to the online one; then, the inventory penalty factor of at least one commodity is generated according to the current inventory information of at least one commodity, the initial inventory information and the at least one predicted transaction information corresponding to at least one commodity.

[0096] After obtaining the inventory penalty factor of at least one commodity by using the above optional embodiments, the expected profit of at least one commodity can be corrected based on the inventory penalty factor of at least one commodity; and then the target commodity can be recommended to the target user based on the corrected expected profit of at least one commodity. These operations are the same as those in the above embodiments and will not be repeated here.

[0097] In the above optional embodiments, by introducing additional predicted transaction information and taking into account the presence of prediction deviations, the inventory penalty factor can be calculated more accurately, which is conducive to improving the accuracy of product recommendations based on the inventory penalty factor, more reasonably recommending products to users, and reducing the probability of invalid product recommendations.

[0098] Furthermore, in the embodiment of the present application, an offline simulation is performed on the embodiment of the present application, and it is obtained that the embodiment of the present application has the following beneficial results in terms of indicators such as merchant revenue and inventory:

[0099] 1. The solution of the embodiment of the present application can take into account both user preferences and the stability of user arrival, and as the stability of user preferences and the intensity of user arrival becomes worse, the improvement of merchant GMV becomes more obvious; in simulation experiments, compared with the existing recommendation method, the improvement of GMV after adopting the solution of the embodiment of the present application can reach up to 1% or even 2%.

[0100] 2. When the initial inventory is equivalent to the number of goods that the user expects to purchase on the day, the improvement effect of the embodiment of the present application on the GMV indicator is particularly obvious.

[0101] Figure 3 The following is a flow chart of an information recommendation method provided by an exemplary embodiment of the present application. Figure 3 As shown, the method includes:

[0102] 31. Receive a page request sent by a terminal device, where the page request includes a user identifier, and the user identifier is used to identify a target user who initiates the page request operation;

[0103] 32. Obtain the shortage degree information and current inventory information of at least one inventory resource that can be traded online, and predict the preference information of the target user for at least one inventory resource;

[0104] 33. Select a target inventory resource from at least one inventory resource according to the shortage degree information, current inventory information of at least one inventory resource, and the preference information of the target user for at least one inventory resource;

[0105] 34. Send the information of the target inventory resource to the terminal device for the terminal device to display the information of the target inventory resource on the page requested by the target user.

[0106] In an optional embodiment, obtaining the shortage degree information of at least one inventory resource that can be operated online includes: estimating the shadow price of at least one inventory resource based on the historical preference information of historical users for at least one inventory resource that can be traded online; wherein, the shadow price of each inventory resource reflects the shortage degree of the inventory resource.

[0107] Further optionally, estimating the shadow price of at least one inventory resource based on the historical preference information of historical users for at least one inventory resource that can be traded online includes: sampling the historical preference information of historical users for at least one inventory resource to obtain the sampled historical preference information of sampled historical users for at least one inventory resource; estimating the preference of future arriving users for at least one inventory resource based on the sampled historical preference information of sampled historical users for at least one inventory resource to obtain the shadow price of at least one inventory resource.

[0108] Further optionally, sampling the historical preference information of historical users for at least one inventory resource to obtain the sampled historical preference information of sampled historical users for at least one inventory resource includes: for the current time window, predicting the number of users who may arrive in the current time window according to the number of historical users who appeared in the historical same - period time window; sampling from the historical users who appeared in the historical same - period time window according to the number of users who may arrive in the current time window to obtain sampled historical users;

[0109] Obtain the sampled historical preference information of sampled historical users for at least one inventory resource from the historical preference information of historical users for at least one inventory resource.

[0110] Further optionally, estimating the preference of future arriving users for at least one inventory resource based on the sampled historical preference information of sampled historical users for at least one inventory resource to obtain the shadow price of at least one inventory resource includes:

[0111] Based on the historical preference information of at least one inventory resource of sampled historical users and the price attributes of at least one inventory resource, construct a linear programming model with the recommendation probability of at least one inventory resource as the decision variable and the goal of maximizing the expected revenue of sampled historical users for at least one inventory resource within the current time window;

[0112] Based on the duality theory, solve the linear programming model to obtain the shadow price of at least one inventory resource. The shadow price of each inventory resource is the dual value of the recommendation probability of this inventory resource, reflecting the shortage degree of this inventory resource.

