Recommendation method and system

Abstract

The invention provides a recommendation method and system. The recommendation method comprises the following steps: the recommendation system obtains an access request of a target user to a target page, wherein the target page comprises a plurality of booths; the recommendation system obtains a target recommendation scheme corresponding to a target user from a scheme library, the scheme library comprises recommendation schemes corresponding to all users in a user set, the user set comprises the target user, and the recommendation schemes represent put products corresponding to a plurality of booths respectively; each recommendation scheme in the scheme library is obtained by performing joint planning on the put products corresponding to the plurality of booths by the recommendation system based on the recommendation evaluation index corresponding to the user set. And then, the recommendation system determines to-be-displayed recommendation content of each booth of the target page based on the target recommendation scheme, and the to-be-displayed recommendation content of each booth is related content of a put product corresponding to the booth.

Description

Technical Field

[0001] This specification relates to the field of Internet technologies, and in particular, to a recommendation method and system. Background Art

[0002] With the development of Internet technologies, more and more Internet products have evolved from supporting single-booth recommendations to supporting multi-booth recommendations. Among them, multi-booth recommendation means that there are multiple booths on the user page of an Internet product, and different booths recommend different content to users. This can not only improve the diversity of user choices but also increase the probability of the recommended content being exposed.

[0003] A recommendation system is deployed in an Internet product, and the recommendation system is responsible for deciding which content to recommend to users. Currently, the recommendation algorithm for a single booth is relatively mature and can ensure the best recommendation effect in the single-booth recommendation scenario. For the multi-booth recommendation scenario, the recommendation system can still use the single-booth recommendation algorithm to implement. Specifically, each booth is regarded as a planning object, and the single-booth recommendation algorithm is used to determine the recommended content corresponding to that booth. However, it is found in actual applications that the overall recommendation effect in the multi-booth recommendation scenario is not ideal and cannot achieve the global optimum, which in turn leads to insufficient utilization of traffic resources on the Internet platform and causes waste of resources.

[0004] The content in the background art section is only the information known to the inventor personally, and does not represent that the above information has entered the public domain before the filing date of this disclosure, nor does it represent that it can become the prior art of this disclosure. Summary of the Invention

[0005] This specification provides a recommendation method and system that can use the overall recommendation evaluation index corresponding to a user set as a planning target and jointly plan multiple booths, so that the overall recommendation effect reaches the global optimum, thereby avoiding waste of traffic resources.

[0006] In a first aspect, this specification provides obtaining an access request of a target user for a target page, where the target page includes multiple booths; in response to the access request, obtaining a target recommendation plan corresponding to the target user from a plan library, where the plan library includes recommendation plans corresponding to each user in a user set, the user set includes the target user, the recommendation plan represents the products to be placed corresponding to the multiple booths respectively, and each recommendation plan in the plan library is obtained by jointly planning the products to be placed corresponding to the multiple booths based on the recommendation evaluation index corresponding to the user set; and based on the target recommendation plan, determining the recommended content to be displayed for each booth on the target page, and the recommended content to be displayed for each booth is the content related to the product to be placed corresponding to that booth.

[0007] In some embodiments, the solution library is generated in the following manner: obtaining the behavior prediction data corresponding to each user in the user set; determining an expression of the recommendation evaluation index based on the recommendation solutions corresponding to each user in the user set and the behavior prediction data corresponding to each user; solving to obtain the recommendation solutions corresponding to each user in the user set with the satisfaction of a preset condition by the expression as the solution target; and generating the solution library according to the recommendation solutions corresponding to each user in the user set.

[0008] In some embodiments, the recommendation evaluation index is the number of clicks. The behavior prediction data corresponding to each user includes: exposure prediction data, representing the number of times each of the multiple booths is exposed to the user; and click prediction data, representing the click probability of the user for the promoted product corresponding to each of the multiple booths.

[0009] In some embodiments, determining the expression of the recommendation evaluation index based on the recommendation solutions corresponding to each user in the user set and the behavior prediction data corresponding to each user includes: for the i-th user in the user set, determining an expression of the number of clicks of the i-th user for the multiple booths based on the recommendation solution corresponding to the i-th user and the behavior prediction data corresponding to the i-th user, where the value of i is an integer from 1 to N, and N is the number of users in the user set; and generating an expression corresponding to the recommendation evaluation index based on the sum of the expressions of the number of clicks corresponding to each user in the user set.

[0010] In some embodiments, the method further includes: predicting the exposure prediction data corresponding to each user in the user set through a pre-trained exposure prediction model; and predicting the click prediction data corresponding to each user in the user set through a pre-trained click prediction model.

[0011] In some embodiments, the method further includes: updating the behavior prediction data corresponding to each user in the user set based on the historical behavior data generated by each user in the user set within a recent preset time period, and generating an updated expression corresponding to the recommendation evaluation index based on the updated behavior prediction data; solving to obtain the updated recommendation solutions corresponding to each user in the user set with the satisfaction of a preset condition by the updated expression as the solution target; and updating the solution library according to the updated recommendation solutions corresponding to each user in the user set.

[0012] In some embodiments, the recommendation evaluation index is the number of clicks. Taking the satisfaction of a preset condition by the expression as the solution target includes: taking maximizing the value of the expression as the solution target under the condition of satisfying preset recommendation constraint conditions.

[0013] In some embodiments, the recommendation constraint conditions include at least one of the following: the total number of exposures corresponding to a single product to be placed is within a first preset range; the total number of clicks corresponding to a single product to be placed is within a second preset range; the number of exposures of a single product to be placed within a single booth is within a third preset range;

[0014] the number of clicks of a single product to be placed within a single booth is within a fourth preset range; the number of products to be placed corresponding to a single booth is within a fifth preset range; the number of booths simultaneously placing the same product to be placed is within a sixth preset range; and the products to be placed corresponding to a single user are within a preset product set.

[0015] In some embodiments, the number of the multiple booths is K. Determining the recommended content to be displayed in each booth of the target page based on the target recommendation scheme includes: for the k-th booth among the multiple booths, where k is an integer less than or equal to K: obtaining the target product to be placed corresponding to the k-th booth from the target recommendation scheme, and determining a plurality of candidate contents corresponding to the target product to be placed. According to the user characteristics corresponding to the target user, determining the recommended content to be displayed in the k-th booth from among the plurality of candidate contents.

[0016] In some embodiments, determining the recommended content to be displayed in the k-th booth from among the plurality of candidate contents according to the user characteristics corresponding to the target user includes: obtaining the degree of fit between the user characteristics and each of the plurality of candidate contents; and based on the number M of contents supported for display by the k-th booth, using the top M candidate contents with the highest degree of fit among the plurality of candidate contents as the recommended content to be displayed in the k-th booth.

[0017] In some embodiments, the user characteristics include sub-characteristics in multiple dimensions. Obtaining the degree of fit between the user characteristics and each of the plurality of candidate contents includes: determining at least one target dimension related to the product to be placed corresponding to the k-th booth among the multiple dimensions; obtaining the sub-characteristics of the at least one target dimension from the user characteristics as reference characteristics; and determining the degree of fit between the reference characteristics and each of the plurality of candidate contents.

