Content recommendation method and device based on large model, electronic equipment, readable storage medium and computer program product

Abstract

The invention provides a content recommendation method and device based on a large model, electronic equipment, a readable storage medium and a computer program product, and relates to the field of intelligent recommendation, in particular to the field of content recommendation based on the large model. According to the implementation scheme, based on browsing content selected by a user for the first time, an initial estimated value of a user behavior index is determined through an index estimation model; for each continuous browsing associated index of the at least one continuous browsing associated index, determining a first estimated value of the continuous browsing associated index based on the initial estimated value of the continuous browsing rate of the user and the initial estimated value of the continuous browsing associated index; based on the browsing content selected by the user for the first time, obtaining an initial weight value of the at least one continuous browsing association index by utilizing a weight learning model; and based on the first estimated value and the initial weight value of the at least one continuous browsing association index and the currently browsed content, determining recommended content of the currently browsed content.

Description

Technical Field

[0001] The present disclosure relates to the field of intelligent recommendation, and particularly to the field of content recommendation based on large models. Specifically, it relates to a content method, device, electronic device, computer-readable storage medium, and computer program product based on large models. Background Art

[0002] Artificial intelligence is a discipline that studies how to make computers simulate certain human thinking processes and intelligent behaviors (such as learning, reasoning, thinking, planning, etc.), including both hardware-level technologies and software-level technologies. Artificial intelligence hardware technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, and big data processing; artificial intelligence software technologies mainly include several major directions such as computer vision technology, speech recognition technology, natural language processing technology, and machine learning / deep learning, big data processing technology, and knowledge graph technology.

[0003] With the continuous development of computer technology, network information has also exploded. For the vast amount of Internet information, how to obtain the content that truly meets the user's interest points from the vast amount of information and recommend the obtained content to the user is a key factor in improving user satisfaction and retention rate; and the rapid development of artificial intelligence technology, especially large models and neural networks, provides a better technical foundation for providing matching recommended content for users.

[0004] The methods described in this section are not necessarily methods that have been previously envisioned or adopted. Unless otherwise specified, no method described in this section should be considered prior art solely because it is included in this section. Similarly, unless otherwise specified, the problems mentioned in this section should not be considered to have been recognized in any prior art. Summary of the Invention

[0005] The present disclosure provides a content recommendation method, device, electronic device, computer-readable storage medium, and computer program product.

[0006] According to one aspect of the present disclosure, there is provided a content recommendation method, including: based on the browsing content initially selected by a user, using a pre-trained metric estimation model to determine an initial pre-estimation value of user behavior metrics, where the user behavior metrics include a user continuation browsing rate and at least one continuation browsing associated metric, the user continuation browsing rate is the probability that the user continues to browse associated content having a set relationship with the current browsing content after finishing browsing the current browsing content, and where each of the at least one continuation browsing associated metric is a metric characterizing user behavior corresponding to the associated content; for each of the at least one continuation browsing associated metric, based on the initial pre-estimation value of the user continuation browsing rate and the initial pre-estimation value of this continuation browsing associated metric, determine a first pre-estimation value of this continuation browsing associated metric; based on the browsing content initially selected by the user, using a weight learning model to obtain an initial weight value of each of the at least one continuation browsing associated metric, where the weight learning model is trained using a large model; and based on the first pre-estimation value of the at least one continuation browsing associated metric, the initial weight value corresponding to the first pre-estimation value of the at least one continuation browsing associated metric, and the current browsing content, determine the recommended content corresponding to the current browsing content.

[0007] According to another aspect of the present disclosure, there is provided a training method for a metric estimation model for content recommendation, including: obtaining a sample user browsing content and a marked value of a sample user behavior metric corresponding to the sample user browsing content; inputting the sample user browsing content into the metric estimation model to obtain a sample pre-estimation value of the sample user behavior metric, where the metric estimation model is constructed based on a multi-task learning model; based on the marked value and the sample pre-estimation value, determine a loss value of the metric estimation model; and in response to determining that the loss value satisfies a preset loss value threshold, end the training of the metric estimation model; otherwise, based on the loss value, adjust the parameters of the metric estimation model; execute the steps of obtaining the sample user browsing content and the marked value of the sample user behavior metric corresponding to the sample user browsing content and subsequent steps until the loss value satisfies the preset loss value threshold.

[0008] According to another aspect of the present disclosure, there is provided a content recommendation device, including: an initial estimated value acquisition module, configured to determine an initial estimated value of a user behavior metric based on the browsing content initially selected by the user and using a pre-trained metric estimation model, where the user behavior metric includes a user continuation browsing rate and at least one continuation browsing associated metric, the user continuation browsing rate being the probability that the user continues to browse associated content having a set relationship with the current browsing content after ending the browsing of the current browsing content, and where each of the at least one continuation browsing associated metrics is a metric characterizing the user behavior corresponding to the associated content; a first estimated value acquisition module, configured to, for each of the at least one continuation browsing associated metrics, determine a first estimated value of the continuation browsing associated metric based on the initial estimated value of the user continuation browsing rate and the initial estimated value of the continuation browsing associated metric; a weight acquisition module, configured to obtain an initial weight value of each of the at least one continuation browsing associated metrics based on the browsing content initially selected by the user and using a weight learning model, where the weight learning model is trained using a large model; and a recommended content determination module, configured to determine recommended content corresponding to the current browsing content based on the first estimated values of the at least one continuation browsing associated metrics, the initial weight values corresponding to the first estimated values of the at least one continuation browsing associated metrics, and the current browsing content.