[0113] Further optionally, during the process of constructing the linear programming model, it also includes:

[0114] Combined with the number of users that may arrive within the current time window and the current inventory information of at least one inventory resource, determine the allocation quantity of at least one inventory resource within the current time window;

[0115] According to the allocation quantity of at least one inventory resource within the current time window and the maximum number of inventory resources that can be recommended each time, construct the constraint conditions of the linear programming model.

[0116] Further optionally, based on the historical preference information of at least one inventory resource of sampled historical users and the price attributes of at least one inventory resource, construct a linear programming model with the recommendation probability of at least one inventory resource as the decision variable and the goal of maximizing the expected revenue of sampled historical users for at least one inventory resource within the current time window, including:

[0117] According to the historical preference information of at least one inventory resource of sampled historical users, the price attributes of at least one inventory resource, and the recommendation probability of at least one inventory resource, generate the basic expected revenue function of historical users for at least one inventory resource within the current time window;

[0118] According to the salvage value information of at least one inventory resource and the current inventory information of at least one inventory resource, generate the loss expected revenue function of at least one inventory resource within the current time window. The salvage value information of the inventory resource is determined according to the validity period of the inventory resource;

[0119] Take maximizing the sum of the basic expected revenue function and the loss expected revenue function as the objective function of the linear programming model.

[0120] In an alternative embodiment, according to the shortage degree information of at least one inventory resource, the current inventory information, and the preference information of the target user for at least one inventory resource, select the target inventory resource from at least one inventory resource, including:

[0121] Determine the expected revenue of at least one inventory resource according to the shortage degree information of at least one inventory resource and the preference information of the target user for at least one inventory resource;

[0122] Revise the expected revenue of at least one inventory resource according to the current inventory information of at least one inventory resource;

[0123] Select the target inventory resource from at least one inventory resource according to the revised expected revenue of at least one inventory resource.

[0124] Further optionally, predict the preference information of the target user for at least one inventory resource, including: predicting the click-through rate and click conversion rate of the target user for at least one inventory resource based on the portrait data of the target user. Correspondingly, determine the expected revenue of at least one inventory resource according to the shortage degree information of at least one inventory resource and the preference information of the target user for at least one inventory resource, including: revise the price attribute of at least one inventory resource according to the shortage degree information of at least one inventory resource to obtain the revised price of at least one inventory resource; determine the expected revenue of at least one inventory resource according to the click-through rate and click conversion rate of the target user for at least one inventory resource and the revised price of at least one inventory resource.

[0125] Further optionally, revise the expected revenue of at least one inventory resource according to the current inventory information of at least one inventory resource, including:

[0126] Generate an inventory penalty factor for at least one inventory resource according to the current inventory information and initial inventory information of at least one inventory resource;

[0127] Use the inventory penalty factor of at least one inventory resource to revise the expected revenue of at least one inventory resource.

[0128] Further optionally, generate an inventory penalty factor for at least one inventory resource according to the current inventory information and initial inventory information of at least one inventory resource, including:

[0129] Obtain at least one piece of predicted transaction information corresponding to at least one inventory resource, where at least one piece of predicted transaction information includes the predicted transaction information of at least one inventory resource on other transaction channels currently and / or the predicted transaction information of at least one inventory resource during the synchronization period of the current offline inventory information to the online;

[0130] Generate an inventory penalty factor for at least one inventory resource according to the current inventory information, initial inventory information of at least one inventory resource, and at least one piece of predicted transaction information corresponding to at least one inventory resource;

[0131] Among them, other trading channels refer to the trading channels other than the target trading channel among the multiple trading channels supported by at least one inventory resource, and the target trading channel refers to the trading channel used by the terminal device to initiate a page request.

[0132] Further optionally, obtaining the predicted trading information of at least one inventory resource currently on other trading channels, including: predicting the predicted trading information of at least one inventory resource currently on other trading channels based on the historical trading information of at least one inventory resource on other trading channels;

[0133] Correspondingly, obtaining the predicted trading information of at least one inventory resource during the period when the current offline inventory information is synchronized to the online, including: predicting the predicted trading information of at least one inventory resource during the period when the current offline inventory information is synchronized to the online according to the historical trading information of at least one inventory resource during the period when the historical offline inventory information is synchronized to the online.