[0018] In some embodiments, the method is applied to a recommendation system. Obtaining an access request from a target user to a target page includes: receiving the access request from a client; the method further includes: sending the recommended content to be displayed in each booth of the target page to the client, so that the client renders the recommended content into the page area corresponding to the booth in the target page.

[0019] In some embodiments, the recommended content includes at least one of text content, picture content, audio content, video content, product link or jump link.

[0020] In a second aspect, this specification provides a recommendation system, including: at least one storage medium storing at least one instruction set for performing data processing related to content recommendation; and at least one processor communicatively connected to the at least one storage medium, wherein when the recommendation system runs, the at least one processor reads the at least one instruction set and implements the method according to any one of the first aspect based on the indication of the at least one instruction set.

[0021] In a third aspect, this specification also provides a computer-readable non-volatile storage medium, wherein at least one instruction set is stored in the computer-readable non-volatile storage medium, and when the at least one instruction set is executed by at least one processor, the recommendation method provided in the first aspect is implemented.

[0022] Other functions of the recommendation method provided in this specification will be partially listed in the following description. The creative aspects of the recommendation method provided in this specification can be fully explained through practice or use of the methods, devices and combinations described in the detailed examples below. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] To more clearly illustrate the technical solutions in the embodiments of this specification, the following will briefly introduce the drawings required for description in the embodiments. Obviously, the drawings in the following description are only some embodiments of this specification. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0024] Figure 1 FIG. shows a schematic diagram of a recommendation scenario provided according to an embodiment of this specification;

[0025] Figure 2 FIG. shows a hardware structure diagram of a computing system provided according to an embodiment of this specification;

[0026] Figure 3 FIG. shows a flowchart of a recommendation method provided according to an embodiment of this specification;

[0027] Figure 4 FIG. shows a schematic diagram of a target page provided according to an embodiment of this specification;

[0028] Figure 5 FIG. shows a flowchart of generating a solution library in a recommendation method provided according to an embodiment of this specification;

[0029] Figure 6Shows a schematic diagram of a solution library provided according to an embodiment of this specification; and

[0030] Figure 7 Shows a flowchart of the online recommendation stage in a recommendation method provided according to an embodiment of this specification. Detailed implementation manners

[0031] The following description provides specific application scenarios and requirements of this specification, aiming to enable those skilled in the art to manufacture and use the content in this specification. For those skilled in the art, various partial modifications to the disclosed embodiments are obvious, and without departing from the spirit and scope of this specification, the general principles defined here can be applied to other embodiments and applications. Therefore, this specification is not limited to the shown embodiments, but has the broadest scope consistent with the claims.

[0032] The terms used here are only for the purpose of describing specific example embodiments and are not restrictive. For example, unless the context clearly indicates otherwise, the singular forms "a", "an", and "the" used here may also include the plural forms. When used in this specification, the terms "include", "comprise", and / or "contain" mean that the associated integers, steps, operations, elements, and / or components exist, but do not exclude the existence of one or more other features, integers, steps, operations, elements, components, and / or groups, or the addition of other features, integers, steps, operations, elements, components, and / or groups in the system / method.

[0033] Considering the following description, these features of this specification and other features, as well as the operations and functions of the related elements of the structure, and the economy of the combination and manufacture of components can be significantly improved. Referring to the accompanying drawings, all of these form a part of this specification. However, it should be clearly understood that the drawings are only for the purpose of illustration and description and are not intended to limit the scope of this specification. It should also be understood that the drawings are not drawn to scale.

[0034] The flowcharts used in this specification show the operations implemented by a system according to some embodiments in this specification. It should be clearly understood that the operations in the flowchart may not be implemented in sequence. On the contrary, the operations may be implemented in reverse order or simultaneously. In addition, one or more other operations may be added to the flowchart. One or more operations may be removed from the flowchart.

[0035] The following introduces the application scenarios of this specification.

[0036] The technical solution provided in this specification is applicable to the scenario of recommending information to users. In this scenario, when a user accesses a target page through a client, the client displays relevant content of the promoted products in each booth of the target page. Among them, the promoted products can provide services in any form, including but not limited to providing credit services, providing trading services, providing information services, etc. The relevant content of the promoted products is used to display the specific content of the services provided by the promoted products.

[0037] This specification provides a recommendation method that can be executed by a recommendation system. First, the recommendation system can obtain an access request of a target user to a target page, where the target page includes multiple booths. Then, the recommendation system can obtain a target recommendation plan corresponding to the target user from a plan library, where the plan library includes recommendation plans corresponding to each user in a user set, the user set includes the target user, the recommendation plan represents the promoted products corresponding to each of the multiple booths, and each recommendation plan in the plan library is obtained by jointly planning the promoted products corresponding to the multiple booths based on the recommendation evaluation index corresponding to the user set. Finally, the recommendation system can determine the recommended content to be displayed in each booth of the target page based on the target recommendation plan.

[0038] In this specification, the recommendation evaluation index is an evaluation index related to the expected recommendation effect and is used to evaluate whether the current recommendation achieves the expected recommendation effect. The expected recommendation effect is related to the requirements of the actual application scenario, and this specification does not limit the expected recommendation effect. For example, the expected recommendation effect may include, but is not limited to, one or more of the following: high conversion rate, high revenue, high number of clicks, etc.

[0039] In the recommendation plan provided in this specification, the target recommendation plan obtained by the recommendation system comes from a plan library, and the plan library includes recommendation plans corresponding to each user in a user set. Each recommendation plan in the plan library is obtained by jointly planning the promoted products corresponding to the multiple booths based on the recommendation evaluation index corresponding to the user set. Since the recommendation plans in the plan library are jointly planned based on the promoted products corresponding to the multiple booths during generation, considering the coupling relationship between the multiple booths, they are more optimal recommendation plans at the global level. Therefore, when the recommendation system determines the recommended content to be displayed in each booth of the target page based on the target recommendation plan, it can start from the global perspective and use a more reasonable and refined recommended content allocation plan, making the recommendation evaluation index of each recommendation better, thereby improving the utilization rate of traffic by the recommendation system.

[0040] Figure 1 Shows a schematic diagram of a recommendation scenario provided according to an embodiment of this specification. As Figure 1As shown, the scenario 100 may include a recommendation system 11 and N clients 12, where N is an integer greater than or equal to 1. Figure 1 The shown scenario 100 may be a recommendation scenario of an Internet product. For example, an Internet product may include a client and a server. Among them, Figure 1 the client 12 in it may correspond to the client of the Internet product, Figure 1 and the recommendation system 11 in it may correspond to the server of the Internet product or a subsystem in the server of the Internet product.

[0041] Refer to Figure 1 , the target page refers to the page displayed by the client 12 to the user. The target page may include K booths, and each booth is used to display the relevant content of a product to be placed.