[0009] According to another aspect of the present disclosure, there is provided a training device for a metric estimation model for content recommendation, including: a sample acquisition module, configured to acquire sample user browsing content and marked values of sample user behavior metrics corresponding to the sample user browsing content; an estimated value acquisition module, configured to input the sample user browsing content into the metric estimation model to obtain sample estimated values of the sample user behavior metrics, where the metric estimation model is constructed based on a multi-task learning model; a loss value calculation module, configured to determine a loss value of the metric estimation model based on the marked values and the sample estimated values; and a parameter adjustment module, configured to end the training of the metric estimation model in response to determining that the loss value satisfies a preset loss value threshold; otherwise, adjust the parameters of the metric estimation model based on the loss value; and execute the steps of acquiring the sample user browsing content and the marked values of the sample user behavior metrics corresponding to the sample user browsing content and subsequent steps until the loss value satisfies the preset loss value threshold.

[0010] According to another aspect of the present disclosure, there is provided an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the above-mentioned content recommendation method and model training method.

[0011] According to another aspect of the present disclosure, there is provided a computer-readable storage medium having stored thereon a computer program, which when executed by a processor, causes the processor to implement the content recommendation method and model training method as described above.

[0012] According to another aspect of the present disclosure, there is provided a computer program product including a computer program, wherein the computer program implements the above-mentioned content recommendation method and model training method when executed by a processor.

[0013] According to one or more embodiments of the present disclosure, through the browsing content initially selected by the user, an initial predicted value of the user's continued browsing rate and the at least one continued browsing correlation index is determined, and the initial predicted value of the user's continued browsing rate is used to guide the determination of the first predicted value of the at least one continued browsing correlation index, so as to improve the accuracy of the first predicted value of the at least one continued browsing correlation index. At the same time, in combination with the user's current browsing content, recommended content corresponding to the current browsing content is determined, effectively improving the matching degree between the recommended content and the user.

[0014] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it used to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The drawings exemplarily illustrate embodiments and constitute a part of the specification, and are used together with the written description of the specification to explain the exemplary embodiments. The illustrated embodiments are for illustrative purposes only and do not limit the scope of the claims. In all the drawings, the same reference numerals refer to similar but not necessarily identical elements.

[0016] Figure 1 A schematic diagram of an exemplary system in which the various methods described herein can be implemented according to an embodiment of the present disclosure is shown;

[0017] Figure 2 A flowchart of a content recommendation method according to an embodiment of the present disclosure is shown;

[0018] Figure 3 A flowchart of a content recommendation method according to an embodiment of the present disclosure is shown;

[0019] Figure 4 The flowchart of the method for training an indicator prediction model according to an embodiment of the present disclosure is shown;

[0020] Figure 5 The block diagram of the content recommendation device according to an embodiment of the present disclosure is shown;

[0021] Figure 6 The block diagram of the indicator prediction model training device according to an embodiment of the present disclosure is shown;

[0022] Figure 7 The structural block diagram of an exemplary electronic device that can be used to implement the embodiments of the present disclosure is shown. Detailed implementation manners

[0023] The following makes an explanation of the exemplary embodiments of the present disclosure in conjunction with the accompanying drawings. Various details of the embodiments of the present disclosure are included to facilitate understanding, and they should be considered merely exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope of the present disclosure. Similarly, for the sake of clarity and conciseness, the description of well-known functions and structures is omitted below.

[0024] In the present disclosure, unless otherwise specified, the terms "first", "second", etc. are used to describe various elements and are not intended to limit the positional relationship, timing relationship, or importance relationship of these elements. Such terms are only used to distinguish one element from another. In some examples, the first element and the second element may refer to the same instance of the element, and in certain cases, based on the context description, they may also refer to different instances.

[0025] In the description of various examples in the present disclosure, the terms used are only for the purpose of describing specific examples and are not intended to be limiting. Unless the context clearly indicates otherwise, if the number of elements is not specifically limited, the element may be one or more. In addition, the term "and / or" used in the present disclosure covers any one of the listed items and all possible combinations.

[0026] The embodiments of the present disclosure will be described in detail below in conjunction with the accompanying drawings.

[0027] Figure 1 The schematic diagram of an exemplary system 100 in which various methods and devices described herein can be implemented according to an embodiment of the present disclosure is shown. Refer to Figure 1, the system 100 includes one or more client devices 101, 102, 103, 104, 105, and 106, a server 120, and one or more communication networks 110 that couple the one or more client devices to the server 120. The client devices 101, 102, 103, 104, 105, and 106 may be configured to execute one or more applications.

[0028] In an embodiment of the present disclosure, the server 120 may run one or more services or software applications that enable the execution of content recommendation methods and model training methods.

[0029] In certain embodiments, the server 120 may also provide other services or software applications, which may include non-virtual environments and virtual environments. In certain embodiments, these services may be provided as web-based services or cloud services, for example, provided to users of the client devices 101, 102, 103, 104, 105, and / or 106 under a software as a service (SaaS) model.

[0030] In Figure 1 the configuration shown, the server 120 may include one or more components that implement the functions performed by the server 120. These components may include software components, hardware components, or combinations thereof that may be executed by one or more processors. Users operating the client devices 101, 102, 103, 104, 105, and / or 106 may in turn utilize one or more client applications to interact with the server 120 to utilize the services provided by these components. It should be understood that various different system configurations are possible, which may be different from the system 100. Therefore, Figure 1 is an example of a system for implementing the various methods described herein and is not intended to be limiting.