[0134] Further optionally, before using the predicted trading information of at least one inventory resource currently on other trading channels, it further includes: correcting the predicted trading information of at least one inventory resource currently on other trading channels according to the standard deviation of the predicted trading information of at least one inventory resource currently on other trading channels.

[0135] In an alternative embodiment, the above at least one inventory resource is a commodity. Correspondingly, the page requested by the target user is the home page, shopping cart page, group page or user details page of the shopping application.

[0136] In the method embodiment of the present application, the description is carried out with the inventory resource as the description object, but the process is the same or detailed as that with the commodity as the object. Therefore, for the detailed implementation and description of the above steps, reference can be made to the foregoing system embodiment, which will not be elaborated herein.

[0137] In this embodiment, for the inventory resources that can be traded online, the front-end recommendation is combined with the inventory information at the supply chain end, and at the same time, multi-dimensional information such as the shortage degree information of the inventory resources, the current inventory information, and the preference information of the target user for the inventory resources is integrated to implement an information recommendation method based on inventory balance. This method can integrate the inventory information from the perspective of supply chain management with almost no increase in computational burden, which is beneficial to personalized recommendation for users from a global perspective, is beneficial to more accurately recommend inventory resources to users, reduces the recommendation probability of invalid resources, and improves the user experience.

[0138] Figure 4 It is a schematic flow chart of an information acquisition method provided for an exemplary embodiment of the present application. As Figure 4 shown, the method includes:

[0139] 41. In response to a page request operation, send a page request to the server device. The page request includes a user identifier, which is used to identify the target user who initiated the page request operation.

[0140] 42. Receive the information of the target inventory resource returned by the server device, and display the information of the target inventory resource on the page requested by the target user. Among them, the target inventory resource is selected by the server device from at least one inventory resource that can be traded online according to the shortage information, current inventory information, and the preference information of the target user for at least one inventory resource.

[0141] For the detailed implementation process of the server device selecting the target inventory resource according to the shortage information, current inventory information, and the preference information of the target user for at least one inventory resource and returning the information of the target inventory resource to the terminal device, reference can be made to the foregoing embodiments, and details are not described herein again.

[0142] It should be noted that the execution subject of each step of the method provided in the foregoing embodiments can be the same device, or the method can also be executed by different devices as the execution subject. For example, the execution subject of steps 41 to 42 can be device A; for another example, the execution subject of step 41 can be device A, and the execution subject of step 42 can be device B; and so on.

[0143] In addition, in some processes described in the foregoing embodiments and the accompanying drawings, a plurality of operations appear in a specific order. However, it should be clearly understood that these operations can be executed not in the order in which they appear in this article or in parallel. The operation numbers such as 41 and 42 are only used to distinguish different operations, and the numbers themselves do not represent any execution order. In addition, these processes can include more or fewer operations, and these operations can be executed in sequence or in parallel. It should be noted that the descriptions such as "first" and "second" in this article are used to distinguish different messages, devices, modules, etc., and do not represent a sequence, nor do they limit that "first" and "second" are of different types.

[0144] Figure 5 This is a schematic structural diagram of a server device provided by an exemplary embodiment of the present application. As Figure 5 shown, the server device includes: a memory 51, a processor 52, and a communication component 53.

[0145] The memory 51 is used to store computer programs and can be configured to store various other data to support operations on the server device. Examples of these data include instructions for any application program or method for operating on the server device, messages, pictures, videos, etc.

[0146] The memory 51 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disc.

[0147] The processor 52, coupled to the memory 51, is configured to execute a computer program in the memory 51 for:

[0148] Receiving a page request sent by a terminal device through the communication component 53, the page request including a user identifier for identifying a target user who initiates the page request operation; obtaining information on the shortage degree and current inventory information of at least one inventory resource available for online transactions, and predicting preference information of the target user for at least one inventory resource; selecting a target inventory resource from at least one inventory resource according to the shortage degree information, current inventory information of at least one inventory resource, and the preference information of the target user for at least one inventory resource; and sending information on the target inventory resource to the terminal device through the communication component 53 for the terminal device to display information on the target inventory resource on the page requested by the target user.