[0042] Refer to Figure 1 , the recommendation system 11 may include an offline operation research module. The offline operation research module can pre-jointly plan the products to be placed corresponding to multiple booths based on the recommendation evaluation indicators corresponding to the user set, and obtain the recommendation plan corresponding to each user in the user set. Among them, the recommendation plan corresponding to each user represents the products to be placed corresponding to each of the multiple booths. For example, taking user A as an example, the recommendation plan corresponding to him may represent: recommend product P1 to be placed in booth 1, recommend product P2 to be placed in booth 2, and recommend product P3 to be placed in booth 3. Furthermore, the offline operation research module can generate a plan library based on the recommendation plans corresponding to each user. That is to say, the plan library includes the recommendation plans corresponding to each user in the user set.

[0043] Among them, the user set in this specification may be the set composed of all users of the current Internet product, or the set composed of some users of the current Internet product. This specification does not limit this.

[0044] Refer to Figure 1 , in some embodiments, the recommendation system 11 may further include an online recommendation module. The online recommendation module can be directly or indirectly communicatively connected to the client 12. For example, when the client 12 detects that the target user requests to access the target page, it sends an access request of the target user to the target page to the recommendation system 11. The online recommendation module in the recommendation system 11 receives the access request, and in response to the access request, obtains the target recommendation plan corresponding to the target user from the plan library, and then determines the recommended content to be displayed in each booth of the target page based on the target recommendation plan. Further, the online recommendation module in the recommendation system 11 can send the recommended content to be displayed in each booth of the target page to the client 12, so that the client 12 can render and display the target page based on the above recommended content..

[0045] In some embodiments, the recommendation method provided in this specification can be executed on the recommendation system 11. For example, the recommendation method provided in this specification can be executed by the online recommendation module in the recommendation system 11. At this time, the recommendation system 11 can store data or instructions for executing the recommendation method described in this specification, and can execute or be used to execute the data or instructions. In some embodiments, the recommendation system 11 can include a hardware device with data information processing capabilities and the necessary programs for driving the operation of the hardware device.

[0046] The recommendation system 11 can correspond to a single computing device or a computing cluster composed of multiple computing devices.

[0047] Reference Figure 1 , each client 12 corresponds to a user in the user set, that is, the number of users in the user set is also N. The client 12 can, in response to the operation of the target user, send a request to access the target page to the recommendation system based on the identification of the target user in the recommendation system.

[0048] In some embodiments, one or more application programs (APPs) can be installed on the client 12. The APP can provide the ability to receive user operations and an interface. The APP includes, but is not limited to: financial APP programs, web browser APP programs, search APP programs, chat APP programs, shopping APP programs, video APP programs, wealth management APP programs, instant messaging tools, email clients, social platform software, and so on.

[0049] In some embodiments, a target APP can be installed on the client 12. The client 12 can receive user operations through the target APP. In some embodiments, the target APP can, in response to receiving a user operation, send a request to access the target page to the recommendation system 11. The target APP can also receive data of the target page from the recommendation system 11 and display the target page on the client 12 (display relevant content of the corresponding promoted products in each booth of the target page).

[0050] Since the recommendation schemes corresponding to the users in the user set are pre-generated, in some embodiments, the recommendation method provided in this specification can also be executed by the client 12. At this time, the client 12 can store data or instructions for executing the recommendation method described in this specification, and can execute or be used to execute the data or instructions. In some embodiments, the client 12 can include a hardware device with data information processing capabilities and the necessary programs for driving the operation of the hardware device.

[0051] For example, after the recommendation system 11 generates a solution library through the offline operation research module, it can distribute each recommended solution in the solution library to the client 12 corresponding to each user. The client 12 stores the recommended solution locally. In this way, when the client 12 obtains an access request from the target user to the target page, it can obtain the target recommended solution corresponding to the target user from the local, and then determine the recommended content to be displayed in each booth of the target page based on the target recommended solution. For another example, the solution library is stored in the recommendation system 11. When the client 12 obtains an access request from the target user to the target page, it can request the target recommended solution corresponding to the target user from the recommendation system 11. Then, the client 12 determines the recommended content to be displayed in each booth of the target page based on the target recommended solution.

[0052] It should be understood that Figure 1 the number of clients 12 in

[0053] Figure 2 shows a hardware structure diagram of a computing system provided according to an embodiment of the present specification. The computing system 200 can be used as Figure 1 the recommendation system 11 in

[0054] As Figure 2 shown, the computing system 200 may include at least one storage medium 230 and at least one processor 220. In some embodiments, the computing system 200 may further include a communication port 250 and an internal communication bus 210. The computing system 200 may further include I / O components 260.

[0055] The internal communication bus 210 can connect different system components. For example, the internal communication bus 210 can connect the storage medium 230, the processor 220, the communication port 250, and the I / O components 260, etc.

[0056] The I / O components 260 support input / output between the computing system 200 and other components.

[0057] The communication port 250 is used for data communication between the computing system 200 and the outside world. For example, the communication port 250 can be used for data communication between the computing system 200 and the network. The communication port 250 can be a wired communication port or a wireless communication port.

[0058] The storage medium 230 may include a data storage device. The data storage device may be a non-transitory storage medium or a transitory storage medium. For example, the data storage device may include one or more of a magnetic disk 232, a read-only storage medium (ROM) 234, or a random access storage medium (RAM) 235. The storage medium 230 further includes at least one instruction set stored in the data storage device. The instruction set may include computer program code, and the computer program code may include programs, routines, objects, components, data structures, procedures, modules, and so on.

[0059] At least one processor 220 may be communicatively connected to at least one storage medium 230. When the computing system 200 is running, at least one processor 220 reads the at least one instruction set and, according to the instructions of the at least one instruction set, executes the recommendation method provided in this specification. The processor 220 may execute the steps included in the recommendation method. The processor 220 may be in the form of one or more processors. In some embodiments, the processor 220 may include one or more hardware processors, such as a microcontroller, a microprocessor, a reduced instruction set computer (RISC), an application-specific integrated circuit (ASIC), an application-specific instruction set processor (ASIP), a central processing unit (CPU), a graphics processing unit (GPU), a physics processing unit (PPU), a microcontroller unit, a digital signal processor (DSP), a field-programmable gate array (FPGA), an advanced RISC machine (ARM), a programmable logic device (PLD), any circuit or processor capable of executing one or more functions, or any combination thereof.

[0060] For illustrative purposes only, only one processor 220 is shown in the computing system 200 in the drawings. However, it should be noted that the computing system 200 in this specification may also include multiple processors. Therefore, the operations and / or method steps disclosed in this specification may be executed by one processor or jointly executed by multiple processors. For example, if it is described in this specification that the processor 220 of the computing system 200 executes step A and step B, it should be understood that step A and step B may also be jointly or separately executed by two different processors 220 (for example, the first processor executes step A, the second processor executes step B, or the first and second processors jointly execute steps A and B).

[0061] Figure 3 A flowchart of a recommendation method provided according to an embodiment of this specification is shown. As before, the recommendation system 11 or the client 12 may execute this recommendation method. For convenience of description, the recommendation system 11 is used as an example of the execution entity in the following description.

