Page recommendation method, device, equipment and storage medium

Abstract

The present disclosure relates to a page recommendation method, apparatus, device and storage medium, and relates to the field of computer technology. The embodiments of the present disclosure at least solve the technical problem in the related art that the optimal page cannot be recommended to the user. The method includes: obtaining characteristic information of the user account and page characteristics of each preset page in the application; the characteristic information includes historical behavior data of the user account in the application; inputting the characteristic information and the page characteristics of each preset page into a pre-trained prediction model to obtain the predicted access time of each preset page; in the training label of the prediction model, the access time of each sample is determined according to the access time of the user account in each preset page of the application during the historical startup cycle, and the access time of the user account in each preset page of the application during the target historical startup cycle after the historical startup cycle; based on the predicted access time, determining the page to be recommended from each preset page.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to a page recommendation method, apparatus, device, and storage medium. Background Art

[0002] At present, mobile Internet applications (Application, APP) usually contain multiple different content areas, which are mostly presented by switching tabs to enrich the user account experience, expand the business boundaries, and thus improve the retention stickiness of the user account. When a user account opens the APP, the page presented for the first time directly determines the initial user account experience and is crucial for the continuous access of the user account. In the related art, in the process of determining the page presented for the first time by the APP, the page presented for the first time is usually recommended to the user account based on the historical behavior data of the user account in each page when logging in or accessing the APP in the historical stage.

[0003] However, in the above page recommendation method, since the historical behavior data of each page in the historical period is used as the basis, only the user account stay time or stickiness in the short term is considered, it is difficult to predict the retention stickiness of the user account in the long term, which may result in the page recommended to the user account being suboptimal. Summary of the Invention

[0004] The present disclosure provides a page recommendation method, apparatus, device, and storage medium to at least solve the technical problem in related technologies of being unable to recommend optimal pages to user accounts. The technical solutions of the present disclosure are as follows:

[0005] According to a first aspect of an embodiment of the present disclosure, a page recommendation method is provided, comprising: obtaining feature information of a user account and page features of each preset page in an application; the feature information comprises historical behavior data of the user account in the application; inputting the feature information and the page features of each preset page into a pre-trained prediction model to predict the access time, thereby obtaining the predicted access time of each preset page; the training label of the prediction model comprises a plurality of sample access times; each sample access time is determined based on the access time of the user account in each preset page of the application during a historical startup cycle, and the access time of the user account in each preset page of the application during a target historical startup cycle after the historical startup cycle; based on the predicted access time, determining a page to be recommended from each preset page.

[0006] Optionally, the above method also includes: for each historical startup cycle in multiple historical startup cycles, obtaining the access time of each preset page of the user account in each historical startup cycle, and obtaining the access time corresponding to each historical startup cycle; based on the access time corresponding to each historical startup cycle, determining the independent access time of each historical startup cycle; the independent access time is used to indicate the total access time of the user account in each preset page of the application in each historical startup cycle; based on the startup time sequence of multiple historical startup cycles, determining the target historical startup cycle after each historical startup cycle from multiple historical startup cycles, and obtaining the independent access time of the target historical startup cycle; according to the independent access time of each historical startup cycle and the independent access time of the target historical startup cycle, determining the sample access time corresponding to each historical startup cycle, and obtaining multiple sample access times; using multiple sample access times as training labels, and using the characteristic information of the sample user accounts in multiple historical startup cycles and the page features of each preset page as sample features, training the preset neural network model, and obtaining a prediction model.

[0007] Optionally, the above-mentioned determination of the independent access duration of each historical startup cycle based on the access duration corresponding to each historical startup cycle includes: determining the valid access duration and the invalid access duration from the access duration corresponding to each historical startup cycle; the valid access duration is the access duration greater than the first threshold; the invalid access duration is the access duration less than the second threshold; the first threshold is greater than or equal to the second threshold; based on the valid access duration, determining the effective comprehensive duration of each historical startup cycle; the effective comprehensive duration is used to characterize the sum of the effective access durations; based on the invalid access duration, determining the invalid comprehensive duration of each historical startup cycle; the invalid comprehensive duration is used to characterize the sum of the invalid access durations; based on the valid comprehensive duration and the invalid comprehensive duration of each historical startup cycle, determining the independent access duration of each historical startup cycle.

[0008] Optionally, the above-mentioned method determines the sample access duration corresponding to each historical startup cycle based on the independent access duration of each historical startup cycle and the independent access duration of the target historical startup cycle, and obtains multiple sample access durations, including: determining the sample access duration corresponding to each historical startup cycle based on the independent access duration of each historical startup cycle, the independent access duration of the target historical startup cycle and the access parameters, and obtaining multiple sample access durations; the access parameters include an attenuation factor, and the attenuation factor is used to indicate that the influence of the independent access duration of the target historical startup cycle on the sample access duration corresponding to each historical startup cycle gradually decreases.

[0009] Optionally, the above-mentioned sample access duration corresponding to each historical startup cycle is determined based on the independent access duration of each historical startup cycle, the independent access duration of the target historical startup cycle and the access parameters, and multiple sample access durations are obtained, including: based on the independent access duration of each historical startup cycle, the independent access duration of the target historical startup cycle and the attenuation factor, the comprehensive access duration of each historical startup cycle is determined as the sample access duration corresponding to each historical startup cycle.

[0010] Optionally, the above-mentioned access parameters also include a third threshold; based on the independent access duration of each historical startup cycle, the independent access duration of the target historical startup cycle and the access parameters, the sample access duration corresponding to each historical startup cycle is determined to obtain multiple sample access durations, including: based on the independent access duration of each historical startup cycle, the independent access duration of the target historical startup cycle and the attenuation factor, the comprehensive access duration of each historical startup cycle is determined; based on the comprehensive access duration of each historical startup cycle and the third threshold, multiple sample startup cycles are determined from multiple historical startup cycles, and the comprehensive access duration of the multiple sample startup cycles is determined as multiple sample access durations; the comprehensive access duration of each sample startup cycle is greater than the third threshold.

[0011] Optionally, the above-mentioned determination of the comprehensive access duration of each historical startup cycle based on the independent access duration of each historical startup cycle, the independent access duration of the target historical startup cycle and the attenuation factor includes: determining the cumulative impact duration of the target historical startup cycle according to the independent access duration of the target historical startup cycle and the attenuation factor; the cumulative impact duration includes the independent access duration of each target historical startup cycle and the sum of the impact durations of the independent access duration of each historical startup cycle affected by the attenuation factor; determining the independent access duration of each historical startup cycle and the sum of the cumulative impact duration of the target historical startup cycle as the comprehensive access duration of each historical startup cycle.

[0012] Optionally, the above-mentioned determination of the page to be recommended from each preset page based on the predicted access time includes: determining the page to be recommended from each preset page based on the predicted access time of each preset page and the page click information of each preset page; the page click information is determined based on a preset beta distribution function, and the page click information is used to represent the probability of the page being clicked by the user account.