[0149] In an alternative embodiment, when obtaining information on the shortage degree of at least one inventory resource available for online operations, the processor 52 is specifically configured to: estimate the shadow price of at least one inventory resource based on historical preference information of historical users for at least one inventory resource available for online transactions; wherein the shadow price of each inventory resource reflects the shortage degree of the inventory resource.

[0150] Further optionally, when estimating the shadow price of at least one inventory resource based on historical preference information of historical users for at least one inventory resource available for online transactions, the processor 52 is specifically configured to: sample the historical preference information of historical users for at least one inventory resource to obtain sampled historical preference information of historical users for at least one inventory resource; and estimate the preference of future arriving users for at least one inventory resource based on the sampled historical preference information of historical users for at least one inventory resource to obtain the shadow price of at least one inventory resource.

[0151] Further, when sampling the historical preference information of historical users for at least one inventory resource, the processor 52 is specifically configured to:

[0152] For the current time window, predict the number of users likely to arrive in the current time window according to the number of historical users that appeared in the historical time window of the same period.

[0153] Sample historical users from the historical users who appeared in the historical same - period time window according to the number of users who may arrive within the current time window to obtain sampled historical users;

[0154] Obtain the historical preference information of the sampled historical users for at least one inventory resource from the historical preference information of historical users for at least one inventory resource.

[0155] Furthermore, when the processor 52 estimates the preference of future - arriving users for at least one inventory resource based on the historical preference information of the sampled historical users for at least one inventory resource to obtain the shadow price of at least one inventory resource, it is specifically used for:

[0156] Construct a linear - programming model with the recommended probability of at least one inventory resource as the decision variable and the maximum expected benefit of the sampled historical users for at least one inventory resource within the current time window as the objective, based on the historical preference information of the sampled historical users for at least one inventory resource and the price attributes of at least one inventory resource;

[0157] Solve the linear - programming model based on the duality theory to obtain the shadow price of at least one inventory resource. The shadow price of each inventory resource is the dual value of the recommended probability of this inventory resource, reflecting the shortage degree of this inventory resource.

[0158] Further optionally, during the process of constructing the linear - programming model, the processor 52 is also used for: combining the number of users who may arrive within the current time window and the current inventory information of at least one inventory resource to determine the allocation quantity of at least one inventory resource within the current time window; constructing the constraint conditions of the linear - programming model according to the allocation quantity of at least one inventory resource within the current time window and the maximum number of inventory resources that can be recommended each time.

[0159] Further optionally, when the processor 52 constructs a linear - programming model with the recommended probability of at least one inventory resource as the decision variable and the maximum expected benefit of the sampled historical users for at least one inventory resource within the current time window as the objective, it is specifically used for:

[0160] Generate the basic expected - benefit function of historical users for at least one inventory resource within the current time window according to the historical preference information of the sampled historical users for at least one inventory resource, the price attributes of at least one inventory resource, and the recommended probability of at least one inventory resource;

[0161] Generate the loss - expected - benefit function of at least one inventory resource within the current time window according to the residual - value information of at least one inventory resource and the current inventory information of at least one inventory resource. The residual - value information of the inventory resource is determined according to the validity period of the inventory resource;

[0162] Maximize the sum of the basic expected revenue function and the loss expected revenue function as the objective function of the linear programming model.

[0163] In an alternative embodiment, when the processor 52 selects a target inventory resource from at least one inventory resource according to the shortage degree information of at least one inventory resource, the current inventory information, and the preference information of the target user for at least one inventory resource, it is specifically configured to:

[0164] Determine the expected revenue of at least one inventory resource according to the shortage degree information of at least one inventory resource and the preference information of the target user for at least one inventory resource;

[0165] Revise the expected revenue of at least one inventory resource according to the current inventory information of at least one inventory resource;

[0166] Select a target inventory resource from at least one inventory resource according to the revised expected revenue of at least one inventory resource.

[0167] Further optionally, when the processor 52 predicts the preference information of the target user for at least one inventory resource, it is specifically configured to: Based on the portrait data of the target user, predict the click-through rate and click conversion rate of the target user for at least one inventory resource;

[0168] Correspondingly, when the processor 52 determines the expected revenue of at least one inventory resource according to the shortage degree information of at least one inventory resource and the preference information of the target user for at least one inventory resource, it is specifically configured to: Revise the price attribute of at least one inventory resource according to the shortage degree information of at least one inventory resource to obtain the revised price of at least one inventory resource; Determine the expected revenue of at least one inventory resource according to the click-through rate and click conversion rate of the target user for at least one inventory resource, and the revised price of at least one inventory resource.