[0062] As Figure 3 shown, the recommendation method may include:

[0063] S310: Obtain an access request of a target user for a target page, where the target page includes multiple booths.

[0064] Among them, the access request for the target page can be sent by the client to the recommendation system. Triggering the client to send an access request for the target page to the recommendation system can include various situations, which are not limited in this specification. For example, when the client responds to the user's operation of clicking the link corresponding to the target page, it can trigger the client to send an access request for the target page to the recommendation system. Alternatively, the client can also respond to detecting a preset voice instruction, responding to detecting a preset action, or detecting being in a target environment to trigger the client to send an access request for the target page to the recommendation system.

[0065] Figure 4 Shows a schematic diagram of a target page provided according to an embodiment of this specification.

[0066] In some embodiments, referring to Figure 4 , the target page may include K booths, where K can be an integer greater than or equal to 1. Each booth can be used to display relevant information of the corresponding product to be put on the market. The product to be put on the market can be a tangible commodity or an intangible commodity. For example, the product to be put on the market is used to provide corresponding services, such as providing credit services, providing trading services, and providing information services, etc. In Figure 4 , the target page is a page related to the bill, where content related to the bill is shown, such as "Spending this month AAAAA (yuan)", "Total bill BBBBB", and "Total limit CCCCC". Below the area where "Spending this month AAAAA (yuan)" is shown, Booth 1 can be set. Booth 1 can be used to display relevant information of a certain product to be put on the market to the user, such as displaying relevant information of the spending analysis service.

[0067] Continuing to refer to Figure 4 , below the area where "Total bill BBBBB" and "Total limit CCCCC" are shown, Booth 2 can be set. Booth 2 can be used to display relevant information of a certain product to be put on the market to the user, such as displaying relevant information of the installment repayment service or the credit limit increase service.

[0068] In Figure 4 , there are also other multiple booths, such as Booth 3, Booth 4, and Booth 5, and these booths can all be used to display relevant information of a certain product to be put on the market to the user. This specification does not list and explain them one by one.

[0069] S320: In response to the access request, obtain a target recommendation plan corresponding to the target user from the plan library, where the plan library includes recommendation plans corresponding to each user in the user set, the user set includes the target user, the recommendation plan represents the products to be placed corresponding to multiple exhibition booths, and each recommendation plan in the plan library is obtained by jointly planning the products to be placed corresponding to multiple exhibition booths based on the recommendation evaluation indicators corresponding to the user set.

[0070] In this specification, the entire recommendation process of the recommendation system can be regarded as two stages, namely the offline operation research stage and the online recommendation stage. Refer to Figure 1 , the offline operation research stage can be implemented through the offline operation research module in the recommendation system, and the online recommendation stage can be implemented through the online recommendation module in the recommendation system. The purpose of the offline operation research stage is to generate or update the recommendation plans in the plan library, and the purpose of the online recommendation stage is to perform online recommendations using the recommendation plans in the plan library.

[0071] In the offline operation research stage of the recommendation system, the recommendation evaluation indicators corresponding to the user set can be used to jointly plan the products to be placed corresponding to multiple exhibition booths, and generate or update the recommendation plans in the plan library. Hereinafter, the implementation method for generating or updating the recommendation plans in the plan library in the offline operation research stage of the recommendation system will be described.

[0072] Figure 5 shows a flowchart for generating a plan library in a recommendation method provided according to an embodiment of this specification.

[0073] Refer to Figure 5 , the steps for generating the plan library include:

[0074] S510: Obtain the behavior prediction data corresponding to each user in the user set.

[0075] In some embodiments, the users in the user set can be users who have registered in the recommendation system. The recommendation system can collect the historical behavior data of the users with the user's authorization. When generating the plan library, the recommendation system can obtain the behavior prediction data corresponding to each user through a prediction model according to the historical behavior data corresponding to each user. When a user generates new behavior data, the recommendation system can update the historical behavior data of the user according to the newly generated behavior data, and the updated historical behavior data can be used to update the behavior prediction data.

[0076] In some embodiments, the historical behavior data of a user may include the user's characteristic data, the user's historical click data, the user's historical access data, context information, etc. The behavior prediction data corresponding to the user may include exposure prediction data and click prediction data. Among them, the exposure prediction data is used to represent the number of exposures of multiple exhibition booths to the user respectively, and the click prediction data is used to represent the click probability of the user for the promoted products corresponding to each of the multiple exhibition booths.

[0077] In some embodiments, the recommendation system may, through a pre-trained exposure prediction model, predict the exposure prediction data corresponding to each user in the user set, and through a pre-trained click prediction model, predict the click prediction data corresponding to each user in the user set.

[0078] In some embodiments, the exposure prediction model may be one of a linear regression model, a decision tree and a random forest model, a gradient boosting decision tree model, a deep learning model or a time series model. When training the exposure prediction model, historical data affecting the exposure of the exhibition booth may be obtained first, including but not limited to the historical number of exposures of the exhibition booth to the user, the user's historical behavior data, the page features of the target page, the exhibition booth features, the context features, etc. Then, the historical data affecting the exposure of the exhibition booth is split into a training set, a validation set and a test set. As an example, the training set may include 70%-80% of the data, and the validation set and the test set may include 20%-30% of the data. Next, using at least one of the mean squared error, the root mean squared error, and the mean absolute error as the loss, the exposure prediction model is trained based on the training set to obtain a pre-trained exposure prediction model. Finally, the pre-trained exposure prediction model is verified and tested based on the validation set and the test set to evaluate the performance of the pre-trained exposure prediction model. When the performance of the pre-trained exposure prediction model meets the standard, the recommendation system may use the pre-trained exposure prediction model to obtain the exposure prediction data corresponding to each user in the user set.

[0079] In some embodiments, the click prediction model may be one of a logistic regression model, a decision tree and a random forest model, a gradient boosting decision tree model, a deep learning model, a factorization machine and a deep factorization machine model. When training the click prediction model, historical data affecting the click probability may be obtained first, including but not limited to the user's historical behavior data, the exhibition booth features of the exhibition booth in the target page, the context features, etc. The process of training the click prediction model based on the obtained historical data affecting the click probability to obtain a pre-trained click prediction model is similar to the process of obtaining a pre-trained click prediction model, and will not be elaborated here. When the performance of the pre-trained click prediction model meets the standard, the recommendation system may use the pre-trained click prediction model to obtain the click prediction data corresponding to each user in the user set.

[0080] S520: Determine the expression of the recommendation evaluation index based on the recommendation solutions corresponding to each user in the user set and the behavior prediction data corresponding to each user.

[0081] When the recommendation system displays the relevant content of the corresponding promoted products in each booth on the target page, different promoted products are displayed for different users, which can bring different recommendation effects. The effect expected to be achieved by the recommendation solutions provided in this specification is that the promoted products displayed on the target page have better recommendation effects at the global level. Among them, the global level includes at least two aspects. One aspect is to consider all users in the user set, that is, to achieve better recommendation effects in the dimension of all users in the user set. Another aspect is to consider all booths on the target page, that is, to also achieve better recommendation effects in the dimension of all booths.