[0031] Users may use the client devices 101, 102, 103, 104, 105, and / or 106 to implement content recommendations. The client devices may provide an interface that enables a user of the client device to interact with the client device. The client device may also output information to the user via the interface. Although Figure 1 only six client devices are depicted, those skilled in the art will be able to understand that the present disclosure may support any number of client devices.

[0032] Client devices 101, 102, 103, 104, 105, and / or 106 may include various types of computing devices, such as portable handheld devices, general-purpose computers (such as personal computers and laptop computers), workstation computers, wearable devices, smart screen devices, self-service terminal devices, service robots, gaming systems, thin clients, various messaging devices, sensors, or other sensing devices, etc. These computing devices may run various types and versions of software applications and operating systems, such as MICROSOFT Windows, APPLE iOS, UNIX-like operating systems, Linux or Linux-like operating systems (such as GOOGLE Chrome OS); or include various mobile operating systems, such as MICROSOFT WindowsMobile OS, iOS, Windows Phone, Android. Portable handheld devices may include cellular phones, smartphones, tablets, personal digital assistants (PDAs), etc. Wearable devices may include head-mounted displays (such as smart glasses) and other devices. Gaming systems may include various handheld gaming devices, Internet-enabled gaming devices, etc. The client devices are capable of executing various different applications, such as various Internet-related applications, communication applications (such as email applications), short message service (SMS) applications, and may use various communication protocols.

[0033] Network 110 can be any type of network known to those skilled in the art, and it can support data communication using any one of a variety of available protocols (including but not limited to TCP / IP, SNA, IPX, etc.). By way of example only, one or more networks 110 can be a local area network (LAN), an Ethernet-based network, token ring, wide area network (WAN), the Internet, virtual network, virtual private network (VPN), intranet, extranet, blockchain network, public switched telephone network (PSTN), infrared network, wireless network (such as Bluetooth, WIFI), and / or any combination of these and / or other networks.

[0034] Server 120 may include one or more general-purpose computers, dedicated server computers (such as PC (personal computer) servers, UNIX servers, midrange servers), blade servers, mainframes, server clusters, or any other suitable arrangement and / or combination. Server 120 may include one or more virtual machines running a virtual operating system, or other computing architectures involving virtualization (such as one or more flexible pools of logical storage devices that can be virtualized to maintain virtual storage devices for the server). In various embodiments, server 120 may run one or more services or software applications that provide the functions described below.

[0035] The computing unit in server 120 can run one or more operating systems including any of the above operating systems and any commercially available server operating systems. Server 120 can also run any one of a variety of additional server applications and / or middleware applications, including HTTP servers, FTP servers, CGI servers, JAVA servers, database servers, etc.

[0036] In some embodiments, server 120 can include one or more applications to analyze and merge data feeds and / or event updates received from users of client devices 101, 102, 103, 104, 105, and / or 106. Server 120 can also include one or more applications to display data feeds and / or real-time events via one or more display devices of client devices 101, 102, 103, 104, 105, and / or 106.

[0037] In some embodiments, server 120 can be a server of a distributed system or a server incorporating a blockchain. Server 120 can also be a cloud server or an intelligent cloud computing server or intelligent cloud host with artificial intelligence technology. A cloud server is a host product in the cloud computing service system to address the defects of high management difficulty and weak business scalability existing in traditional physical hosts and virtual private server (VPS) services.

[0038] System 100 can also include one or more databases 130. In certain embodiments, these databases can be used to store data and other information. For example, one or more of databases 130 can be used to store information such as audio files and video files. Databases 130 can reside in various locations. For example, the databases used by server 120 can be local to server 120 or can be remote from server 120 and can communicate with server 120 via a network-based or dedicated connection. Databases 130 can be of different types. In certain embodiments, the databases used by server 120 can be, for example, relational databases. One or more of these databases can store, update, and retrieve data to and from the databases in response to commands.

[0039] In certain embodiments, one or more of databases 130 can also be used by applications to store application data. The databases used by applications can be different types of databases, such as key-value repositories, object repositories, or conventional repositories supported by a file system.

[0040] Figure 1The system 100 can be configured and operated in various ways to enable the application of the various methods and apparatuses described in this disclosure.

[0041] Figure 2 is a flowchart showing a content recommendation method 200 according to an embodiment of the present disclosure, as Figure 2 shown, the content recommendation method may include: step S202, based on the browsing content initially selected by the user, using a pre-trained metric estimation model to determine an initial pre-estimation value of the user behavior metric, where the user behavior metric includes a user continuation browsing rate and at least one continuation browsing correlation metric, the user continuation browsing rate is the probability that the user continues to browse associated content having a set relationship with the current browsing content after finishing browsing the current browsing content, and where each of the at least one continuation browsing correlation metric is a metric characterizing the user behavior corresponding to the associated content; step S204, for each of the at least one continuation browsing correlation metric, based on the initial pre-estimation value of the user continuation browsing rate and the initial pre-estimation value of this continuation browsing correlation metric, determine a first pre-estimation value of this continuation browsing correlation metric; step S206, based on the browsing content initially selected by the user, using a weight learning model to obtain an initial weight value of each of the at least one continuation browsing correlation metric, where the weight learning model is trained using a large model; and step S208, based on the first pre-estimation value of the at least one continuation browsing correlation metric, the initial weight value corresponding to the first pre-estimation value of the at least one continuation browsing correlation metric, and the current browsing content, determine the recommended content corresponding to the current browsing content.