[0169] Further optionally, when the processor 52 revises the expected revenue of at least one inventory resource according to the current inventory information of at least one inventory resource, it is specifically configured to: Generate an inventory penalty factor for at least one inventory resource according to the current inventory information and the initial inventory information of at least one inventory resource; Use the inventory penalty factor of at least one inventory resource to revise the expected revenue of at least one inventory resource.

[0170] Further optionally, when the processor 52 generates an inventory penalty factor for at least one inventory resource according to the current inventory information and the initial inventory information of at least one inventory resource, it is specifically configured to:

[0171] Obtain at least one piece of predicted transaction information corresponding to at least one inventory resource, where the at least one piece of predicted transaction information includes at least one piece of predicted transaction information of the at least one inventory resource on other transaction channels currently and / or at least one piece of predicted transaction information of the at least one inventory resource during the synchronization of the current offline inventory information to the online inventory information;

[0172] Generate an inventory penalty factor for at least one inventory resource according to the current inventory information, initial inventory information of the at least one inventory resource, and at least one piece of predicted transaction information corresponding to the at least one inventory resource;

[0173] Among them, other transaction channels refer to the transaction channels other than the target transaction channel among the multiple transaction channels supported by the at least one inventory resource, and the target transaction channel refers to the transaction channel used by the terminal device to initiate a page request.

[0174] Further optionally, when the processor 52 obtains the predicted transaction information of the at least one inventory resource on other transaction channels currently, it is specifically used for: predicting the predicted transaction information of the at least one inventory resource on other transaction channels currently based on the historical transaction information of the at least one inventory resource on other transaction channels;

[0175] Correspondingly, when the processor 52 obtains the predicted transaction information of the at least one inventory resource during the synchronization of the current offline inventory information to the online inventory information, it is specifically used for: predicting the predicted transaction information of the at least one inventory resource during the synchronization of the current offline inventory information to the online inventory information according to the historical transaction information of the at least one inventory resource during the synchronization of the historical offline inventory information to the online inventory information.

[0176] Further optionally, before using the predicted transaction information of the at least one inventory resource on other transaction channels currently, the processor 52 is also used for: correcting the predicted transaction information of the at least one inventory resource on other transaction channels currently according to the standard deviation of the predicted transaction information of the at least one inventory resource on other transaction channels currently.

[0177] In an optional embodiment, the above page is the home page, shopping cart page, group page or user details page of a shopping application; the above at least one inventory resource is a commodity.

[0178] Further, as Figure 5 shown, the server device further includes: a power supply component 54 and other components. Figure 5 Only some components are schematically shown, which does not mean that the server device only includes Figure 5 the components shown.

[0179] Accordingly, an embodiment of the present application further provides a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to implement the steps executable by the server device in the above method embodiments.

[0180] Accordingly, an embodiment of the present application further provides a computer program product, including a computer program / instructions, which, when executed by a processor, enables the processor to implement the steps executable by the server device in the above method embodiments.

[0181] Figure 6 FIG. is a schematic structural diagram of a terminal device provided by an exemplary embodiment of the present application. As Figure 6 shown, the terminal device includes: a memory 61, a processor 62, and a communication component 63.

[0182] The memory 61 is used to store computer programs and can be configured to store various other data to support operations on the terminal device. Examples of these data include instructions for any application program or method for operating on the terminal device, messages, pictures, videos, etc.

[0183] The memory 61 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, a magnetic disk, or an optical disk.

[0184] The processor 62 is coupled to the memory 61 and is used to execute the computer program in the memory 61 for:

[0185] Responding to a page request operation, sending a page request to the server device, where the page request includes a user identifier for identifying the target user who initiated the page request operation;

[0186] Receiving information about the target inventory resource returned by the server device and displaying the information about the target inventory resource on the page requested by the target user;

[0187] Wherein, the target inventory resource is selected by the server device from at least one inventory resource available for online trading according to the shortage degree information, the current inventory information, and the preference information of the target user for the at least one inventory resource.

