Content recommendation method, device and equipment and computer readable storage medium

Abstract

The invention discloses a content recommendation method, device and equipment and a computer readable storage medium. The method comprises the steps of obtaining environment feature information, user feature information of a target user and content feature information of to-be-recommended content; inputting the environment feature information, the user feature information and the content feature information into a content recommendation model, and outputting weight information corresponding to the to-be-recommended content through the content recommendation model; and based on the weight information and the to-be-recommended content, determining target recommended content corresponding to the target user. By adopting the method, dynamic content recommendation can be performed in combination with the environment feature information, the user feature information and the content feature information, and the accuracy of a recommendation result is improved.

Description

Technical Field

[0001] This application relates to the field of data recommendation technology, specifically to a content recommendation method, apparatus, device, and computer-readable storage medium. Background Technology

[0002] With the development of internet technology, the amount of online data has exploded. To enable users to quickly find the content they want or that suits them from this massive amount of data, personalized content recommendations are needed. For example, recommending matching videos from a vast amount of video data. However, current content recommendation and user interaction systems are usually static personalized recommendations, which suffer from inaccuracies. Summary of the Invention

[0003] This application provides a content recommendation method, apparatus, device, and computer-readable storage medium, which can combine environmental feature information, user feature information, and content feature information to perform dynamic content recommendation and improve the accuracy of recommendation results.

[0004] The technical solution adopted by this invention to solve the problem is as follows: Firstly, this application provides a content recommendation method, including: Acquire environmental characteristic information, user characteristic information of target users, and content characteristic information of the content to be recommended; Environmental feature information, user feature information, and content feature information are input into the content recommendation model, and the content recommendation model outputs the weight information corresponding to the content to be recommended. Based on weight information and the content to be recommended, the target recommended content for the target user is determined.

[0005] In some embodiments of this application, when the content recommendation model is deployed in the cloud, environmental feature information, user feature information, and content feature information are input into the content recommendation model, and the content recommendation model outputs the weight information corresponding to the content to be recommended, including: User feature information is anonymized and noise is added to obtain preprocessed user feature information; The hash values ​​corresponding to the preprocessed user feature information, environmental feature information, and content feature information are uploaded to the cloud. The hash values ​​corresponding to the preprocessed user feature information, environmental feature information, and content feature information are then input into the content recommendation model through the cloud to obtain the weight information corresponding to the content to be recommended.

[0006] In some implementations of this application, when the content recommendation model is deployed locally, before inputting environmental feature information, user feature information, and content feature information into the content recommendation model and outputting the weight information corresponding to the content to be recommended by the content recommendation model, the following steps are included: The first network model is trained based on the training data to obtain the second network model; Upload the first model parameters of the second network model to the cloud, and receive the second model parameters returned by the cloud based on the first model parameters; The model parameters of the second network model are updated based on the parameters of the second model to obtain the content recommendation model.

[0007] In some implementation schemes of this application, when the content recommendation model is deployed locally, after determining the target recommendation content corresponding to the target user based on weight information and the content to be recommended, the following steps are taken: Receive feedback data from target users based on target recommended content; The content recommendation model is updated based on feedback data.

[0008] In some implementation schemes of this application, target recommended content for a target user is determined based on weight information and the content to be recommended, including: Based on weight information and the content to be recommended, candidate recommended content is determined; When the target user is an adult, the candidate recommendation content will be determined as the target recommendation content corresponding to the target user; When the target user is a child, the candidate recommendations are filtered to obtain the target recommendations corresponding to the target user.

[0009] In some implementation schemes of this application, after determining the target recommended content for a target user based on weight information and the content to be recommended, the process includes: Responding to the target user's playback command for the target content in the target recommended content, play the target content; If the target user does not watch the target content for more than a first time threshold, the playback of the target content will be paused.

[0010] In some embodiments of this application, after the target content is played, the following are included: Upon receiving a first operation instruction and a second operation instruction targeting the target content, determine whether there is a conflict between the first operation instruction and the second operation instruction; When there is a conflict between the first operation instruction and the second operation instruction, the target operation instruction is determined from the first operation instruction and the second operation instruction based on the target user's user history behavior and / or the target user's gaze focus. Execute the control operation corresponding to the target operation instruction on the target content.

[0011] Secondly, embodiments of the present invention also provide a content recommendation device, comprising: The information acquisition module is used to acquire environmental feature information, user feature information of the target user, and content feature information of the content to be recommended; The weight determination module is used to input environmental feature information, user feature information and content feature information into the content recommendation model, and output the weight information corresponding to the content to be recommended through the content recommendation model. The content recommendation module is used to determine the target recommended content for a target user based on weight information and the content to be recommended.