[0013] According to a second aspect of an embodiment of the present disclosure, a page recommendation device is provided, including an acquisition unit, a prediction unit and a determination unit; the acquisition unit is used to acquire feature information of a user account and page features of each preset page in an application; the feature information includes historical behavior data of the user account in the application; the prediction unit is used to input the feature information and page features of each preset page into a pre-trained prediction model to predict the access time, so as to obtain the predicted access time of each preset page; the training label of the prediction model includes multiple sample access times; each sample access time is determined according to the access time of the user account in each preset page of the application during a historical startup cycle, and the access time of the user account in each preset page of the application during a target historical startup cycle after the historical startup cycle; the determination unit is used to determine the page to be recommended from each preset page based on the predicted access time.

[0014] Optionally, the above-mentioned device also includes a training unit; the acquisition unit is further used to obtain the access time of each preset page of the user account in the application in each historical startup cycle for each historical startup cycle in multiple historical startup cycles, and obtain the access time corresponding to each historical startup cycle; the determination unit is further used to determine the independent access time of each historical startup cycle based on the access time corresponding to each historical startup cycle; the independent access time is used to indicate the total access time of the user account in each preset page of the application in each historical startup cycle; the determination unit is further used to determine the target historical startup cycle after each historical startup cycle from multiple historical startup cycles based on the startup time sequence of multiple historical startup cycles, and obtain the independent access time of the target historical startup cycle; the determination unit is further used to determine the sample access time corresponding to each historical startup cycle based on the independent access time of each historical startup cycle and the independent access time of the target historical startup cycle, and obtain multiple sample access times; the training unit is used to use multiple sample access times as training labels, and use the characteristic information of the sample user accounts in multiple historical startup cycles and the page features of each preset page as sample features to train the preset neural network model to obtain a prediction model.

[0015] Optionally, the above-mentioned determination unit is specifically used to: determine the effective access time and the invalid access time from the access time corresponding to each historical startup cycle; the effective access time is the access time greater than the first threshold; the invalid access time is the access time less than the second threshold; the first threshold is greater than or equal to the second threshold; based on the effective access time, determine the effective comprehensive time of each historical startup cycle; the effective comprehensive time is used to characterize the sum of the effective access times; based on the invalid access time, determine the invalid comprehensive time of each historical startup cycle; the invalid comprehensive time is used to characterize the sum of the invalid access times; based on the effective comprehensive time and the invalid comprehensive time of each historical startup cycle, determine the independent access time of each historical startup cycle.

[0016] Optionally, the above-mentioned determination unit is specifically used to: determine the sample access duration corresponding to each historical startup cycle based on the independent access duration of each historical startup cycle, the independent access duration of the target historical startup cycle and the access parameters, and obtain multiple sample access durations; the access parameters include an attenuation factor, which is used to indicate that the influence of the independent access duration of the target historical startup cycle on the sample access duration corresponding to each historical startup cycle gradually decreases.

[0017] Optionally, the above-mentioned determination unit is specifically used to: determine the comprehensive access duration of each historical startup cycle based on the independent access duration of each historical startup cycle, the independent access duration of the target historical startup cycle and the attenuation factor, and determine the comprehensive access duration of each historical startup cycle as the sample access duration corresponding to each historical startup cycle.

[0018] Optionally, the above-mentioned access parameters also include a third threshold; the above-mentioned determination unit is specifically used to: determine the comprehensive access duration of each historical startup cycle based on the independent access duration of each historical startup cycle, the independent access duration of the target historical startup cycle and the attenuation factor; determine multiple sample startup cycles from multiple historical startup cycles based on the comprehensive access duration of each historical startup cycle and the third threshold, and determine the comprehensive access duration of the multiple sample startup cycles as multiple sample access durations; the comprehensive access duration of each sample startup cycle is greater than the third threshold.

[0019] Optionally, the above-mentioned determination unit is specifically used to: determine the cumulative impact duration of the target historical startup cycle based on the independent access duration of the target historical startup cycle and the attenuation factor; the cumulative impact duration includes the independent access duration of each target historical startup cycle and the sum of the impact durations of the independent access duration of each historical startup cycle affected by the attenuation factor; determine the independent access duration of each historical startup cycle, and the sum of the cumulative impact duration of the target historical startup cycle, as the comprehensive access duration of each historical startup cycle.

[0020] Optionally, the above-mentioned determination unit is specifically used to: determine the page to be recommended from each preset page based on the predicted access time of each preset page and the page click information of each preset page; the page click information is determined based on a preset beta distribution function, and the page click information is used to represent the probability of the page being clicked by the user account.

[0021] According to a third aspect of an embodiment of the present disclosure, an electronic device is provided, comprising: a processor and a memory for storing instructions executable by the processor; wherein the processor is configured to execute instructions to implement a page recommendation method as provided in the first aspect and any possible design thereof.

[0022] According to a fourth aspect of an embodiment of the present disclosure, a computer-readable storage medium is provided. When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the page recommendation method provided in the first aspect and any possible design thereof.

[0023] According to a fifth aspect of an embodiment of the present disclosure, a computer program product is provided, which includes computer instructions. When the computer instructions are run on an electronic device, the electronic device executes the page recommendation method provided in the first aspect and any possible design thereof.

[0024] The technical solution provided by this disclosure provides at least the following beneficial effects: During the training process, the access duration of each sample in the training label is determined based on the access duration of the user account on each preset application page during the historical launch cycle, as well as the access duration of the user account on each preset application page during the target historical launch cycle after the historical launch cycle. As a result, the predicted access duration, based on the above prediction model, includes the influence of access durations within different launch cycles, and can characterize the long-term retention and stickiness of user accounts. Furthermore, the optimal pages to be recommended can be determined based on global and long-term profit considerations.

[0025] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The accompanying drawings herein are incorporated into and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the description are used to explain the principles of the present disclosure, and do not constitute an improper limitation of the present disclosure.

[0027] Figure 1 is a structural diagram of a multi-page recommendation system according to an exemplary embodiment;

[0028] Figure 2This is one of the flowcharts of a page recommendation method according to an exemplary embodiment;

[0029] Figure 3 This is a second flow chart of a page recommendation method according to an exemplary embodiment;

[0030] Figure 4 This is a third flow chart of a page recommendation method according to an exemplary embodiment;

[0031] Figure 5 This is a fourth flowchart of a page recommendation method according to an exemplary embodiment;

[0032] Figure 6 This is a fifth flowchart of a page recommendation method according to an exemplary embodiment;

[0033] Figure 7 This is a sixth flowchart of a page recommendation method according to an exemplary embodiment;

[0034] Figure 8 is a structural diagram of a page recommendation device according to an exemplary embodiment;

[0035] Figure 9 The figure is a schematic structural diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0036] In order to enable ordinary persons in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings.