[0042] In this embodiment, through the browsing content initially selected by the user, the initial pre-estimation values of the user continuation browsing rate and the at least one continuation browsing correlation metric are determined, and the initial pre-estimation value of the user continuation browsing rate is used to guide the determination of the first pre-estimation value of the at least one continuation browsing correlation metric, so as to improve the accuracy of the first pre-estimation value of the at least one continuation browsing correlation metric. At the same time, in combination with the user's current browsing content, the recommended content corresponding to the current browsing content is determined, effectively improving the matching degree between the recommended content and the user.

[0043] In some embodiments, the continuous browsing correlation metric may be a continuous browsing step length, a continuous browsing duration, or other metrics that can reflect the user's continuous browsing behavior; the continuous browsing step length is the number of associated contents that the user continues to browse, which have a set relationship with the current browsing content, after finishing the current browsing content; the continuous browsing duration is the total duration that the user continues to browse the associated contents that have a set relationship with the current browsing content after finishing the current browsing content; the continuous browsing correlation metric can intuitively reflect the attractiveness of the recommended content to the user, thereby indirectly reflecting the matching degree between the recommended content and the user.

[0044] In some embodiments, the at least one continuous browsing correlation metric can also be used to inversely adjust the estimation of the continuous browsing rate; for example, when the at least one continuous browsing correlation metric is the user's stay duration, if the user stays for a long time, it indicates that the user is more interested in the current browsing content and is more likely to continue browsing the content related to the current browsing content, and correspondingly, the user's continuous browsing rate is higher; if the user leaves quickly (short stay duration), to a certain extent, it reflects that the user has a lower interest in the current browsing content, and correspondingly, the user's continuous browsing rate is smaller.

[0045] In some embodiments, the set relationship includes: topic association, knowledge structure association, scenario requirement association, user interest association, etc.; for example, when the user is watching a "fitness teaching video", the associated contents may be about "fitness diet plan", "training videos of the same type of author", and "introduction of home fitness equipment", and the acquisition of the association points of the above associated contents can be realized through the user interest tags pre-set in the content tags, or the collaborative filtering algorithm, or other algorithms that can obtain the association points.

[0046] In some embodiments, the weight learning model is based on a large model and is pre-trained with a large amount of sample data. In a specific application scenario, domain adaptation fine-tuning is adopted, and the model parameters are optimized by using the user behavior samples of the current platform to ensure that the weight distribution is aligned with the business objectives (such as content conversion rate, user retention rate); at the same time, it supports adjusting the weights through a time decay factor. For example, higher weights are assigned to the recently browsed contents.

[0047] In some embodiments, for new users or new contents, generalization initial weights are generated based on content-based collaborative filtering or other algorithms with the same function to alleviate the data sparsity problem during cold start.

[0048] In some embodiments, determining the first estimated value of the continuation browsing associated indicator based on the initial estimated value of the user's continued browsing rate and the initial estimated value of the continued browsing associated indicator includes: for each of the at least one continued browsing associated indicator, taking the initial estimated value of the user's continued browsing rate as the prior probability, and using the initial estimated value of the continued browsing associated indicator and the initial estimated value of the user's continued browsing rate to calculate the first estimated value of the continued browsing associated indicator.

[0049] In this embodiment, the user's continued browsing rate is used as a priori probability to determine the estimated value of the at least one continued browsing related indicator, which effectively constrains the estimation of the at least one continued browsing related indicator and avoids invalid estimation, thereby improving the accuracy of the recommendation method.

[0050] In some embodiments, for low-frequency content (such as niche short dramas), the content exposure is relatively small compared to other types of content. Using the user's continued browsing rate as a priori probability to constrain the estimated value range of the step size and duration can effectively avoid overfitting estimation.

[0051] In some embodiments, determining the recommended content corresponding to the currently browsed content based on the first estimated value of the at least one continued browsing association indicator, the initial weight value corresponding to the first estimated value of the at least one continued browsing association indicator, and the currently browsed content includes: obtaining the associated content of the currently browsed content; in response to determining that the associated content of the currently browsed content does not exist, obtaining a set number of hot content; and using the set number of hot content as recommended content corresponding to the currently browsed content.

[0052] In this embodiment, when it is determined that there is no related content of the currently browsed content, a set number of hot content is recommended to the user, providing the user with an opportunity to explore new content, avoiding the user leaving due to "no related content", and effectively improving the robustness of the content recommendation method.

[0053] In some embodiments, when a user watches a niche documentary, the platform first tries to recommend: other works by the same author, documentary series on the same theme, or award-winning films of the same type; when the platform does not have the related content, the method instead obtains the top 10 videos in terms of the number of views on the entire site that day; illustratively, it can also obtain the 10 most recently watched films by the user's friends or 5 personalized recommendations generated based on the user's historical preferences; and mark the above recommended videos with source tags such as "Everyone is watching" and "Friends like it".

[0054] Figure 3 3 is a flow chart showing a content recommendation method 300 according to an embodiment of the present disclosure. Figure 3As shown, the method includes: Step S302, in response to determining that there is at least one associated content of the current browsing content, determining the number of the at least one associated content; Step S304, in response to determining that the at least one associated content only includes one associated content, using the one associated content as the recommended content corresponding to the current browsing content; Step S306, in response to determining that the at least one associated content includes multiple associated contents, for each of the multiple associated contents, based on the first estimated value of the at least one continuous browsing association indicator, the initial weight value corresponding to the first estimated value of the at least one continuous browsing association indicator, and this associated content, calculating the recommended score of this associated content; and Step S308, based on the sorting order of the recommended scores of each of the multiple associated contents, determining the recommended content corresponding to the current browsing content.