[0012] Thirdly, this application also provides a terminal device, which includes: One or more processors; Memory; and One or more applications, wherein the applications are stored in memory and configured to be executed by a processor to implement the content recommendation method of any of the first aspects.

[0013] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, the computer program being loaded by a processor to perform the steps in the content recommendation method of any of the first aspects.

[0014] The beneficial effects of this invention are as follows: By acquiring environmental feature information, user feature information of the target user, and content feature information of the content to be recommended, the environmental feature information, user feature information, and content feature information are input into the content recommendation model. The content recommendation model outputs the weight information corresponding to the content to be recommended. Based on the weight information and the content to be recommended, the target recommended content corresponding to the target user is determined. Dynamic content recommendation can be performed by combining environmental feature information, user feature information, and content feature information, thereby improving the accuracy of the recommendation results. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of a content recommendation system provided in an embodiment of the present invention; Figure 2 This is a schematic flowchart of an embodiment of the content recommendation method provided by the present invention; Figure 3 This is a schematic diagram of a specific embodiment of the process for determining target recommended content provided by the present invention; Figure 4 This is a schematic diagram of a specific embodiment of the content recommendation device provided in this invention. Figure 5 This is a schematic diagram of an embodiment of the terminal device provided in this invention. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0018] In the description of this application, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first," "second," etc., may explicitly or implicitly include one or more features.

[0019] In this application, the term "exemplary" is used to mean "used as an example, illustration, or description." Any embodiment described as "exemplary" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use this application. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that this application can be made without using these specific details. In other instances, well-known structures and processes are not described in detail to avoid obscuring the description of this application with unnecessary detail. Therefore, this application is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0020] It should be noted that since the method in this application embodiment is executed in a terminal device, the processing objects of each terminal device exist in the form of data or information, such as time, which is essentially time information. It can be understood that if size, quantity, position, etc. are mentioned in subsequent embodiments, they are all corresponding data that exist so that the terminal device can process them. Specific details will not be elaborated here.

[0021] This application provides a content recommendation method, apparatus, device, and computer-readable storage medium, which will be described in detail below.

[0022] Please see Figure 1 , Figure 1 This is a schematic diagram of a content recommendation system provided in an embodiment of this application. The content recommendation system may include a terminal device 100, which integrates a content recommendation device, such as... Figure 1 Terminal devices in the process.

[0023] In this embodiment, the terminal device 100 is mainly used to acquire environmental feature information, user feature information of the target user, and content feature information of the content to be recommended; input the environmental feature information, user feature information, and content feature information into the content recommendation model, and output the weight information corresponding to the content to be recommended through the content recommendation model; based on the weight information and the content to be recommended, determine the target recommended content corresponding to the target user, and can combine environmental feature information, user feature information, and content feature information to perform dynamic content recommendation, thereby improving the accuracy of the recommendation results.

[0024] In this embodiment, the terminal device 100 can be an independent server, a server network, or a server cluster. For example, the terminal device 100 described in this embodiment includes, but is not limited to, a computer, a network host, a single network server, a set of multiple network servers, or a cloud server composed of multiple servers. The cloud server is composed of a large number of computers or network servers based on cloud computing.

[0025] It is understood that the terminal device 100 used in the embodiments of this application can be a device that includes both receiving and transmitting hardware, that is, a device having receiving and transmitting hardware capable of performing bidirectional communication on a bidirectional communication link. Such a device may include: cellular or other communication devices having a single-line display, a multi-line display, or a cellular or other communication device without a multi-line display. Specifically, the terminal device 100 may be a desktop terminal or a mobile terminal, and the terminal device 100 may also be one of a smart TV, mobile phone, tablet computer, laptop computer, etc.

[0026] Those skilled in the art will understand that Figure 1 The application environment shown is merely one application scenario of the solution in this application and does not constitute a limitation on the application scenario of the solution in this application. Other application environments may include those that are more specific to this application. Figure 1 The number of more or fewer terminal devices shown, for example Figure 1 Only one terminal device is shown in the image. It is understood that the content recommendation system may also include one or more other services, which are not specified here.

[0027] In addition, such as Figure 1 As shown, the content recommendation system may also include a memory 200 for storing data, such as feature information, such as environmental feature information, user feature information, content feature information, etc., and recommended content, such as content to be recommended, target recommended content, etc.

[0028] It should be noted that, Figure 1The schematic diagram of the content recommendation system shown is merely an example. The content recommendation system and scenario described in this application are intended to more clearly illustrate the technical solutions of this application and do not constitute a limitation on the technical solutions provided in this application. As those skilled in the art will know, with the evolution of content recommendation systems and the emergence of new business scenarios, the technical solutions provided in this application are also applicable to similar technical problems.