[0082] Based on the above analysis, the process of determining the recommendation solution can be transformed into a process of modeling and solving. That is, taking the promoted products corresponding to each booth as unknowns, an expression of the recommendation evaluation index (an index related to the expected recommendation effect) is modeled based on the expected recommendation effect, and then, by solving this expression, the promoted products corresponding to each booth are obtained.

[0083] Furthermore, the recommendation effect can be quantified by the recommendation evaluation index. For example, whether the promoted product can be clicked is the main factor determining the recommendation effect. In this specification, the number of clicks can be used as the recommendation evaluation index to evaluate the recommendation effect. Therefore, in order to make the expression of the recommendation evaluation index more accurately reflect the expected recommendation effect, the following method is adopted in modeling in this specification: determine the expression of the number of clicks of the i-th user on multiple booths, and generate the corresponding expression of the recommendation evaluation index based on the sum of the expressions of the number of clicks corresponding to each user in the user set.

[0084] Those skilled in the art can understand that the recommendation solution obtained by jointly planning the promoted products corresponding to multiple booths based on the recommendation evaluation index corresponding to the user set. Such a recommendation solution considers the coupling relationship between multiple booths and is a better recommendation solution at the global level. Furthermore, when the recommendation system displays the relevant content of the corresponding promoted products in each booth on the target page based on the target recommendation solution, it can start from the global perspective and use a more reasonable and refined recommendation content distribution plan, making the recommendation evaluation index of each recommendation better, thereby improving the utilization rate of traffic by the recommendation system.

[0085] In some embodiments, based on the recommendation solutions corresponding to each user in the user set and the behavior prediction data corresponding to each user, an expression of the recommendation evaluation index is determined, including: for the i-th user in the user set, based on the recommendation solution corresponding to the i-th user and the behavior prediction data corresponding to the i-th user, an expression of the number of clicks of the i-th user on multiple exhibition booths is determined, where the value of i is an integer from 1 to N, and N is the number of users in the user set; and based on the sum of the click number expressions corresponding to each user in the user set, an expression corresponding to the recommendation evaluation index is generated.

[0086] In the modeling process, the products placed in each exhibition booth are used as unknowns. This unknown can be denoted as x ijk , indicating whether product j is displayed at booth k for user i. x ijk is a binary variable, x ijk ∈{0,1}. When the value of x ijk is 0, it means that product j is not displayed at booth k for user i, and when the value of x ijk is 1, it means that product j is displayed at booth k for user i.

[0087] In some embodiments, based on the expected recommendation effect, an expression of the recommendation evaluation index is modeled and can be represented by the following formula:

[0088]

[0089] where I represents the user set, J represents the product set, and K represents the booth set in the target page. expo ik represents the number of exposures of booth k to user i, ctr ijk represents the click probability of user i when product j is displayed at booth k. x ijk ·expo ik ·ctr ijk then represents the number of clicks of user i when product j is displayed at booth k for user i. Therefore, the above formula represents the sum of the number of clicks of all users in the user set I on all booths.

[0090] S530: Taking the expression satisfying the preset conditions as the solution target, the recommendation solutions corresponding to each user in the user set are obtained by solving.

[0091] In some embodiments, when the recommendation evaluation index has a positive correlation with the expected recommendation effect (that is, the greater the recommendation evaluation index, the better the recommendation effect), the recommendation system can, under the condition of satisfying the preset recommendation constraint conditions, take maximizing the value of the expression as the solution target. That is, the recommendation system can use the following formula as the solution target:

[0092]

[0093] Among them, the recommended constraint conditions may include one or more of Conditions 1 to 6 described below.

[0094] Condition 1

[0095]

[0096] Among them, el j represents the lower limit of the total exposure times corresponding to a single placed product, and eu j represents the upper limit of the total exposure times corresponding to a single placed product. The recommended constraint condition represented by this formula is that the total exposure times corresponding to a single placed product are within the first preset range (that is, greater than or equal to el j , and less than or equal to eu j ). In the formula corresponding to Condition 1, the total exposure times corresponding to a single placed product refer to the sum of the exposure times of a single placed product considering all users in the user set and considering all exhibition booths.

[0097] It should be noted that the formula corresponding to the above Condition 1 is only taken as a possible example. In actual applications, based on the requirements of different recommendation scenarios, there may be some deformation forms of the formula of the above Condition 1, which are not listed in this specification. In addition, the values of el j and eu j can depend on the actual requirements of the recommendation scenario.

[0098] Condition 2

[0099]

[0100] Among them, cl j represents the lower limit of the total click times corresponding to a single placed product, and cu j represents the upper limit of the total click times corresponding to a single placed product. The recommended constraint condition represented by this formula is that the total click times corresponding to a single placed product are within the second preset range (that is, greater than or equal to cl j , and less than or equal to cu j ). In the formula corresponding to Condition 2, the total click times corresponding to a single placed product refer to the sum of the click times of a single placed product considering all users in the user set and considering all exhibition booths.

[0101] It should be noted that the formula corresponding to the above Condition 2 is only taken as a possible example. In actual applications, based on the requirements of different recommendation scenarios, there may be some deformation forms of the formula corresponding to the above Condition 2, which are not listed in this specification. In addition, cl j and cuj The value of

[0102] Condition Three

[0103]

[0104] where el′ kj represents the lower limit of the number of exposures of a single product to be placed in a single booth, and eu′ kj represents the upper limit of the number of exposures of a single product to be placed in a single booth. The recommended constraint condition represented by this formula is that the number of exposures of a single product to be placed in a single booth is within the third preset range (i.e., greater than or equal to el′ kj , and less than or equal to eu′ kj ). In the formula corresponding to Condition Three, the number of exposures of a single product to be placed in a single booth refers to the sum of the number of exposures of a single product to be placed in a single booth considering all users in the user set.

[0105] It should be noted that the formula corresponding to the above Condition Three is only taken as a possible example. In actual applications, based on the requirements of different recommendation scenarios, there may be some deformation forms of the formula corresponding to the above Condition Three, which are not listed in this specification. In addition, the values of el′ kj and eu′ kj can depend on the actual requirements of the recommendation scenario.

[0106] Condition Four

[0107]

[0108] where cl′ kj represents the lower limit of the number of clicks of a single product to be placed in a single booth, and cu′ kj represents the upper limit of the number of clicks of a single product to be placed in a single booth. The recommended constraint condition represented by this formula is that the number of clicks of a single product to be placed in a single booth is within the fourth preset range (i.e., greater than or equal to cl′ kj , and less than or equal to cu′ kj ). In the formula corresponding to Condition Four, the number of clicks of a single product to be placed in a single booth refers to the sum of the number of clicks of a single product to be placed in a single booth considering all users in the user set.

[0109] It should be noted that the formula corresponding to the above Condition Four is only taken as a possible example. In actual applications, based on the requirements of different recommendation scenarios, there may be some deformation forms of the formula corresponding to the above Condition Four, which are not listed in this specification. In addition, the values of cl′ kj and cu′ kjIts value can depend on the actual requirements of the recommendation scenario.