[0037] It should be noted that the terms "first," "second," and the like in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or precedence. It should be understood that the numbers used in this manner are interchangeable where appropriate so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of apparatus and methods consistent with certain aspects of the present disclosure as detailed in the appended claims.

[0038] In addition, in the description of the embodiments of the present disclosure, unless otherwise specified, " / " means or. For example, A / B can mean A or B. "And / or" in this document is simply a description of the association relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present disclosure, "plurality" means two or more than two.

[0039] A page recommendation method provided by an embodiment of the present disclosure may be applicable to a page recommendation system. Figure 1 A structural diagram of the page recommendation system is shown in FIG. Figure 1 As shown, a page recommendation system 10 is used to address the technical problem in related arts of being unable to recommend optimal pages to user accounts. Page recommendation system 10 includes a page recommendation device 11 and an electronic device 12. The page recommendation device 11 and the electronic device 12 are connected. The connection between the page recommendation device 11 and the electronic device 12 can be wired or wireless, and this is not limited in the present embodiment.

[0040] The electronic device 12 is used to receive a login request from a user account requesting to log in to an application, and obtain user account information, historical behavior data, context information of pages in the application, etc. of the user account, and send the obtained information to the page recommendation device.

[0041] The page recommendation device 11 can be used to interact with the electronic device 12 for data. For example, the page recommendation device 11 can obtain user account information of the user account, historical behavior data of the user account, and context information of the page in the application from the electronic device 12 .

[0042] The page recommendation device 11 can also execute the page recommendation method in the embodiment of the present disclosure, for example, input the acquired feature information and the page features of each preset page into a pre-trained prediction model to predict the predicted access time of each preset page, and determine the page to be recommended to the user account based on the predicted access time.

[0043] It should be noted that the page recommendation device 11 and the electronic device 12 can be independent devices or integrated into the same device, and the present invention does not specifically limit this. The electronic device 12 can be a terminal with an application installed, or a server or server cluster used for developing, testing, and hosting application programs.

[0044] When the page recommendation device 11 and the electronic device 12 are integrated into the same device, the communication between the page recommendation device 11 and the electronic device 12 is performed as between the internal modules of the device. In this case, the communication process between the two is the same as the communication process between the page recommendation device 11 and the electronic device 12 when they are independent of each other.

[0045] In the following embodiments provided by the present invention, the present invention is described by taking an example in which the page recommendation device 11 and the electronic device 12 are independently configured.

[0046] In practical applications, the page recommendation method provided by the embodiment of the present invention can be applied to a page recommendation device, or to an electronic device or other similar device that includes a page recommendation device. Below, in conjunction with the accompanying drawings, the page recommendation method provided by the embodiment of the present invention is described by taking the application of the page recommendation method to the page recommendation device as an example.

[0047] like Figure 2 As shown, the page recommendation method provided by the embodiment of the present disclosure includes the following S201-S203.

[0048] S201 : The page recommendation device obtains characteristic information of a user account and page characteristics of each preset page in an application.

[0049] The feature information includes the historical behavior data of the user account in the application. The page feature is used to identify the context information of each preset page in the application.

[0050] As a possible implementation method, the page recommendation device can obtain historical behavior data of the user account in the application and / or account information of the user account from the electronic device, and convert the obtained historical behavior data and / or user account information into feature information.

[0051] It should be noted that the characteristic information also includes the account information of the user account.

[0052] For example, historical user account behavior data within an application includes the user account's visit duration, video playback duration, likes, follows, and friend additions on each preset page within the application. User account information includes information such as the user's age, gender, region, and group. Preset page context information includes the page's type, scene, display time period, and network information.

[0053] S202 : The page recommendation device inputs the feature information and the page features of each preset page into a pre-trained prediction model to predict the access duration, and obtains the predicted access duration of each preset page.

[0054] The training labels for the prediction model include multiple sample access durations. Each sample access duration is determined based on the user account's access duration on each preset page of the application during the historical launch cycle, as well as the user account's access duration on each preset page of the application during a target historical launch cycle following the historical launch cycle. The predicted access duration includes the cumulative access duration of the user account in the application, assuming the preset page is the first page displayed.

[0055] As a possible implementation manner, the page recommendation device inputs the feature information and the page features of each preset page into a prediction model, and determines the prediction result output by the prediction model as the predicted access duration of each preset page.

[0056] As an example, the prediction model may be a double Deep Q Network (double DQN) model.

[0057] S203: The page recommendation device determines a page to be recommended from each preset page based on the predicted access duration.

[0058] As a possible implementation manner, the page recommendation device determines, based on the predicted access time, a preset page with the largest predicted access time as the page to be recommended.

[0059] For other implementations of this step, please refer to the subsequent description of the embodiments of the present disclosure, which will not be repeated here.

[0060] The technical solution provided by this disclosure provides at least the following beneficial effects: During the training process, the access duration of each sample in the training label is determined based on the access duration of the user account on each preset application page during the historical launch cycle, as well as the access duration of the user account on each preset application page during the target historical launch cycle after the historical launch cycle. As a result, the predicted access duration, based on the above prediction model, includes the influence of access durations within different launch cycles, and can characterize the long-term retention and stickiness of user accounts. Furthermore, the optimal pages to be recommended can be determined based on global and long-term profit considerations.

[0061] In one design, in order to be able to train a prediction model based on multiple sample access durations, such as Figure 3 As shown, the page recommendation method provided by the embodiment of the present disclosure further includes the following S301-S306.

[0062] S301. For each of a plurality of historical startup cycles, the page recommendation device obtains the access duration of each preset page in the application by the user account in each historical startup cycle to obtain the access duration corresponding to each historical startup cycle.

[0063] The historical startup cycle is the period of time during which a user account launches an application and stays in the application in the historical stage.

[0064] As a possible implementation method, the page recommendation device determines multiple historical startup cycles and obtains the access time of the user account in each preset page of the application in each historical startup cycle.

[0065] It should be noted that the access duration can be the length of time the user account stays on the preset page.

[0066] Exemplarily, the application includes 5 pages. For any historical startup cycle, the page recommendation device obtains the length of time the user account stays on the 5 pages as the visit length of the user account on the 5 pages.

[0067] S302: The page recommendation device determines the independent access duration of each historical startup cycle based on the access duration corresponding to each historical startup cycle.

[0068] The independent access duration is used to indicate the total access duration of the user account in each preset page of the application in each historical startup cycle.

[0069] As a possible implementation, for any historical startup cycle, the page recommendation device calculates the sum of the access duration of the user account on each preset page during the historical startup cycle as the independent access duration of the historical startup cycle.

[0070] The specific implementation of this step can also refer to the subsequent description of the embodiment of the present disclosure, and will not be repeated here.