[0055] In this embodiment, in response to there being one associated content of the current browsing content, using this associated content as the recommended content corresponding to the current browsing content; in response to there being multiple associated contents, sorting and recommending the associated contents according to their recommended scores, effectively improving the satisfaction of the user's continuous browsing.

[0056] In some embodiments, the calculating the recommended score of this associated content based on the first estimated value of the at least one continuous browsing association indicator, the initial weight value corresponding to the first estimated value of the at least one continuous browsing association indicator, and this associated content includes: for each of the multiple associated contents, based on the initial weight value corresponding to the first estimated value of the at least one continuous browsing association indicator and this associated content, using an adaptive evolutionary strategy to determine the real-time weight value of the at least one continuous browsing association indicator corresponding to this associated content; performing a weighted sum of the first estimated value and the real-time weight value of the at least one continuous browsing association indicator; and using the value of the weighted sum as the recommended score of this associated content.

[0057] In this embodiment, based on the initial weight value corresponding to the first estimated value of the at least one continuous browsing association indicator and this associated content, determining the real-time weight of the at least one continuous browsing association indicator through an adaptive evolutionary strategy, thereby adaptively capturing the user's real-time interest, effectively improving the reflection of the recommended score of the associated content on the user's real-time interest, and improving the adaptability of the recommended content.

[0058] In some embodiments, the adaptive evolution strategy includes calculating fitness using the user behavior data of a recent period (such as 1 hour) to ensure that the weights adapt to short-term changes; in this embodiment, the covariance matrix adaptation evolution strategy (CMA-ES) can be used to learn the correlation between the continuous browsing association metrics. For example, in browsing behavior analysis, if there is a negative correlation between "user click-through rate" and "user stay duration" (the user may leave early after quickly clicking), CMA-ES will automatically adjust the covariance matrix to avoid suboptimal solutions caused by independent optimization; the above description of the adaptive evolution strategy is only for illustrative purposes and is not restrictive, and other evolution strategies that can determine the real-time weight values of the at least one continuous browsing association metric can also be used.

[0059] In some embodiments, the method includes: in response to the browsing duration of the user for the current browsing content meeting a set condition, based on the current browsing content, using the metric prediction model to determine the real-time prediction value of the user's continuous browsing rate and the real-time prediction values of the at least one continuous browsing association metric; and based on the real-time prediction value of the user's continuous browsing rate and the real-time prediction values of the at least one continuous browsing association metric, determining a first prediction value of the at least one continuous browsing association metric.

[0060] In this embodiment, the browsing duration of the user for the current browsing content reflects to a certain extent the user's interest in the current browsing content. In response to the browsing duration of the user for the current browsing content meeting the set condition, the prediction value of the user behavior metric is re-determined according to the current browsing content, effectively improving the matching degree between the recommended content and the user.

[0061] In some embodiments, the set condition includes at least one of the following conditions: the browsing duration is greater than or equal to the total duration of the current browsing content; and the browsing duration is greater than a set duration threshold.

[0062] In this embodiment, by determining whether the user has completely browsed the current browsing content or whether the user has browsed for a set duration for the current browsing content with a longer duration, the user's interest in the current browsing content can be effectively determined.

[0063] In some embodiments, when the user's stay duration on the current browsing content reaches the average browsing duration of platform users, it is confirmed that the user is interested in the current browsing content, thereby triggering the real-time prediction of the user behavior metric. It should be noted that the setting of the duration threshold can be adaptively adjusted in combination with the average total browsing duration of the current platform content and the type of content.

[0064] In some embodiments, the method includes: in response to determining that the browsing content initially selected by the user is the content initially recommended by the platform, taking the average value of the user behavior metrics of the same type of content as the initial pre-estimated value of the user behavior metrics.

[0065] In this embodiment, for the content recommended for the first time, due to the lack of historical data, it is impossible to obtain the pre-estimated value of the user behavior metrics through learning. At this time, using the average value of the user behavior metrics of the same type of content as the pre-estimated value of the user behavior metrics of the user helps the initial exposure of the new content.

[0066] In some embodiments, the type of the content can be implemented based on the rules of the field to which the content belongs, content embedding clustering, collaborative filtering similarity, or other algorithms that can achieve type confirmation.

[0067] In some embodiments, the metric estimation model is constructed based on a multi-task learning model, including a backbone network, the user continuous browsing rate learning network, and at least one continuous browsing correlation metric learning network; wherein the backbone network, the user continuous browsing rate learning network, and at least one continuous browsing correlation metric learning network are respectively constructed using neural networks; the backbone network extracts the general features of the input data and uses the general features as the input of the user continuous browsing rate learning network and the at least one continuous browsing correlation metric learning network.

[0068] In some embodiments, the method further includes: for each of the at least one continuous browsing correlation metric, based on the initial pre-estimated value of the user continuous browsing rate and the initial pre-estimated value of this continuous browsing correlation metric, using an improved metric estimation model to determine the first estimate of this continuous browsing correlation metric; wherein, in the improved metric estimation model, the output of the user continuous browsing rate learning network and the output features of the backbone network are simultaneously used as the input of the at least one continuous browsing correlation metric learning network. In this embodiment, by using the improved metric estimation model to determine the first pre-estimated value of the at least one continuous browsing correlation metric, on the basis of sacrificing a certain amount of computing resources, the accuracy of the first pre-estimated value of the at least one continuous browsing correlation metric is effectively improved.