[0188] Further, as Figure 6 shown, the terminal device further includes other components such as a display 64, an audio component 65, and a power supply component 66. Figure 6Only some components are schematically shown, which does not mean that the terminal device only includes Figure 6 the components shown.

[0189] Correspondingly, an embodiment of the present application further provides a computer-readable storage medium storing a computer program, which when executed by a processor, causes the processor to be able to implement the steps executable by the terminal device in the above method embodiments.

[0190] Correspondingly, an embodiment of the present application further provides a computer program product, including a computer program / instructions, which when executed by a processor, causes the processor to be able to implement the steps executable by the terminal device in the above method embodiments.

[0191] The above Figure 5 and Figure 6 The communication components therein are configured to facilitate communication between the device where the communication components are located and other devices in a wired or wireless manner. The device where the communication components are located can access a wireless network based on a communication standard, such as WiFi, 2G, 3G, 4G / LTE, 5G and other mobile communication networks, or a combination thereof. In an exemplary embodiment, the communication component receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component further includes a Near Field Communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on Radio Frequency Identification (RFID) technology, Infrared Data Association (IrDA) technology, Ultra Wideband (UWB) technology, Bluetooth (BT) technology and other technologies.

[0192] The above Figure 6 The display in the above includes a screen, and the screen can include a Liquid Crystal Display (LCD) and a Touch Panel (TP). If the screen includes a touch panel, the screen can be implemented as a touch screen to receive input signals from a user. The touch panel includes one or more touch sensors to sense touches, swipes and gestures on the touch panel. The touch sensors can sense not only the boundaries of touch or swipe actions, but also detect the duration and pressure associated with the touch or swipe operation.

[0193] The above Figure 5 and Figure 6 The power supply component in the above provides power for various components of the device where the power supply component is located. The power supply component can include a power management system, one or more power supplies, and other components associated with generating, managing and distributing power for the device where the power supply component is located.

[0194] The above Figure 6The audio component therein can be configured to output and / or input audio signals. For example, the audio component includes a microphone (MIC), which is configured to receive external audio signals when the device where the audio component is located is in an operation mode, such as a call mode, a recording mode, and a voice recognition mode. The received audio signals can be further stored in a memory or sent via a communication component. In some embodiments, the audio component further includes a speaker for outputting audio signals.

[0195] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0196] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0197] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means implements the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0198] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0199] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0200] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0201] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0202] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0203] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. An information recommendation method, characterized in that, Including: Receiving a page request sent by a terminal device, where the page request includes a user identifier for identifying a target user who initiates the page request operation; Based on the historical preference information of historical users for at least one inventory resource that can be traded online, using the linear programming method to estimate the preferences of future arriving users for the at least one inventory resource, so as to obtain the shadow price of the at least one inventory resource; wherein, the shadow price of each inventory resource is the dual value of the recommended probability of this inventory resource, reflecting the shortage degree of this inventory resource; Obtaining the current inventory information of the at least one inventory resource and predicting the preference information of the target user for the at least one inventory resource; Selecting target inventory resources from the at least one inventory resource according to the shortage degree information, current inventory information of the at least one inventory resource, and the preference information of the target user for the at least one inventory resource; Sending the information of the target inventory resources to the terminal device for the terminal device to display the information of the target inventory resources on the page requested by the target user.

2. The method according to claim 1, wherein Based on the historical preference information of historical users for at least one inventory resource that can be traded online, using the linear programming method to estimate the preferences of future arriving users for the at least one inventory resource, so as to obtain the shadow price of the at least one inventory resource, including: Sampling the historical preference information of historical users for the at least one inventory resource to obtain the historical preference information of sampled historical users for the at least one inventory resource; Based on the historical preference information of the sampled historical users for the at least one inventory resource, using the linear programming method to estimate the preferences of future arriving users for the at least one inventory resource, so as to obtain the shadow price of the at least one inventory resource.

3. The method according to claim 2, characterized in that, Sampling the historical preference information of historical users for the at least one inventory resource to obtain the historical preference information of sampled historical users for the at least one inventory resource, including: For the current time window, predicting the number of users who may arrive in the current time window according to the number of historical users who appeared in the historical same-period time window; Sampling from the historical users who appeared in the historical same-period time window according to the number of users who may arrive in the current time window to obtain sampled historical users; Obtaining the historical preference information of the sampled historical users for the at least one inventory resource from the historical preference information of historical users for the at least one inventory resource.