[0071] S303: The page recommendation apparatus determines a target historical startup cycle following each historical startup cycle from the multiple historical startup cycles based on the startup time sequence of the multiple historical startup cycles.

[0072] As a possible implementation manner, for any historical startup cycle, the page recommendation device determines a target historical startup cycle that follows the historical startup cycle based on the time sequence of the historical startup cycle among multiple historical startup cycles.

[0073] Exemplarily, the number of the plurality of historical startup cycles is 10. For the fifth historical startup cycle, the target historical startup cycles corresponding to the fifth historical startup cycle are the sixth historical startup cycle to the tenth historical startup cycle.

[0074] S304: The page recommendation device obtains the independent access duration of the target historical startup cycle.

[0075] As a possible implementation manner, the page recommendation apparatus obtains the independent access duration of each target historical startup cycle from the independent access duration of each historical startup cycle in the multiple historical startup cycles in S302 according to the identifier of the target historical startup cycle.

[0076] S305 , the page recommendation device determines the sample access duration corresponding to each historical startup cycle according to the independent access duration of each historical startup cycle and the independent access duration of the target historical startup cycle, and obtains multiple sample access durations.

[0077] As one possible implementation, for any historical boot cycle, the page recommendation device calculates the sum of the independent access duration of the historical boot cycle and the independent access duration of the target historical boot cycle as the sample access duration corresponding to the historical boot cycle. Furthermore, the page recommendation device calculates the sample access durations corresponding to all historical boot cycles as multiple sample access durations.

[0078] The specific implementation of this step can also refer to the subsequent description of the embodiment of the present disclosure, and will not be repeated here.

[0079] S306. The page recommendation device uses multiple sample access durations as training labels, and uses feature information of sample user accounts in multiple historical startup cycles and page features of each preset page as sample features to train a preset neural network model to obtain a prediction model.

[0080] As a possible implementation method, the page recommendation device uses multiple sample access times as sample labels in the training samples, uses the historical behavior data of the user account corresponding to each sample access time, and / or the user account information of the user account, and the page features of each preset page as the sample features in the training samples, and performs supervised training on the preset neural network model to train a prediction model.

[0081] In this step, the implementation method of training the initial prediction model can refer to the existing technology and will not be described in detail here.

[0082] The technical solution provided by this disclosure provides at least the following beneficial effects: It can determine, based on historical startup cycles, the independent access duration corresponding to each historical startup cycle and the independent access duration of a target startup cycle. Based on the determined independent access durations, multiple sample access durations are determined to represent the cumulative access duration of a user account within an application. Furthermore, the multiple sample access durations are used as sample labels to train the aforementioned prediction model, ensuring that the prediction model can predict the predicted access duration of a user account within an application.

[0083] In one design, each preset page in the application is of different types, such as a product display page and a video playback page. For different types of preset pages, the access duration of a user account on each preset page has different comparability, which may result in inaccurate independent access duration. Therefore, in order to ensure the accuracy of the independent access duration, Figure 4 As shown, the above S302 provided in the embodiment of the present disclosure may include the following S401-S404.

[0084] S401. The page recommendation apparatus determines a valid access duration and an invalid access duration from the access duration corresponding to each historical startup cycle.

[0085] The valid access duration is an access duration greater than a first threshold, and the invalid access duration is an access duration less than a second threshold. The first threshold is greater than or equal to the second threshold.

[0086] As a possible implementation method, for the preset pages included in the application in any historical startup cycle, the page recommendation device determines that the preset page with an access time greater than a first threshold is a valid page based on the access time of each preset page, and determines the access time of the valid page as the effective access time.

[0087] At the same time, the page recommendation device further determines that a preset page with an access duration less than a second threshold is an invalid page, and determines the access duration of the invalid page as the invalid duration.

[0088] It should be noted that the first threshold and the second threshold can be set in advance in the page recommendation device by the operation and maintenance personnel of the page recommendation system.

[0089] Exemplarily, the first threshold may be 7 seconds, and the second threshold may also be 7 seconds.

[0090] S402: The page recommendation device determines the effective comprehensive duration of each historical startup cycle based on the effective access duration.

[0091] The effective comprehensive duration is used to represent the sum of the effective access durations.

[0092] As a possible implementation manner, for any historical startup cycle, the page recommendation device determines the sum of all valid access durations corresponding to the historical startup cycle as the effective comprehensive duration of the historical startup cycle.

[0093] As another possible implementation, for any historical startup cycle, the page recommendation device weights all effective access durations corresponding to the historical startup cycle to obtain the effective comprehensive duration of the historical startup cycle.

[0094] In this step, the weights corresponding to the effective durations of different preset pages are also different. For example, the weight corresponding to the product display page is greater than the weight corresponding to the video playback page.

[0095] It is understandable that since the page feature of the video playback page is to play videos, the user account stays in this type of preset page for a longer time. Therefore, in order to unify the comparability of the access time of different types of preset pages, different weights are set for different types of preset pages, which can achieve the comparability of the access time of different types of preset pages.

[0096] S403: The page recommendation device determines the invalid comprehensive duration of each historical startup cycle based on the determined invalid access duration.

[0097] The invalid comprehensive duration is used to represent the sum of invalid access durations.

[0098] As a possible implementation manner, for any historical startup cycle, the page recommendation device determines the sum of all invalid access durations corresponding to the historical startup cycle as the invalid comprehensive duration of the historical startup cycle.

[0099] As another possible implementation, for any historical startup cycle, the page recommendation device weights all invalid access durations corresponding to the historical startup cycle to obtain the invalid comprehensive duration of the historical startup cycle.

[0100] In this step, the weights corresponding to the inactive time of different pages are also different. For example, the weight corresponding to the product display page is greater than the weight corresponding to the video playback page.

[0101] It is understandable that since the page feature of the video playback page is to play videos, the user account stays in this type of preset page for a longer time. Therefore, in order to unify the comparability of the access time of different types of pages, different weights are set for different types of preset pages, which can achieve the comparability of the access time of different types of preset pages.

[0102] It should be noted that, in actual application, S402 may be executed first and then S403, or S403 may be executed first and then S402, or S402 and S403 may be executed simultaneously.

[0103] S404: The page recommendation device determines the independent access duration of each historical startup cycle based on the effective integrated duration and the invalid integrated duration of each historical startup cycle.

[0104] As a possible implementation, for any historical startup cycle, the page recommendation device may sum the effective integrated duration and the invalid integrated duration corresponding to the historical startup cycle, and use the calculation result as the independent access duration of the historical startup cycle.

[0105] As another possible implementation, the page recommendation device may further weight the effective integrated duration and the invalid integrated duration of the historical startup cycle to obtain the independent access duration of the historical startup cycle.

[0106] It should be noted that the weight of the effective comprehensive duration is greater than the weight of the invalid comprehensive duration.