[0069] Figure 4 is a flowchart showing a training method 400 of a metric estimation model for content recommendation according to an embodiment of the present disclosure, as Figure 4As shown, the training method includes: Step S402, obtaining the browsing content of a sample user and the marked value of the sample user behavior index corresponding to the browsing content of the sample user; Step S404, inputting the browsing content of the sample user into the index prediction model to obtain a sample predicted value of the sample user behavior index, where the index prediction model is constructed based on a multi-task learning model; Step S406, determining the loss value of the index prediction model based on the marked value and the sample predicted value; and Step S408, in response to determining that the loss value meets a preset loss value threshold, ending the training of the index prediction model; otherwise, adjusting the parameters of the index prediction model based on the loss value; performing the steps of obtaining the browsing content of the sample user and the marked value of the sample user behavior index corresponding to the browsing content of the sample user and subsequent steps until the loss value meets the preset loss value threshold.

[0070] In this embodiment, the index prediction model is trained by using the browsing content of the sample user and the marked value of the sample user behavior index corresponding to the browsing content of the sample user, so as to improve the calculation accuracy of the index prediction model, and thus, based on the calculation result of the index prediction model, the predicted value of the user behavior index can be determined more efficiently and accurately.

[0071] Figure 5 is a block diagram showing a content recommendation device 500 according to an embodiment of the present disclosure, as Figure 5As shown, the content recommendation device includes: an initial pre-estimation value acquisition module 502, configured to determine an initial pre-estimation value of a user behavior metric based on the browsing content initially selected by the user and using a pre-trained metric estimation model, where the user behavior metric includes a user continued browsing rate and at least one continued browsing associated metric, the user continued browsing rate is the probability that the user continues to browse associated content having a set relationship with the current browsing content after finishing browsing the current browsing content, and where each of the at least one continued browsing associated metrics is a metric characterizing user behavior corresponding to the associated content; a first pre-estimation value acquisition module 504, configured to, for each of the at least one continued browsing associated metrics, determine a first pre-estimation value of the continued browsing associated metric based on the initial pre-estimation value of the user continued browsing rate and the initial pre-estimation value of the continued browsing associated metric; a weight acquisition module 506, configured to, based on the browsing content initially selected by the user and using a weight learning model, obtain an initial weight value of each of the at least one continued browsing associated metrics, where the weight learning model is trained using a large model; and a recommended content determination module 508, configured to determine recommended content corresponding to the current browsing content based on the first pre-estimation values of the at least one continued browsing associated metrics, the initial weight values corresponding to the first pre-estimation values of the at least one continued browsing associated metrics, and the current browsing content.

[0072] Figure 6 is a block diagram showing a training device 600 for a metric estimation model for content recommendation according to an embodiment of the present disclosure, as Figure 6 shown, the training device includes: a sample acquisition module 602, configured to acquire sample user browsing content and marked values of sample user behavior metrics corresponding to the sample user browsing content; a pre-estimation value acquisition module 604, configured to input the sample user browsing content into the metric estimation model to obtain sample pre-estimation values of the sample user behavior metrics, where the metric estimation model is constructed based on a multi-task learning model; a loss value calculation module 606, configured to determine a loss value of the metric estimation model based on the marked values and the sample pre-estimation values; and a parameter adjustment module 608, in response to determining that the loss value satisfies a preset loss value threshold, ending the training of the metric estimation model; otherwise, adjusting parameters of the metric estimation model based on the loss value; performing the steps of acquiring the sample user browsing content and the marked values of the sample user behavior metrics corresponding to the sample user browsing content and subsequent steps until the loss value satisfies the preset loss value threshold.

[0073] It should be noted that Figure 5 each module of the device 500 shown in Figure 2Each step in the described method 200 corresponds. Thus, the operations, features, and advantages described above for method 200 also apply to apparatus 500 and the modules and units included therein. Figure 6 Each module in the apparatus 600 shown in Figure 4 can correspond to each step in the described method 400. Thus, the operations, features, and advantages described above for method 400 also apply to apparatus 600 and the modules and units included therein. For the sake of brevity, some operations, features, and advantages are not described herein again.

[0074] According to an embodiment of the present disclosure, an electronic device, a readable storage medium, and a computer program product are also provided.

[0075] Referring to Figure 7 , a block diagram of an electronic device 700 that can be a server or a client of the present disclosure will now be described. It is an example of a hardware device that can be applied to various aspects of the present disclosure. The electronic device is intended to represent various forms of digital electronic computer devices, such as, a laptop computer, a desktop computer, a workbench, a personal digital assistant, a server, a blade server, a mainframe computer, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as, a personal digital processor, a cellular phone, a smart phone, a wearable device, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are only examples and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0076] As Figure 7 shown, the electronic device 700 includes a computing unit 701, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 702 or a computer program loaded from a storage unit 708 into a random access memory (RAM) 703. In the RAM 703, various programs and data required for the operation of the electronic device 700 can also be stored. The computing unit 701, the ROM 502, and the RAM 703 are connected to each other through a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0077] Multiple components in the electronic device 700 are connected to the I / O interface 705, including: an input unit 706, an output unit 707, a storage unit 708, and a communication unit 709. The input unit 706 can be any type of device capable of inputting information into the electronic device 700. The input unit 706 can receive input digital or character information, and generate key signal inputs related to the user settings and / or function controls of the electronic device, and can include, but is not limited to, a mouse, a keyboard, a touch screen, a trackpad, a trackball, a joystick, a microphone, and / or a remote control. The output unit 707 can be any type of device capable of presenting information, and can include, but is not limited to, a display, a speaker, a video / audio output terminal, a vibrator, and / or a printer. The storage unit 708 can include, but is not limited to, magnetic disks and optical discs. The communication unit 709 allows the electronic device 700 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks, and can include, but is not limited to, a modem, a network card, an infrared communication device, a wireless communication transceiver, and / or a chipset, such as a Bluetooth device, an 802.11 device, a WiFi device, a WiMax device, a cellular communication device, and / or the like.