4. The method according to claim 3, wherein Based on the historical preference information of the sampled historical users for the at least one inventory resource, using the linear programming method to estimate the preferences of future arriving users for the at least one inventory resource, so as to obtain the shadow price of the at least one inventory resource, including: Based on the historical preference information of the sampled historical users for the at least one inventory resource and the price attributes of the at least one inventory resource, constructing a linear programming model with the recommended probability of the at least one inventory resource as the decision variable and the maximum expected revenue of the sampled historical users for the at least one inventory resource in the current time window as the goal; Based on the duality theory, solve the linear programming model to obtain the shadow price of the at least one inventory resource.

5. The method according to claim 4, characterized in that, In the process of constructing the linear programming model, it also includes: Combined with the number of users that may arrive within the current time window and the current inventory information of the at least one inventory resource, determine the allocation quantity of the at least one inventory resource within the current time window; According to the allocation quantity of the at least one inventory resource within the current time window and the maximum quantity of inventory resources that can be recommended each time, construct the constraint conditions of the linear programming model.

6. The method according to claim 5, wherein Based on the historical preference information of the at least one inventory resource of the sampled historical users and the price attributes of the at least one inventory resource, construct a linear programming model with the recommendation probability of the at least one inventory resource as the decision variable and the maximum expected benefit of the sampled historical users for the at least one inventory resource within the current time window as the objective, including: According to the historical preference information of the at least one inventory resource of the sampled historical users, the price attributes of the at least one inventory resource, and the recommendation probability of the at least one inventory resource, generate the basic expected benefit function of the historical users for the at least one inventory resource within the current time window; According to the salvage value information of the at least one inventory resource and the current inventory information of the at least one inventory resource, generate the loss expected benefit function of the at least one inventory resource within the current time window, and the salvage value information of the inventory resource is determined according to the validity period of the inventory resource; Maximize the sum of the basic expected benefit function and the loss expected benefit function as the objective function of the linear programming model.

7. The method according to any one of claims 1-6, characterized in that According to the shortage degree information of the at least one inventory resource, the current inventory information, and the preference information of the target user for the at least one inventory resource, select the target inventory resource from the at least one inventory resource, including: According to the shortage degree information of the at least one inventory resource and the preference information of the target user for the at least one inventory resource, determine the expected benefit of the at least one inventory resource; According to the current inventory information of the at least one inventory resource, correct the expected benefit of the at least one inventory resource; According to the corrected expected benefit of the at least one inventory resource, select the target inventory resource from the at least one inventory resource.

8. The method according to claim 7, wherein Predict the preference information of the target user for the at least one inventory resource, including: Based on the portrait data of the target user, predict the click-through rate and click conversion rate of the target user for the at least one inventory resource; Correspondingly, according to the shortage degree information of the at least one inventory resource and the preference information of the target user for the at least one inventory resource, determine the expected benefit of the at least one inventory resource, including: Correct the price attributes of the at least one inventory resource according to the shortage degree information of the at least one inventory resource to obtain the corrected price of the at least one inventory resource; Determine the expected revenue of the at least one inventory resource according to the click-through rate and click conversion rate of the at least one inventory resource by the target user, and the revised price of the at least one inventory resource.

9. The method according to claim 7, wherein Revise the expected revenue of the at least one inventory resource according to the current inventory information of the at least one inventory resource, including: Generate an inventory penalty factor for the at least one inventory resource according to the current inventory information and initial inventory information of the at least one inventory resource; Use the inventory penalty factor of the at least one inventory resource to revise the expected revenue of the at least one inventory resource.

10. The method according to claim 9, wherein Generate an inventory penalty factor for the at least one inventory resource according to the current inventory information and initial inventory information of the at least one inventory resource, including: Obtain at least one piece of predicted transaction information corresponding to the at least one inventory resource, where the at least one piece of predicted transaction information includes the predicted transaction information of the at least one inventory resource currently on other transaction channels and / or the predicted transaction information of the at least one inventory resource during the synchronization of the current offline inventory information to the online; Generate an inventory penalty factor for the at least one inventory resource according to the current inventory information, initial inventory information of the at least one inventory resource, and at least one piece of predicted transaction information corresponding to the at least one inventory resource; Among them, other transaction channels refer to the transaction channels other than the target transaction channel among the multiple transaction channels supported by the at least one inventory resource, and the target transaction channel refers to the transaction channel used by the terminal device to initiate a page request.