[0110] Condition Five

[0111]

[0112] Among them, r k represents the number of products to be placed corresponding to a single booth. The recommendation constraint condition represented by this formula is: the number of products to be placed corresponding to a single booth is within the fifth preset range (i.e., greater than or equal to 1 and less than or equal to r k ). In the formula corresponding to Condition Five, the number of products to be placed corresponding to a single booth refers to the number of products to be placed corresponding to a single booth when considering a single user in the user set.

[0113] It should be noted that the formula corresponding to the above Condition Five is only taken as a possible example. In actual applications, based on the requirements of different recommendation scenarios, there may be some variant forms of the formula corresponding to the above Condition Five, which are not listed in this specification. In addition, the value of r k can depend on the actual requirements of the recommendation scenario. For example, the value of r k can be 1, or an integer greater than 1.

[0114] Condition Six

[0115]

[0116] Among them, s represents the number of booths that simultaneously place the same product to be placed on the target page. The recommendation constraint condition represented by this formula is: the number of booths that simultaneously place the same product to be placed is within the sixth preset range (less than or equal to s). In the formula corresponding to Condition Six, the number of booths that simultaneously place the same product to be placed refers to the number of booths that simultaneously place the same product to be placed in a single target page when considering a single user in the user set.

[0117] It should be noted that the formula corresponding to the above Condition Six is only taken as a possible example. In actual applications, based on the requirements of different recommendation scenarios, there may be some variant forms of the formula corresponding to the above Condition Six, which are not listed in this specification. In addition, the value of s can depend on the actual requirements of the recommendation scenario. For example, the value of s can be 1, or an integer greater than 1.

[0118] In some embodiments, the recommendation constraint condition may further include a preset product set corresponding to a single user, and the preset product set is used to represent the products to be placed corresponding to a single user. When the recommendation system confirms the products to be placed corresponding to the user, it can select from the products to be placed within the preset product set corresponding to the single user.

[0119] In this specification, the recommendation system can, when meeting the preset recommendation constraint conditions, take maximizing the value of the expression of the recommendation evaluation index as the solution objective, and solve to obtain the recommendation solution corresponding to each user. The expression of the recommendation evaluation index of the recommendation system and the recommendation constraint conditions constrain multiple exhibition positions on the target page and the products to be placed corresponding to each exhibition position, fully considering the coupling relationship between the various exhibition positions on the target page, so as to be able to generate a more reasonable and refined recommendation solution from a global perspective. When using these recommendation solutions to display the products to be placed, the recommendation evaluation index is better, thereby improving the utilization rate of traffic by the recommendation system.

[0120] S540: Generate a solution library according to the recommendation solutions corresponding to the users in the user set.

[0121] In some embodiments, the recommendation system can store the recommendation solutions generated according to steps S510 - S530 in the target database to form a solution library. When storing the recommendation solutions, the recommendation system can also store the identifiers of the users corresponding to the recommendation solutions. In this way, when receiving an access request from a target user for the target page, the corresponding target recommendation solution can be matched in the solution library according to the identifier of the target user.

[0122] Figure 6 Shows a schematic diagram of the solution library provided according to the embodiments of this specification. As Figure 6 shown, the solution library includes the identifiers of the users in the user set and the recommendation solutions corresponding to each user. Among them, the recommendation solution corresponding to each user includes: the products to be placed corresponding to multiple exhibition positions on the target page.

[0123] In some embodiments, after the solution library is generated, the recommendation system can also update the behavior prediction data corresponding to the users in the user set based on the historical behavior data generated by the users in the user set within the recent preset time period, and generate an updated expression corresponding to the recommendation evaluation index based on the updated behavior prediction data; take the updated expression satisfying the preset conditions as the solution objective, solve to obtain the updated recommendation solutions corresponding to the users in the user set; and update the solution library according to the updated recommendation solutions corresponding to the users in the user set.

[0124] In some embodiments, the recommendation system may update the solution library at a preset time duration. The preset time duration can be adjusted according to the requirements in actual applications, and this specification does not limit it. In some embodiments, the recommendation system may also update the solution library based on other triggering conditions. The recommendation system updates the behavior prediction data corresponding to each user in the user set, and generates an updated expression corresponding to the recommendation evaluation index based on the updated behavior prediction data, and re-solves to obtain an updated recommendation solution. The manner of updating the solution library is the same as the manner of generating the solution library described above, and will not be elaborated here.

[0125] S330: Based on the target recommendation solution, determine the recommended content to be displayed for each booth on the target page. The recommended content to be displayed for each booth is the content related to the product corresponding to the placement of this booth.

[0126] In some embodiments, S330 is the online recommendation stage of the recommendation system. In the online recommendation stage, the recommendation system can, according to the target recommendation solution corresponding to the target user and based on the user characteristics corresponding to the target user, determine the target content corresponding to each placement product. Hereinafter, the implementation manner of the recommendation system determining the target content corresponding to each placement product in the online recommendation stage will be described.

[0127] Figure 7 The flowchart of the online recommendation stage in a recommendation method provided according to an embodiment of this specification is shown.

[0128] In some embodiments, the number of multiple booths on the target page is K. Refer to Figure 7 , for the k-th booth among the multiple booths, where k is an integer less than or equal to K, the recommendation method includes:

[0129] S710: Obtain the target placement product corresponding to the k-th booth from the target recommendation solution, and determine multiple candidate contents corresponding to the target placement product.

[0130] As described above, the target recommendation solution includes the target placement products corresponding to multiple booths on the target page. For example, booth 1 corresponds to placement product P1, booth 2 corresponds to placement product P2, and booth 3 corresponds to placement product P3.

[0131] Generally, each placement product may correspond to multiple candidate contents. Therefore, in S710, the recommendation system can recall multiple candidate contents corresponding to the target placement product based on the target placement product corresponding to the k-th booth.

[0132] Among them, the candidate content refers to the content related to the corresponding product to be delivered. The candidate content may include at least one of text content, picture content, audio content, video content, product link or jump link. For example, when the target product to be delivered is a service, the candidate content may be the access link of the service, the news of the service, the text, picture or video for introducing the service content, etc. For example, when the target product to be delivered is a product, the candidate content may be the access link of the product, the picture of the product, the text or video for introducing the product, the trial evaluation related to the product, etc.

[0133] In some embodiments, multiple candidate contents corresponding to the target product to be delivered can be configured according to the actual situation. For example, the recommendation system can configure multiple candidate contents for the target product to be delivered based on the behavior dimension of the target user, the time dimension of accessing the target page, the geographical dimension of accessing the target page, the customization requirements of the candidate content, etc. Among them, the recommendation system can store the relevant data of each candidate content locally or in the corresponding cloud, and this specification does not limit this.

[0134] S720: Determine the recommended content to be displayed in the k-th booth from multiple candidate contents according to the user characteristics corresponding to the target user.