[0078] The computing unit 701 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 701 executes the various methods and processes described above, such as method 200 and method 400. For example, in some embodiments, method 200 and method 400 can be implemented as computer software programs that are tangibly contained in a machine-readable medium, such as the storage unit 708. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 700 via the ROM 702 and / or the communication unit 709. When the computer program is loaded into the RAM 703 and executed by the computing unit 701, one or more steps of method 200 and method 400 described above can be executed. Alternatively, in other embodiments, the computing unit 701 can be configured to execute method 200 and method 400 in any other suitable manner (e.g., by means of firmware).

[0079] The various embodiments of the systems and techniques described above in this specification can be implemented in digital electronic circuitry, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on a chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be a special-purpose or general-purpose programmable processor that receives data and instructions from, and transmits data and instructions to, a storage system, at least one input device, and at least one output device.

[0080] The program code for implementing the methods of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing device, such that the program codes, when executed by the processor or controller, cause the functions / operations specified in the flowchart and / or block diagram to be implemented. The program code can be executed entirely on the machine, partly on the machine, as a stand-alone software package partly on the machine and partly on a remote machine, or entirely on the remote machine or server.

[0081] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, 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 foregoing.

[0082] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and a pointing device (e.g., a mouse or a trackball) through which the user can provide input to the computer. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0083] The systems and techniques described herein can be implemented in a computing system including backend components (e.g., as a data server), or a computing system including middleware components (e.g., an application server), or a computing system including frontend components (e.g., a user computer having a graphical user interface or a web browser through which the user can interact with an implementation of the systems and techniques described herein), or a computing system including any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected to each other by digital data communication in any form or medium (e.g., a communication network). Examples of communication networks include: local area network (LAN), wide area network (WAN), the Internet, and blockchain networks.

[0084] A computer system can include a client and a server. The client and the server are generally far from each other and usually interact through a communication network. The client - server relationship is created by computer programs running on the respective computers and having a client - server relationship with each other. The server can be a cloud server, can also be a server of a distributed system, or a server incorporating blockchain.

[0085] It should be understood that various forms of the processes shown above can be used, with steps reordered, added, or deleted. For example, the steps recited in this disclosure can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and no limitation is made herein.

[0086] Although embodiments or examples of the present disclosure have been described with reference to the accompanying drawings, it should be understood that the above methods, systems, and devices are merely exemplary embodiments or examples, and the scope of the present invention is not limited by these embodiments or examples, but is only defined by the authorized claims and their equivalent scope. Various elements in the embodiments or examples may be omitted or replaced by their equivalent elements. In addition, the steps may be executed in an order different from that described in the present disclosure. Further, the various elements in the embodiments or examples may be combined in various ways. Importantly, with the evolution of technology, many of the elements described herein may be replaced by equivalent elements that emerge after the present disclosure.

Claims

1. A content recommendation method based on a large model, comprising: Based on the browsing content initially selected by the user, using a pre-trained metric estimation model to determine an initial pre-estimation value of the user behavior metric, wherein the user behavior metric includes a user continuation browsing rate and at least one continuation browsing correlation metric, the user continuation browsing rate is the probability that the user continues to browse associated content having a set relationship with the current browsing content after finishing browsing the current browsing content, and wherein each continuation browsing correlation metric in the at least one continuation browsing correlation metric is a metric characterizing the user behavior corresponding to the associated content; For each continuation browsing correlation metric of the at least one continuation browsing correlation metric, Based on the initial pre-estimation value of the user continuation browsing rate and the initial pre-estimation value of this continuation browsing correlation metric, determine a first pre-estimation value of this continuation browsing correlation metric; Based on the browsing content initially selected by the user, using a weight learning model to obtain an initial weight value of each continuation browsing correlation metric of the at least one continuation browsing correlation metric, wherein the weight learning model is trained using a large model; and Based on the first pre-estimation value of the at least one continuation browsing correlation metric, the initial weight value corresponding to the first pre-estimation value of the at least one continuation browsing correlation metric, and the current browsing content, determine the recommended content corresponding to the current browsing content.

2. The content recommendation method according to claim 1, wherein, The determining the first pre-estimation value of the continuation browsing correlation metric based on the initial pre-estimation value of the user continuation browsing rate and the initial pre-estimation value of this continuation browsing correlation metric includes: For each continuation browsing correlation metric of the at least one continuation browsing correlation metric, Using the initial pre-estimation value of the user continuation browsing rate as the prior probability, and using the initial pre-estimation value of this continuation browsing correlation metric and the initial pre-estimation value of the user continuation browsing rate, calculate the first pre-estimation value of this continuation browsing correlation metric.

3. The content recommendation method according to claim 1, wherein The determining the recommended content corresponding to the current browsing content based on the first pre-estimation value of the at least one continuation browsing correlation metric, the initial weight value corresponding to the first pre-estimation value of the at least one continuation browsing correlation metric, and the current browsing content includes: Obtain the associated content of the current browsing content; In response to determining that there is no associated content of the current browsing content, obtain a set number of hot content; and Use the set number of hot content as the recommended content corresponding to the current browsing content.