11. The method according to claim 10, characterized in that, Obtain the predicted transaction information of the at least one inventory resource currently on other transaction channels, including: predicting the predicted transaction information of the at least one inventory resource currently on the other transaction channels based on the historical transaction information of the at least one inventory resource on the other transaction channels; Obtain the predicted transaction information of the at least one inventory resource during the synchronization of the current offline inventory information to the online, including: predicting the predicted transaction information of the at least one inventory resource during the synchronization of the current offline inventory information to the online according to the historical transaction information of the at least one inventory resource during the synchronization of the historical offline inventory information to the online.

12. The method according to claim 11, wherein Before using the predicted transaction information of the at least one inventory resource currently on the other transaction channels, it further includes: Revise the predicted transaction information of the at least one inventory resource currently on the other transaction channels according to the standard deviation of the predicted transaction information of the at least one inventory resource currently on the other transaction channels.

13. The method according to any one of claims 1-6, characterized in that, The page is the home page, shopping cart page, group page or user details page of the shopping application; the at least one inventory resource is a commodity.

14. An information acquisition method, characterized in that, Include: Respond to the page request operation, send a page request to the server device, where the page request includes a user identifier, and the user identifier is used to identify the target user who initiates the page request operation; Receive the information of the target inventory resource returned by the server device, and display the information of the target inventory resource on the page requested by the target user; The target inventory resource is selected by the server device according to scarcity information of at least one inventory resource that can be traded online, current inventory information, and preference information of the target user for the at least one inventory resource; The scarcity information of the at least one inventory resource is reflected by the shadow price of each inventory resource; the shadow price of each inventory resource is the dual value of the recommendation probability of the inventory resource; the shadow price of the at least one inventory resource is obtained by estimating the preference of future arriving users for the at least one inventory resource based on historical preference information of historical users for the at least one inventory resource using a linear programming method.

15. A server device, characterized in that, include: Memory and processor; The memory is used to store computer programs or instructions; The processor, coupled to the memory, is configured to execute the computer program or instructions for: Receiving a page request sent by a terminal device, wherein the page request includes a user identifier, and the user identifier is used to identify a target user who initiates the page request operation; Based on historical user preference information for at least one inventory resource that can be traded online, a linear programming method is used to estimate the preference of future arriving users for the at least one inventory resource, so as to obtain a shadow price of the at least one inventory resource; wherein the shadow price of each inventory resource is the dual value of the recommendation probability of the inventory resource, reflecting the scarcity of the inventory resource; Acquire current inventory information of the at least one inventory resource, and predict preference information of the target user for the at least one inventory resource; selecting a target inventory resource from the at least one inventory resource according to scarcity information of the at least one inventory resource, current inventory information, and preference information of the target user for the at least one inventory resource; The information of the target inventory resource is sent to the terminal device, so that the terminal device displays the information of the target inventory resource on the page requested by the target user.

16. A terminal device, characterized in that, include: Memory, processor and display; The memory is used to store computer programs or instructions; The processor, coupled to the memory, is configured to execute the computer program or instructions for: In response to a page request operation, a page request is sent to a server device, wherein the page request includes a user identifier, and the user identifier is used to identify a target user who initiates the page request operation; receiving the target inventory resource information returned by the server device, and displaying the target inventory resource information on the page requested by the target user; The display is used to display the page requested by the target user; The target inventory resource is selected by the server device according to scarcity information of at least one inventory resource that can be traded online, current inventory information, and preference information of the target user for the at least one inventory resource; The shortage degree information of the at least one inventory resource is reflected by the shadow price of each inventory resource; the shadow price of each inventory resource is the dual value of the recommendation probability of the inventory resource; the shadow price of the at least one inventory resource is obtained by estimating the preference of future arriving users for the at least one inventory resource by using the linear programming method based on the historical preference information of historical users for the at least one inventory resource.

17. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it causes the processor to implement the steps in the method according to any one of claims 1-14.

Images