[0107] The technical solution provided by this disclosure provides at least the following beneficial effects: By assigning different weights to different types of preset page valid access durations and different weights to different types of invalid access durations, the access durations corresponding to different preset page types are made comparable, resulting in more accurate independent access durations. Furthermore, by weighting the valid and invalid combined durations and assigning different weights to them, the calculated independent access durations can be further ensured to be more accurate.

[0108] In one design, in order to obtain multiple sample access durations, such as Figure 5 As shown, S305 provided in the embodiment of the present disclosure may specifically include the following S501.

[0109] S501. The page recommendation apparatus determines a sample access duration corresponding to each historical startup cycle based on the independent access duration of each historical startup cycle, the independent access duration of a target historical startup cycle, and access parameters, and obtains multiple sample access durations.

[0110] The access parameter includes a decay factor, which is used to indicate that the influence of the independent access duration of the target historical startup cycle on the sample access duration corresponding to each historical startup cycle gradually decreases.

[0111] As a possible implementation method, the page recommendation device can determine the comprehensive access duration of each historical startup cycle based on the independent access duration of each historical startup cycle, the independent access duration of the target historical startup cycle and the attenuation factor, and determine the comprehensive access duration of each historical startup cycle as the sample access duration corresponding to each historical startup cycle.

[0112] In some embodiments, in order to calculate the comprehensive access time of each historical startup cycle, in the page recommendation method provided by the embodiment of the present disclosure, the above-mentioned step of determining the comprehensive access time may specifically include the following S1-S2.

[0113] S1. The page recommendation device determines the cumulative impact duration of the target historical startup period according to the independent access duration of the target historical startup period and the attenuation factor.

[0114] The cumulative impact duration includes the independent access duration of each target historical startup cycle and the sum of the impact durations of the independent access duration of each historical startup cycle affected by the attenuation factor.

[0115] As a possible implementation method, the cumulative impact duration of the target historical launch cycle satisfies the following formula:

[0116]

[0117] Among them, T 累计影响 is the cumulative access time of the target startup cycle, a is the decay factor, j is the order of the target historical startup cycle, T 独立j is the independent access duration of the jth historical startup cycle.

[0118] S2. The page recommendation device determines the sum of the independent access duration of each historical startup cycle and the cumulative impact duration of the target historical startup cycle as the comprehensive access duration of each historical startup cycle.

[0119] It can be understood that, combined with the above S1-S2, the comprehensive access duration of each historical startup cycle satisfies the following formula:

[0120]

[0121] Among them, T 综合i is the comprehensive access duration of the i-th historical startup cycle, T 独立i is the independent access duration of the i-th historical startup cycle, a is the decay factor, j is the order of the target historical startup cycle, T 独立j is the independent access duration of the jth historical startup cycle.

[0122] The technical solution provided by the present disclosure brings at least the following beneficial effects: in the process of determining multiple sample access durations, since each sample access duration includes the influence of the independent access duration of the target historical startup cycle on the sample access duration, it can be ensured that the determined sample access duration can truly represent the cumulative access duration of the user account within the application. At the same time, using the comprehensive access duration of each historical startup cycle as the sample access duration can further ensure the accuracy of the determined sample access duration. Furthermore, using the above formula, a method for determining the comprehensive access duration of the historical startup cycle is provided.

[0123] In one design, since there may be a certain amount of noise data in the determined comprehensive access duration, in order to further ensure the accuracy of the sample data, such as Figure 6 As shown, S501 provided in the embodiment of the present disclosure may further include the following S601-S603.

[0124] S601. The page recommendation apparatus determines a comprehensive access duration of each historical startup cycle based on the independent access duration of each historical startup cycle, the independent access duration of a target historical startup cycle, and a decay factor.

[0125] It should be noted that the specific implementation method of this step can refer to the formulas in S1-S2 provided in this embodiment of the present disclosure, and will not be repeated here.

[0126] S602: The page recommendation apparatus determines a plurality of sample startup cycles from a plurality of historical startup cycles based on the comprehensive access duration of each historical startup cycle and a third threshold.

[0127] The third threshold is located in the access parameter, and the comprehensive access duration of each sample startup cycle is greater than the third threshold.

[0128] As a possible implementation manner, the page recommendation device determines, based on the third threshold, from the determined comprehensive access durations of multiple historical startup cycles, a historical startup cycle with a comprehensive access duration greater than the third threshold as a sample startup cycle.

[0129] It should be noted that the third threshold can be set in advance by the operation and maintenance personnel in the page recommendation device, and the third threshold can also be determined by the page recommendation device by sorting the multiple comprehensive access times determined from large to small, and based on the sorting results and the preset number of samples, determining the minimum value of the sampled number of comprehensive access times as the third threshold.

[0130] Exemplarily, a comprehensive access duration greater than the third threshold may be a positive sample, and a comprehensive access duration less than the third threshold may be a negative sample.

[0131] S603: The page recommendation device determines the comprehensive access durations of the multiple sample startup cycles as multiple sample access durations.

[0132] The technical solution provided by the present disclosure brings at least the following beneficial effects: determining positive samples from the determined comprehensive access duration, and using the determined positive samples as multiple sample access durations, which can remove noise data in the sample data and thus ensure the accuracy of the trained prediction model.

[0133] In one design, in the page recommendation method provided by the embodiment of the present disclosure, in order to make the determined recommended pages more consistent with the behavior of the user account, such as Figure 7 As shown, the above S203 provided in the embodiment of the present disclosure may specifically include S701-S702.

[0134] S701: The page recommendation device obtains page click information of each preset page.

[0135] The page click information is determined based on a preset Beta distribution function and is used to represent the probability of a page being clicked by a user account.

[0136] As a possible implementation manner, the page recommendation device may obtain page click information of each preset page from its memory.

[0137] It should be noted that the page recommendation device maintains a beta distribution function for each preset page. The beta distribution function for each preset page includes a first parameter and a second parameter. The first parameter is used to indicate the number of times the preset page was hit, and the second parameter is used to indicate the number of times the preset page was not hit.

[0138] In some embodiments, when the page recommendation device determines that any of the preset pages has been clicked, followed, or favorited by a user account, or has been viewed for a period exceeding a preset threshold, the first parameter is incremented by 1. Conversely, if any of the preset pages has not been clicked, followed, or favorited by a user account, or has been viewed for a period less than the preset threshold, the page recommendation device increments the second parameter by 1.

[0139] Furthermore, the page recommendation device determines page click information of each preset page according to the Beta distribution function including the updated first parameter and the second parameter, and stores the page click information of each preset page.

[0140] S702: The page recommendation device determines a page to be recommended from each preset page based on the predicted access duration of each preset page and page click information of each preset page.

[0141] As a possible implementation method, the page recommendation device weights the predicted access time and the click probability in the page click information, and determines the preset page corresponding to the maximum value of the weighted result as the page to be recommended based on the size of the weighted result.