4. The content recommendation method according to claim 3, wherein, The method includes: In response to determining that there is at least one of the associated content of the current browsing content, judge the number of the at least one of the associated content; In response to determining that the at least one of the associated content only includes one of the associated content, use the one of the associated content as the recommended content corresponding to the current browsing content; In response to determining that the at least one of the associated content includes a plurality of the associated content, for each of the plurality of the associated content, Calculate a recommendation score for the associated content based on a first estimated value of the at least one continuous browsing association metric, an initial weight value corresponding to the first estimated value of the at least one continuous browsing association metric, and the associated content; and Determine recommended content corresponding to the current browsing content based on the sorting order of the recommendation scores of each of the multiple pieces of associated content.

5. The content recommendation method according to claim 4, wherein, The calculating a recommendation score for the associated content based on a first estimated value of the at least one continuous browsing association metric, an initial weight value corresponding to the first estimated value of the at least one continuous browsing association metric, and the associated content includes: For each of the multiple pieces of associated content, Based on the initial weight value corresponding to the first estimated value of the at least one continuous browsing association metric and the associated content, use an adaptive evolutionary strategy to determine real-time weight values of the at least one continuous browsing association metric corresponding to the associated content; Perform a weighted sum of the first estimated value and the real-time weight values of the at least one continuous browsing association metric; and Take the value of the weighted sum as the recommendation score of the associated content.

6. The content recommendation method according to any one of claims 1-5, wherein, The method includes: In response to the browsing duration of the user for the current browsing content satisfying a set condition, Based on the current browsing content, use the metric estimation model to determine a real-time estimated value of the user's continuous browsing rate and real-time estimated values of the at least one continuous browsing association metric; and Based on the real-time estimated value of the user's continuous browsing rate and the real-time estimated values of the at least one continuous browsing association metric, determine a first estimated value of the at least one continuous browsing association metric.

7. The content recommendation method according to claim 6, wherein, The set condition includes at least one of the following conditions: The browsing duration is greater than or equal to the total duration of the current browsing content; and The browsing duration is greater than a set duration threshold.

8. The content recommendation method according to any one of claims 1-7, wherein The method includes: In response to determining that the browsing content initially selected by the user is content initially recommended by the platform, use the average value of the user behavior metrics of the same type of content as the initial estimated value of the user behavior metric.

9. A training method for an index prediction model for content recommendation, wherein, The metric estimation model is used to implement the content recommendation method described in any one of claims 1-8, wherein the training method includes: Obtain sample user browsing content and marked values of sample user behavior metrics corresponding to the sample user browsing content; Input the sample user browsing content into the metric estimation model to obtain sample estimated values of the sample user behavior metrics, where the metric estimation model is constructed based on a multi-task learning model; Based on the marked values and the sample estimated values, determine the loss value of the metric estimation model; and In response to determining that the loss value satisfies a preset loss value threshold, end the training of the metric estimation model; otherwise Based on the loss value, adjust the parameters of the metric estimation model; Execute the obtaining of sample user browsing content and marked values of sample user behavior metrics corresponding to the sample user browsing content and subsequent steps until the loss value satisfies the preset loss value threshold.

10. A content recommendation device based on a large model, including: An initial pre - estimate acquisition module, configured to determine an initial pre - estimate of user behavior metrics based on the browsing content initially selected by the user and using a pre - trained metric estimation model. Wherein, the user behavior metrics include a user continuation browsing rate and at least one continuation browsing associated metric. The user continuation browsing rate is the probability that the user continues to browse associated content having a set relationship with the current browsing content after finishing browsing the current browsing content. And wherein each of the at least one continuation browsing associated metrics is a metric characterizing user behavior corresponding to the associated content; A first pre - estimate acquisition module, configured to, for each of the at least one continuation browsing associated metrics, determine a first pre - estimate of the continuation browsing associated metric based on the initial pre - estimate of the user continuation browsing rate and the initial pre - estimate of the continuation browsing associated metric; A weight acquisition module, configured to, based on the browsing content initially selected by the user and using a weight learning model, obtain an initial weight value for each of the at least one continuation browsing associated metrics. Wherein, the weight learning model is trained using a large model; and A recommended content determination module, configured to determine recommended content corresponding to the current browsing content based on the first pre - estimates of the at least one continuation browsing associated metrics, the initial weight values corresponding to the first pre - estimates of the at least one continuation browsing associated metrics, and the current browsing content.

11. A training device for an index prediction model for content recommendation, wherein, The metric estimation model is used to implement the content recommendation device described in claim 9. Wherein, the training device includes: A sample acquisition module, configured to acquire sample user browsing content and marked values of sample user behavior metrics corresponding to the sample user browsing content; A pre - estimate acquisition module, configured to input the sample user browsing content into the metric estimation model to obtain sample pre - estimates of the sample user behavior metrics. Wherein, the metric estimation model is constructed based on a multi - task learning model; A loss value calculation module, configured to determine a loss value of the metric estimation model based on the marked values and the sample pre - estimates; and A parameter adjustment module, in response to determining that the loss value meets a preset loss value threshold, ends the training of the metric estimation model; otherwise Based on the loss value, perform parameter adjustment on the metric estimation model; Execute the steps of acquiring sample user browsing content and marked values of sample user behavior metrics corresponding to the sample user browsing content and subsequent steps until the loss value meets the preset loss value threshold.

12. An electronic device, comprising: At least one processor; And A memory communicatively connected to the at least one processor; Wherein The memory stores instructions executable by the at least one processor. The instructions are executed by the at least one processor so that the at least one processor can execute the method described in any one of claims 1 - 9.

13. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to execute the method described in any one of claims 1 - 9.

14. A computer program product comprising a computer program, wherein, The computer program, when executed by a processor, implements the method described in any one of claims 1 - 9.

Images