[0135] In some embodiments, the recommendation system can first obtain the adaptation degrees between the user characteristics and multiple candidate contents respectively. Furthermore, based on the number M of contents supported to be displayed in the k-th booth, the top M candidate contents with the highest adaptation degrees among the multiple candidate contents are used as the recommended content to be displayed in the k-th booth. The higher the adaptation degree between the candidate content and the user characteristics, the more it means that the candidate content is more in line with the interests and hobbies of the target user or easier to meet the needs of the user. This can increase the probability that the recommended content is clicked by the target user.

[0136] In some embodiments, the user characteristics corresponding to the target user may include multiple dimensions. For example, the user characteristics may include the basic information of the user and historical behaviors. Among them, the historical behaviors may include user interest preferences, user historical click behaviors, user historical retrieval behaviors, etc.

[0137] As an example, the recommendation system can determine at least one target dimension related to the product to be delivered corresponding to the k-th booth among multiple dimensions. Then, the recommendation system can obtain the sub-characteristics of at least one target dimension from the user characteristics as reference characteristics, and determine the adaptation degrees between the reference characteristics and multiple candidate contents respectively. In this solution, when the recommendation system determines the adaptation degree corresponding to each candidate content, it is not based on all the content of the user characteristics, but on the sub-characteristics related to the product to be delivered corresponding to the k-th booth in the user characteristics. This makes the adaptation degrees determined by the recommendation system more targeted and can determine more accurate recommended content for different booths.

[0138] For example, assume that the product to be placed corresponding to the k-th booth is for recommending products. Then, the at least one target dimension determined by the recommendation system for the k-th booth may include user interest preferences and user historical retrieval behaviors.

[0139] The recommendation system can obtain sub-features corresponding to user interest preferences and user historical retrieval behaviors from user features. The sub-feature corresponding to user interest preferences is used to characterize the user preferences extracted based on the user's historical purchase records. For example, the sub-feature corresponding to user interest preferences may include "electronic products", "beauty products", "food", etc. The sub-feature corresponding to user historical retrieval behaviors is used to characterize the types of products searched by the user within a period of time, and the sub-feature corresponding to user historical retrieval behaviors may include "mobile phones", "lipsticks", "sunflower seeds", etc.

[0140] In some embodiments, the recommendation system can calculate the fitness between the reference feature and multiple candidate contents through a preset fitness algorithm. As an example, the fitness algorithm may include collaborative filtering algorithms, matrix factorization algorithms, feature representation and matching degree calculation methods, Neural Collaborative Filtering (NCF) algorithms, etc.

[0141] In some embodiments, a booth may support displaying multiple contents. As an example, when the booth is displayed in a dynamic manner such as scrolling display, sliding display, etc., or supports multiple windows during static display, the booth can support displaying multiple contents. For example, when the product to be placed corresponding to the k-th booth is for recommending products, multiple products can be respectively displayed through multiple windows in the booth. That is, when determining the target content to be displayed in the k-th booth, it is necessary to confirm M candidate contents from multiple candidate contents as the target content to be displayed in the k-th booth according to the number M of contents supported by the k-th booth for display.

[0142] In this embodiment, the recommendation system taking the top M candidate contents with the highest fitness among multiple candidate contents as the target content to be displayed in the k-th booth can make the k-th booth more likely to be clicked by users when displaying the target content, thereby improving the utilization rate of traffic by the recommendation system.

[0143] In some embodiments, when the recommendation method described in this specification is executed by a recommendation system, the method may further include S730.

[0144] S730: Sending the recommended content to be displayed in each booth in the target page to the client, so that the client renders the recommended content to the page area corresponding to the booth in the target page.

[0145] For the k-th booth in the target page, the client renders the recommended content corresponding to the k-th booth to the page area corresponding to the k-th booth according to the display mode corresponding to the k-th booth. The display mode includes at least one of static display, dynamic display, or pop-up display.

[0146] In summary, in the recommendation method and system provided in this specification, the target recommendation plan obtained by the recommendation system comes from the plan library, and the plan library includes the recommendation plans corresponding to each user in the user set. Each recommendation plan in the plan library is obtained by jointly planning the products to be placed corresponding to multiple booths based on the recommendation evaluation metrics corresponding to the user set. Since the recommendation plans in the plan library are jointly planned based on the products to be placed corresponding to multiple booths and consider the coupling relationship between multiple booths, they are better recommendation plans at the global level. Therefore, when the recommendation system determines the recommended content for each booth on the target page based on the target recommendation plan, it can start from the global perspective and use a more reasonable and refined recommended content distribution plan, making the recommendation evaluation metrics of each recommendation better, thereby improving the utilization rate of traffic by the recommendation system.

[0147] On the other hand, this specification provides a computer-readable non-transitory storage medium storing at least one instruction set for content recommendation. When the at least one instruction set is executed by a processor, the at least one instruction set directs the processor to perform the steps of the recommendation method described in this specification. In some possible implementation manners, each aspect of this specification may also be implemented in the form of a program product, which includes program code. When the program product runs on a computing system 200, the program code is used to cause the computing system 200 to perform the steps of the recommendation method described in this specification. The program product for implementing the above method may adopt a portable compact disc read-only memory (CD-ROM) including program code and may run on the computing system 200. However, the program product of this specification is not limited thereto. In this specification, the readable storage medium may be any tangible medium that contains or stores a program, and this program may be used by or combined with an instruction execution system. The program product may adopt any combination of one or more readable media. The readable media may be a readable signal medium or a readable storage medium. The readable storage medium may, for example, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. The computer-readable storage medium may include a data signal propagated in a baseband or as part of a carrier wave, where the data signal carries the readable program code. Such a propagated data signal may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The readable storage medium may also be any readable medium other than the readable storage medium, and this readable medium may send, propagate, or transmit a program for use by or combined with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium may be transmitted by any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the above. The program code for performing the operations of this specification may be written in any combination of one or more programming languages, including object-oriented programming languages such as Java, C++, etc., and also including conventional procedural programming languages such as the "C" language or similar programming languages. The program code may be executed entirely on the computing system 200, partially on the computing system 200, executed as an independent software package, partially on the computing system 200 and partially on a remote computing device, or entirely on a remote computing device.

[0148] The above description has been made of specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the acts or steps recited in the claims may be performed in a different order than in the embodiments and still achieve the desired result. Additionally, the processes depicted in the drawings do not necessarily require a particular order or a sequential order to achieve the desired result. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0149] In summary, after reading this detailed disclosure, those skilled in the art will appreciate that the foregoing detailed disclosure may be presented by way of example only and is not necessarily limiting. Although not explicitly stated herein, those skilled in the art will understand that this specification is intended to embrace various reasonable changes, improvements, and modifications to the embodiments. These changes, improvements, and modifications are intended to be proposed by this specification and are within the spirit and scope of the exemplary embodiments of this specification.

[0150] Furthermore, certain terms in this specification have been used to describe embodiments of this specification. For example, "one embodiment", "an embodiment", and / or "some embodiments" mean that the specific features, structures, or characteristics described in connection with that embodiment may be included in at least one embodiment of this specification. Thus, it should be emphasized and understood that two or more references to "an embodiment" or "one embodiment" or "alternative embodiments" in various parts of this specification do not necessarily all refer to the same embodiment. Additionally, the specific features, structures, or characteristics may be appropriately combined in one or more embodiments of this specification.