[0142] The technical solution provided by the present disclosure brings at least the following beneficial effects: in the process of determining the pages to be recommended, the page click information of each preset page can be maintained based on the Beta distribution function, and the impact of the page click information on the recommendation results can be taken into account, which can avoid the convergence of the recommendation results to a certain extent, and at the same time ensure that the recommendation results have a certain degree of exploratory nature.

[0143] Figure 8 FIG. 1 is a structural diagram of a page recommendation device according to an exemplary embodiment. Figure 8 As shown, the page recommendation device 80 provided by the embodiment of the present disclosure includes an acquisition unit 801 , a prediction unit 802 and a determination unit 803 .

[0144] The acquisition unit 801 is used to acquire characteristic information of the user account and page characteristics of each preset page in the application. The characteristic information includes historical behavior data of the user account in the application.

[0145] Prediction unit 802 is configured to input feature information and page features of each preset page into a pre-trained prediction model to predict access duration, thereby obtaining a predicted access duration for each preset page. The training labels of the prediction model include multiple sample access durations. Each sample access duration is determined based on the user account's access duration on each preset page of the application during a historical startup cycle, as well as the user account's access duration on each preset page of the application during a target historical startup cycle following the historical startup cycle.

[0146] The determining unit 803 is configured to determine a page to be recommended from among the preset pages based on the predicted access duration.

[0147] Optional, such as Figure 8 As shown, the device 80 provided by the embodiment of the present disclosure further includes a training unit 804.

[0148] The acquisition unit 801 is further configured to acquire, for each of the multiple historical startup cycles, the access duration of each preset page in the application by the user account in each historical startup cycle, and obtain the access duration corresponding to each historical startup cycle.

[0149] The determining unit 803 is further configured to determine an independent access duration for each historical startup cycle based on the access duration corresponding to each historical startup cycle. The independent access duration indicates the total access duration of the user account in each preset page of the application program in each historical startup cycle.

[0150] The determining unit 803 is further configured to determine a target historical boot cycle following each historical boot cycle from the multiple historical boot cycles based on the boot time sequence of the multiple historical boot cycles, and obtain an independent access duration of the target historical boot cycle.

[0151] The determining unit 803 is further configured to determine a sample access duration corresponding to each historical boot cycle according to the independent access duration of each historical boot cycle and the independent access duration of the target historical boot cycle, thereby obtaining a plurality of sample access durations.

[0152] The training unit 804 is used to use multiple sample access durations as training labels, feature information of sample user accounts in multiple historical startup cycles and page features of each preset page as sample features, train a preset neural network model, and obtain a prediction model.

[0153] Optional, such as Figure 8 As shown, the determining unit 803 provided in the embodiment of the present disclosure is specifically configured to:

[0154] Determine the valid access duration and invalid access duration from the access duration corresponding to each historical startup cycle. The valid access duration is the access duration greater than a first threshold. The invalid access duration is the access duration less than a second threshold. The first threshold is greater than or equal to the second threshold.

[0155] Based on the effective access duration, the effective comprehensive duration of each historical startup cycle is determined. The effective comprehensive duration is used to represent the sum of the effective access durations.

[0156] Based on the invalid access duration, the invalid comprehensive duration of each historical startup cycle is determined. The invalid comprehensive duration is used to represent the sum of the invalid access durations.

[0157] Based on the effective integrated duration and the invalid integrated duration of each historical startup cycle, the independent access duration of each historical startup cycle is determined.

[0158] Optional, such as Figure 8 As shown, the determining unit 803 provided in the embodiment of the present disclosure is specifically configured to:

[0159] Based on the independent access duration of each historical boot cycle, the independent access duration of the target historical boot cycle, and access parameters, a sample access duration corresponding to each historical boot cycle is determined to obtain multiple sample access durations. The access parameters include a decay factor, which is used to indicate that the influence of the independent access duration of the target historical boot cycle on the sample access duration corresponding to each historical boot cycle gradually decreases.

[0160] Optional, such as Figure 8 As shown, the determining unit 803 provided in the embodiment of the present disclosure is specifically configured to:

[0161] Based on the independent access duration of each historical startup cycle, the independent access duration of the target historical startup cycle and the attenuation factor, the comprehensive access duration of each historical startup cycle is determined, and the comprehensive access duration of each historical startup cycle is determined as the sample access duration corresponding to each historical startup cycle.

[0162] Optional, such as Figure 8 As shown, the access parameter provided by the embodiment of the present disclosure also includes a third threshold. The determination unit 803 is specifically configured to:

[0163] The comprehensive access duration of each historical boot cycle is determined based on the independent access duration of each historical boot cycle, the independent access duration of the target historical boot cycle, and the decay factor.

[0164] Based on the comprehensive access duration of each historical boot cycle and a third threshold, multiple sample boot cycles are determined from the multiple historical boot cycles, and the comprehensive access duration of the multiple sample boot cycles is determined as multiple sample access durations. The comprehensive access duration of each sample boot cycle is greater than the third threshold.

[0165] Optional, such as Figure 8 As shown, the determining unit 803 provided in the embodiment of the present disclosure is specifically configured to:

[0166] The cumulative impact duration of the target's historical startup cycles is determined based on the independent access duration of the target's historical startup cycles and the attenuation factor. The cumulative impact duration includes the sum of the independent access duration of each target's historical startup cycle and the impact duration of each historical startup cycle affected by the attenuation factor.

[0167] Determine the independent access duration of each historical startup cycle and the sum of the cumulative impact duration of the target historical startup cycle to obtain the comprehensive access duration of each historical startup cycle.

[0168] Optional, such as Figure 8 As shown, the determining unit 803 provided in the embodiment of the present disclosure is specifically configured to:

[0169] Based on the predicted visit duration of each preset page and the page click information of each preset page, the page to be recommended is determined from each preset page. The page click information is determined based on a preset beta distribution function and is used to represent the probability of the page being clicked by the user account.

[0170] Regarding the apparatus in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0171] Figure 9 : is a schematic diagram of the structure of an electronic device provided by the present disclosure. The electronic device can be the first node or the second node mentioned above. Figure 9 The electronic device 90 may include at least one processor 901 and a memory 903 for storing processor-executable instructions. The processor 901 is configured to execute the instructions in the memory 903 to implement the page recommendation method in the above embodiment.

[0172] In addition, the electronic device 90 may further include a communication bus 902 and at least one communication interface 904 .

[0173] The processor 901 may be a central processing unit (CPU), a microprocessor unit, an ASIC, or one or more integrated circuits for controlling the execution of the program of the disclosed solution.

[0174] The communication bus 902 may include a pathway for transmitting information between the aforementioned components.

[0175] The communication interface 904 uses any transceiver or other device for communicating with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area network (WLAN), etc.