[0029] First, this application provides a content recommendation method. The execution subject of the content recommendation method is a content recommendation device, which is applied to a terminal device. The content recommendation method includes: acquiring environmental feature information, user feature information of the target user, and content feature information of the content to be recommended; inputting the environmental feature information, user feature information, and content feature information into a content recommendation model, and outputting weight information corresponding to the content to be recommended through the content recommendation model; and determining the target recommended content corresponding to the target user based on the weight information and the content to be recommended.

[0030] like Figure 2 The diagram shown is a flowchart of an embodiment of the content recommendation method in this application. The content recommendation method may include the following steps S201 to S203, as detailed below: Step S201: Obtain environmental feature information, user feature information of the target user, and content feature information of the content to be recommended.

[0031] In this embodiment, the content recommendation method is applied to a terminal device. Environmental feature information is used to characterize the environmental characteristics of the terminal device. The environmental feature information includes one or more of the following: time information, location information, light intensity, temperature, humidity, noise level, device status information, and network status information. Specifically, time information is used to characterize the time period during which the target user uses the terminal device, such as daytime or nighttime; location information is used to characterize the geographical location of the target user using the terminal device, such as home, hotel, or hospital; device status information is used to characterize the current device status of the terminal device, such as computing resources, CPU resources, and network connection status; and network status information is used to characterize the network status of the network currently connected to the terminal device, such as network bandwidth.

[0032] Optionally, environmental characteristic information can be collected by modules configured on the terminal device itself, or it can be obtained from other computer devices via networks, Bluetooth, infrared, etc. For example, light intensity can be obtained through sensors configured on the terminal device, and ambient noise can be obtained through the microphone of the terminal device.

[0033] Furthermore, the target user is the user currently using the terminal device. User characteristic information is used to characterize the user profile of the target user. User characteristic information includes one or more of the following: age, gender, facial expression, body movement, gaze focus, content preference, and real-time behavior. Among them, content preference information includes one or more of the following: historical viewing records, favorites, and interest group information; real-time behavior information includes one or more of the following: current channel viewing duration, interaction repetition, and voice commands.

[0034] Optionally, user characteristic information can be collected through modules configured on the terminal device itself, or it can be obtained from other computer devices via networks, Bluetooth, infrared, etc. For example, facial images can be acquired through the terminal device's camera, and facial recognition can be performed on the facial images to obtain the target user's age, gender, and facial expression information. Voice commands from the target user can also be collected through the terminal device's microphone.

[0035] In this embodiment, the content to be recommended is content that needs to be recommended. The content to be recommended includes one or more of video data, text data, audio data, and image data. For example, the content to be recommended can be a collection of video data to be recommended. Content feature information is used to characterize the content features of the content to be recommended. Content feature information includes at least one of content tag information and contextual features. Content tag information includes one or more of content duration, suitable age group, and rating. Contextual features include compatible playback devices.

[0036] S202. Input environmental feature information, user feature information and content feature information into the content recommendation model, and output the weight information corresponding to the content to be recommended through the content recommendation model.

[0037] In this embodiment, the content recommendation model is a pre-trained neural network model used to predict weight information based on environmental feature information, user feature information, and content feature information. The content recommendation model can be constructed based on lightweight neural network models such as MobileNetV3+LSTM. The weight information is used to characterize the degree of matching between each piece of content to be recommended and the target user. Optionally, the process of determining the weight information can be expressed as follows: ,in, Represents weight information, , and Indicates the dynamic adjustment coefficient. It is a function of environmental feature information E, used to calculate the contribution value of environmental feature information to the weight information. It's about user characteristic information. The function is used to calculate user feature information. The contribution value of the weight information It is about content feature information The function is used to calculate content feature information. The contribution value to the weighting information.

[0038] Furthermore, the content recommendation model can be deployed locally on the terminal device or in the cloud. In some embodiments, when the content recommendation model is deployed in the cloud, the step S202 above, which involves inputting environmental feature information, user feature information, and content feature information into the content recommendation model and outputting the weight information corresponding to the content to be recommended, specifically includes: anonymizing and adding noise to the user feature information to obtain preprocessed user feature information; uploading the hash values ​​corresponding to the preprocessed user feature information, environmental feature information, and content feature information to the cloud, so that the hash values ​​corresponding to the preprocessed user feature information, environmental feature information, and content feature information are input into the content recommendation model through the cloud to obtain the weight information corresponding to the content to be recommended. This embodiment protects user privacy and avoids user information leakage by anonymizing and adding noise to the user feature information and then uploading the preprocessed user feature information to the cloud; by uploading the hash values ​​corresponding to the content feature information to the cloud, the amount of data uploaded can be reduced, and the speed of content recommendation can be improved.