[0151] It should be understood that in the foregoing description of the embodiments of this specification, for the purpose of helping to understand a feature and for the purpose of simplifying this specification, this specification combines various features in a single embodiment, drawing, or its description. However, this does not mean that the combination of these features is necessary, and it is entirely possible for those skilled in the art, when reading this specification, to mark out some of the devices as separate embodiments for understanding. That is to say, the embodiments in this specification can also be understood as the integration of multiple sub - embodiments. And the content of each sub - embodiment is also valid when it has fewer features than all the features of a single foregoing disclosed embodiment.

[0152] Each patent, patent application, publication of patent application, and other materials cited herein, such as articles, books, specifications, publications, documents, items, etc., except for those that are inconsistent with or conflict with this document, or those that have a limiting effect on the broadest scope of the claims, may be incorporated herein by reference and used for all purposes now or hereafter related to this document. In addition, in the event of any inconsistency or conflict between the description, definition, and / or use of relevant terms in any material and the description, definition, and / or use of relevant terms in this document, the terms in this document shall prevail.

[0153] Finally, it should be understood that the embodiments of the application disclosed herein are illustrative of the principles of the embodiments of this specification. Other modified embodiments are also within the scope of this specification. Therefore, the embodiments disclosed in this specification are merely examples and not limitations. Those skilled in the art can adopt alternative configurations based on the embodiments in this specification to implement the application in this specification. Therefore, the embodiments of this specification are not limited to the embodiments precisely described in the application.

Claims

1. A recommendation method, comprising: Obtaining a target user's access request to a target page, wherein the target page includes a plurality of booths; In response to the access request, a target recommendation scheme corresponding to the target user is obtained from a scheme library, wherein the scheme library includes recommendation schemes corresponding to respective users in a user set, the user set includes the target user, the recommendation schemes represent the products corresponding to the multiple booths, and each recommendation scheme in the scheme library is obtained by jointly planning the recommendation evaluation index corresponding to the user set and the products corresponding to the multiple booths; as well as Based on the target recommendation scheme, the recommended content to be displayed at each booth of the target page is determined, and the recommended content to be displayed at each booth is content related to the product placed at the booth.

2. The method according to claim 1, wherein: The solution library is generated in the following way: Obtaining behavior prediction data corresponding to each user in the user set; Determining an expression of the recommendation evaluation index based on the recommendation scheme corresponding to each user in the user set and the behavior prediction data corresponding to each user; Taking the expression satisfying a preset condition as a solution goal, solving to obtain a recommendation solution corresponding to each user in the user set; as well as The solution library is generated according to the recommendation solution corresponding to each user in the user set.

3. The method according to claim 2, wherein: The recommendation evaluation index is the number of clicks, and the behavior prediction data corresponding to each user includes: Exposure prediction data, representing the number of times each of the multiple booths is exposed to the user; and The click prediction data represents the click probability of the user on the product placed in each of the multiple booths.

4. The method according to claim 3, wherein: The step of determining the expression of the recommendation evaluation index based on the recommendation scheme corresponding to each user in the user set and the behavior prediction data corresponding to each user includes: For the ith user in the user set, based on the recommendation scheme corresponding to the ith user and the behavior prediction data corresponding to the ith user, determine an expression for the number of clicks of the ith user on the plurality of booths, where the value of i is an integer between 1 and N, and N is the number of users in the user set; and Based on the sum of the click count expressions corresponding to the users in the user set, an expression corresponding to the recommendation evaluation index is generated.

5. The method according to claim 3, wherein: The method further comprises: Predicting the exposure prediction data corresponding to each user in the user set by using a pre-trained exposure prediction model; and The click prediction data corresponding to each user in the user set is predicted by using a pre-trained click prediction model.

6. The method according to claim 2, wherein: The method further comprises: Based on the historical behavior data generated by each user in the user set within a recent preset time period, the behavior prediction data corresponding to each user in the user set is updated, and an updated expression corresponding to the recommendation evaluation index is generated based on the updated behavior prediction data; Taking the updated expression satisfying a preset condition as a solving goal, solving to obtain an updated recommendation scheme corresponding to each user in the user set; and The solution library is updated according to the updated recommendation solution corresponding to each user in the user set.

7. The method according to claim 2, wherein: The recommendation evaluation index is the number of clicks, and the solution goal is to satisfy the preset condition of the expression, including: Under the condition that the preset recommended constraint conditions are met, the solution goal is to maximize the value of the expression.

8. The method according to claim 7, wherein: The recommended constraint condition includes at least one of the following: The total number of exposures corresponding to a single product is within the first preset range; The total number of clicks corresponding to a single delivery product is within the second preset range; The number of exposures of a single product in a single booth is within the third preset range; The number of clicks on a single product in a single booth is within the fourth preset range; The number of products placed in a single booth is within the fifth preset range; The number of booths simultaneously placing the same product is within the sixth preset range; as well as The products corresponding to a single user are within the preset product set.

9. The method according to claim 1, wherein: The number of the plurality of booths is K, and determining the recommended content to be displayed at each booth of the target page based on the target recommendation scheme includes: For the kth booth among the plurality of booths, k is an integer less than or equal to K: Obtaining the target product corresponding to the k-th booth from the target recommendation scheme, and determining a plurality of candidate contents corresponding to the target product. According to the user characteristics corresponding to the target user, the recommended content to be displayed at the k-th booth is determined from the multiple candidate contents.

10. The method according to claim 9, wherein: The step of determining the recommended content to be displayed at the k-th booth from the plurality of candidate contents according to the user characteristics corresponding to the target user includes: Obtaining the degree of compatibility between the user features and the plurality of candidate contents respectively; and Based on the number M of contents supported for display at the k-th booth, first M candidate contents with the highest adaptability among the multiple candidate contents are used as recommended contents to be displayed at the k-th booth.

11. The method according to claim 10, wherein: The user feature includes sub-features of multiple dimensions, and obtaining the degree of compatibility between the user feature and the multiple candidate contents includes: Determining at least one target dimension related to the product placed corresponding to the kth booth from among the multiple dimensions; Acquire the sub-feature of the at least one target dimension from the user feature as a reference feature; and Determine the degree of fit between the reference features and the plurality of candidate contents respectively.

12. The method according to claim 9, wherein: The method is applied to a recommendation system, and obtaining a target user's access request to a target page includes: receiving the access request from a client; The method further includes: sending the recommended content to be displayed at each booth in the target page to the client, so that the client renders the recommended content to a page area corresponding to the booth in the target page.

13. The method according to claim 12, wherein: The recommended content includes at least one of text content, picture content, audio content, video content, product link or jump link.

14. A recommendation system comprising: at least one storage medium storing at least one instruction set for performing data processing related to content recommendation; as well as At least one processor is communicatively connected to the at least one storage medium, wherein when the recommendation system is running, the at least one processor reads the at least one instruction set and implements the method according to any one of claims 1 to 13 according to the instructions of the at least one instruction set.

Images