[0176] The memory 903 may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program codes in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory may be independent and connected to the processing unit via a bus. The memory may also be integrated with the processing unit.

[0177] The memory 903 is used to store instructions for executing the solution of the present disclosure, and the execution is controlled by the processor 901. The processor 901 is used to execute the instructions stored in the memory 903, thereby realizing the functions of the method of the present disclosure.

[0178] As an example, combining Figure 8 The functions implemented by the acquisition unit 801, the prediction unit 802, the determination unit 803 and the training unit 804 in the page recommendation device 80 are similar to those implemented by the Figure 9 The function of processor 901 in is the same.

[0179] In a specific implementation, as an embodiment, the processor 901 may include one or more CPUs, such as Figure 9 CPU0 and CPU1 in.

[0180] In a specific implementation, as an embodiment, the electronic device 90 may include multiple processors, such as Figure 9901 and processor 907 in FIG. Each of these processors may be a single-CPU processor or a multi-CPU processor. A processor herein may refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0181] In a specific implementation, as an embodiment, the electronic device 90 may further include an output device 905 and an input device 906. The output device 905 communicates with the processor 901 and can display information in a variety of ways. For example, the output device 905 can be a liquid crystal display (LCD), a light emitting diode (LED) display device, a cathode ray tube (CRT) display device, or a projector. The input device 906 communicates with the processor 901 and can accept user account input in a variety of ways. For example, the input device 906 can be a mouse, a keyboard, a touch screen device, or a sensor device.

[0182] Those skilled in the art will understand that Figure 9 The structure shown in the figure does not constitute a limitation on the electronic device 90, and the electronic device 90 may include more or fewer components than shown in the figure, or combine certain components, or adopt a different component arrangement.

[0183] In addition, the present disclosure also provides a computer-readable storage medium. When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device can execute the page recommendation method provided in the above embodiment.

[0184] In addition, the present disclosure also provides a computer program product, including computer instructions. When the computer instructions are executed on an electronic device, the electronic device executes the page recommendation method provided in the above embodiment.

[0185] Other embodiments of the present disclosure will readily occur to those skilled in the art after considering the specification and practicing the invention disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, with the true scope and spirit of the present disclosure being indicated by the claims.

Claims

1. A page recommendation method, characterized in that: include: Obtain user account feature information and page features of each preset page in the application; The characteristic information includes historical behavior data of the user account in the application; Input the feature information and the page features of each of the preset pages into a pre-trained prediction model to predict the access duration, thereby obtaining the predicted access duration of each of the preset pages; the training labels of the prediction model include a plurality of sample access durations; the sample access durations are determined based on the independent access durations of each historical startup cycle and the independent access durations of a target historical startup cycle; the target historical startup cycle is a historical startup cycle after each historical startup cycle determined from the plurality of historical startup cycles based on the startup time sequence of the plurality of historical startup cycles; the independent access durations are used to indicate the total access duration of the user account in each of the preset pages of the application in each historical startup cycle; The independent access duration is determined based on the access duration of each preset page in the application by the user account in each historical startup cycle; the sample features of the prediction model include feature information of sample user accounts in multiple historical startup cycles and the page features of each preset page; Based on the predicted access duration, a page to be recommended is determined from each of the preset pages.

2. The page recommendation method according to claim 1, characterized in that: The method further comprises: For each of the plurality of historical startup cycles, obtaining a duration of access to each of the preset pages in the application by the user account in each historical startup cycle, to obtain a duration of access corresponding to each historical startup cycle; Determine, based on the access duration corresponding to each historical startup cycle, an independent access duration for each historical startup cycle; the independent access duration is used to indicate a total access duration of the user account on each of the preset pages of the application during each historical startup cycle; Based on the startup time sequence of the multiple historical startup cycles, determining the target historical startup cycle after each historical startup cycle from the multiple historical startup cycles, and obtaining the independent access duration of the target historical startup cycle; Determine the sample access duration corresponding to each historical startup cycle according to the independent access duration of each historical startup cycle and the independent access duration of the target historical startup cycle, to obtain the multiple sample access durations; The multiple sample access durations are used as training labels, the feature information of the sample user accounts in the multiple historical startup cycles and the page features of each of the preset pages are used as sample features, and a preset neural network model is trained to obtain the prediction model.

3. The page recommendation method according to claim 2, characterized in that: The determining, based on the access duration corresponding to each historical startup period, the independent access duration of each historical startup period includes: Determining a valid access duration and an invalid access duration from the access duration corresponding to each historical startup cycle; the valid access duration is the access duration greater than a first threshold; the invalid access duration is the access duration less than a second threshold; the first threshold is greater than or equal to the second threshold; Determine the effective comprehensive duration of each historical startup cycle based on the effective access duration; the effective comprehensive duration is used to represent the sum of the effective access durations; Determine the invalid comprehensive duration of each historical startup cycle based on the invalid access duration; the invalid comprehensive duration is used to represent the sum of the invalid access durations; The independent access duration of each historical startup cycle is determined based on the effective integrated duration and the invalid integrated duration of each historical startup cycle.

4. The page recommendation method according to claim 2, characterized in that: Determining the sample access duration corresponding to each historical startup cycle according to the independent access duration of each historical startup cycle and the independent access duration of the target historical startup cycle to obtain the multiple sample access durations includes: Based on the independent access duration of each historical startup cycle, the independent access duration of the target historical startup cycle and the access parameters, the sample access duration corresponding to each historical startup cycle is determined to obtain the multiple sample access durations; the access parameters include an attenuation factor, and the attenuation factor is used to indicate that the influence of the independent access duration of the target historical startup cycle on the sample access duration corresponding to each historical startup cycle gradually decreases.

5. The page recommendation method according to claim 4, characterized in that: The determining, based on the independent access duration of each historical startup cycle, the independent access duration of the target historical startup cycle, and the access parameter, the sample access duration corresponding to each historical startup cycle to obtain the multiple sample access durations includes: Based on the independent access duration of each historical startup cycle, the independent access duration of the target historical startup cycle and the attenuation factor, the comprehensive access duration of each historical startup cycle is determined, and the comprehensive access duration of each historical startup cycle is determined as the sample access duration corresponding to each historical startup cycle.

6. The page recommendation method according to claim 4, characterized in that: The access parameter further includes a third threshold; determining the sample access duration corresponding to each historical startup cycle based on the independent access duration of each historical startup cycle, the independent access duration of the target historical startup cycle, and the access parameter, to obtain the multiple sample access durations, including: Determine a comprehensive access duration of each historical boot cycle based on the independent access duration of each historical boot cycle, the independent access duration of the target historical boot cycle, and the decay factor; Based on the comprehensive access duration of each historical startup cycle and the third threshold, multiple sample startup cycles are determined from the multiple historical startup cycles, and the comprehensive access duration of the multiple sample startup cycles is determined as the multiple sample access durations; the comprehensive access duration of each sample startup cycle is greater than the third threshold.

7. The page recommendation method according to claim 5 or 6, characterized in that: The determining, based on the independent access duration of each historical boot cycle, the independent access duration of the target historical boot cycle, and the attenuation factor, of the comprehensive access duration of each historical boot cycle includes: Determining a cumulative impact duration of the target historical startup period based on the independent access duration of the target historical startup period and the attenuation factor; the cumulative impact duration includes the sum of the independent access duration of each target historical startup period and the impact duration of the independent access duration of each historical startup period affected by the attenuation factor; Determine the sum of the independent access duration of each historical startup cycle and the accumulated impact duration of the target historical startup cycle as the comprehensive access duration of each historical startup cycle.

8. The page recommendation method according to any one of claims 1 to 6, characterized in that: The determining of the to-be-recommended page from the preset pages based on the predicted access duration includes: Based on the predicted access time of each of the preset pages and the page click information of each of the preset pages, the page to be recommended is determined from each of the preset pages; the page click information is determined based on a preset beta distribution function, and the page click information is used to represent the probability of the page being clicked by the user account.

9. A page recommendation device, characterized in that: include: Acquisition unit, prediction unit and determination unit; The acquisition unit is configured to acquire characteristic information of the user account and page characteristics of each preset page in the application; the characteristic information includes historical behavior data of the user account in the application; The prediction unit is used to input the feature information and the page features of each of the preset pages into a pre-trained prediction model to predict the access time, so as to obtain the predicted access time of each of the preset pages; the training labels of the prediction model include multiple sample access times; the sample access time is determined based on the independent access time of each historical startup cycle and the independent access time of a target historical startup cycle; the target historical startup cycle is a historical startup cycle after each historical startup cycle determined from the multiple historical startup cycles based on the startup time sequence of multiple historical startup cycles; the independent access time is used to indicate the total access time of the user account in each of the preset pages of the application in each historical startup cycle; The independent access duration is determined based on the access duration of each preset page in the application by the user account in each historical startup cycle; the sample features of the prediction model include feature information of sample user accounts in multiple historical startup cycles and the page features of each preset page; The determining unit is configured to determine a page to be recommended from the preset pages based on the predicted access duration.

10. The page recommendation device according to claim 9, characterized in that: The device further comprises a training unit; The acquisition unit is further configured to acquire, for each historical startup cycle in a plurality of historical startup cycles, a duration of access of each preset page of the application by the user account in each historical startup cycle, to obtain the access duration corresponding to each historical startup cycle; The determining unit is further configured to determine an independent access duration for each historical startup period based on the access duration corresponding to each historical startup period; the independent access duration is configured to indicate a total access duration of the user account on each of the preset pages of the application program during each historical startup period; The determining unit is further configured to determine, based on a startup time sequence of the multiple historical startup cycles, the target historical startup cycle following each of the multiple historical startup cycles, and obtain the independent access duration of the target historical startup cycle; The determining unit is further configured to determine the sample access duration corresponding to each historical startup cycle according to the independent access duration of each historical startup cycle and the independent access duration of the target historical startup cycle, to obtain the multiple sample access durations; The training unit is used to use the multiple sample access durations as training labels, the feature information of the sample user accounts in the multiple historical startup cycles and the page features of each of the preset pages as sample features, train a preset neural network model, and obtain the prediction model.

11. The page recommendation device according to claim 10, characterized in that: The determining unit is specifically configured to: Determining a valid access duration and an invalid access duration from the access duration corresponding to each historical startup cycle; the valid access duration is the access duration greater than a first threshold; the invalid access duration is the access duration less than a second threshold; the first threshold is greater than or equal to the second threshold; Determine the effective comprehensive duration of each historical startup cycle based on the effective access duration; the effective comprehensive duration is used to represent the sum of the effective access durations; Determine the invalid comprehensive duration of each historical startup cycle based on the invalid access duration; the invalid comprehensive duration is used to represent the sum of the invalid access durations; The independent access duration of each historical startup cycle is determined based on the effective integrated duration and the invalid integrated duration of each historical startup cycle.

12. The page recommendation device according to claim 10, characterized in that: The determining unit is specifically configured to: Based on the independent access duration of each historical startup cycle, the independent access duration of the target historical startup cycle and the access parameters, the sample access duration corresponding to each historical startup cycle is determined to obtain the multiple sample access durations; the access parameters include an attenuation factor, and the attenuation factor is used to indicate that the influence of the independent access duration of the target historical startup cycle on the sample access duration corresponding to each historical startup cycle gradually decreases.

13. The page recommendation device according to claim 12, characterized in that: The determining unit is specifically configured to: Based on the independent access duration of each historical startup cycle, the independent access duration of the target historical startup cycle and the attenuation factor, the comprehensive access duration of each historical startup cycle is determined, and the comprehensive access duration of each historical startup cycle is determined as the sample access duration corresponding to each historical startup cycle.

14. The page recommendation device according to claim 12, characterized in that: The access parameter further includes a third threshold; the determining unit is specifically configured to: Determine a comprehensive access duration of each historical boot cycle based on the independent access duration of each historical boot cycle, the independent access duration of the target historical boot cycle, and the decay factor; Based on the comprehensive access duration of each historical startup cycle and the third threshold, multiple sample startup cycles are determined from the multiple historical startup cycles, and the comprehensive access duration of the multiple sample startup cycles is determined as the multiple sample access durations; the comprehensive access duration of each sample startup cycle is greater than the third threshold.

15. The page recommendation device according to claim 13 or 14, characterized in that: The determining unit is specifically configured to: Determining a cumulative impact duration of the target historical startup period based on the independent access duration of the target historical startup period and the attenuation factor; the cumulative impact duration includes the sum of the independent access duration of each target historical startup period and the impact duration of the independent access duration of each historical startup period affected by the attenuation factor; Determine the sum of the independent access duration of each historical startup cycle and the accumulated impact duration of the target historical startup cycle as the comprehensive access duration of each historical startup cycle.

16. The page recommendation device according to any one of claims 9 to 14, characterized in that: The determining unit is specifically configured to: Based on the predicted access time of each of the preset pages and the page click information of each of the preset pages, the page to be recommended is determined from each of the preset pages; the page click information is determined based on a preset beta distribution function, and the page click information is used to represent the probability of the page being clicked by the user account.

17. An electronic device, characterized in that: include: A processor and a memory for storing instructions executable by the processor; wherein the processor is configured to execute instructions to implement the page recommendation method according to any one of claims 1 to 8.

18. A computer-readable storage medium, characterized in that When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the page recommendation method according to any one of claims 1 to 8.

19. A computer program product comprising instructions, characterized in that The computer program product includes computer instructions, and when the computer instructions are executed on an electronic device, the electronic device executes the page recommendation method according to any one of claims 1 to 8.

Images