[0039] In this embodiment, anonymization coding is a coding method that processes personal identification information in data using technical means, making the processed data impossible to directly or indirectly associate with a specific natural person, while preserving the original business value of the data as much as possible (such as statistical, analytical, and research purposes). Optionally, hash coding, permutation coding, or k-anonymization can be used to anonymize user feature information.

[0040] Furthermore, noise injection / addition is a technique that actively introduces controllable, meaningless interference signals into the original user feature data. Its core objective is to blur sensitive user feature details, reduce data identifiability, and simultaneously preserve the overall distribution characteristics and statistical regularities of the data to the greatest extent possible, ensuring that the data can still be used for subsequent tasks such as data analysis and model training while protecting privacy. Optionally, local differential privacy (LDP) technology, random replacement technology, or fuzzy merging technology can be used to add noise to user feature information.

[0041] In some embodiments, when the content recommendation model is deployed locally, before step S202 above, which inputs environmental feature information, user feature information, and content feature information into the content recommendation model and outputs the weight information corresponding to the content to be recommended, includes: training a first network model based on training data to obtain a second network model; uploading the first model parameters of the second network model to the cloud and receiving the second model parameters returned by the cloud based on the first model parameters; and updating the model parameters of the second network model based on the second model parameters to obtain the content recommendation model. This embodiment trains the first network model locally based on training data to obtain the first model parameters, and then uploads the first model parameters to the cloud, which can protect user privacy; uploading the first model parameters of the second network model to the cloud and then updating the model parameters of the second network model based on the second model parameters returned by the cloud can improve the accuracy of the content recommendation model, thereby improving the accuracy of the content recommendation results.

[0042] In this embodiment, the first network model can be constructed using a lightweight neural network model such as MobileNetV3+LSTM. The model structure of the first network model is the same as that of the content recommendation model. The difference between the first network model and the content recommendation model is that the model parameters of the first network model are the initial model parameters, while the model parameters of the content recommendation model are the trained model parameters. Optionally, the training data information includes content feature information and tag weight information corresponding to the sample content, user feature information corresponding to the tag weight information, and environmental feature information corresponding to the tag weight information. The steps of training the first network model based on the training data information to obtain the second network model specifically include: inputting the content feature information corresponding to the sample content, the user feature information corresponding to the tag weight information, and the environmental feature information corresponding to the tag weight information into the first network model; outputting the prediction weight information corresponding to the sample content through the first network model; and training the first network model based on the tag weight information, the prediction weight information, and the loss function of the first network model to obtain the second network model.

[0043] In some embodiments, the step of training the first network model based on label weight information, prediction weight information, and the loss function of the first network model to obtain the second network model specifically includes: determining the loss value based on the label weight information, prediction weight information, and the loss function of the first network model; updating the model parameters of the first network model when the loss value has not converged or the number of training iterations of the first network model has not reached the preset number of iterations, and continuing to execute the steps of inputting the content feature information corresponding to the sample content, the user feature information corresponding to the label weight information, and the environmental feature information corresponding to the label weight information into the first network model, and outputting the prediction weight information corresponding to the sample content through the first network model, until the loss value converges or the number of training iterations of the first network model reaches the preset number of iterations; and determining the updated first network model as the content recommendation model.

[0044] S203. Based on weight information and the content to be recommended, determine the target recommended content for the target user.

[0045] In this embodiment, the target recommended content is content that matches the target user, selected from the content to be recommended based on weight information. The target recommended content includes one or more of video data, text data, audio data, and image data. This embodiment inputs environmental feature information, user feature information, and content feature information into the content recommendation model. The content recommendation model outputs the weight information corresponding to the content to be recommended. Based on the weight information and the content to be recommended, the target recommended content is determined. This approach combines environmental feature information, user feature information, and content feature information for dynamic content recommendation, improving the accuracy of the recommendation results.

[0046] In some embodiments, the step of determining the target recommended content corresponding to the target user based on weight information and the content to be recommended specifically includes: filtering the content to be recommended based on weight information to obtain the target recommended content corresponding to the target user; wherein, the target recommended content is the content to be recommended whose weight value corresponding to the weight information satisfies a first condition, the content to be recommended whose weight value corresponding to the weight information satisfies the first condition can be a preset number of content to be recommended with the highest weight value corresponding to the weight information, and the content to be recommended whose weight value corresponding to the weight information satisfies the first condition can also be content to be recommended whose weight value corresponding to the weight information is greater than a third threshold.

[0047] In other embodiments, reference is made to Figure 3 As shown, step S203 above, which determines the target recommended content for the target user based on weight information and the content to be recommended, may include steps S301 to S303, as detailed below: S301. Based on weight information and the content to be recommended, determine the candidate recommended content.

[0048] In this embodiment of the application, the candidate recommended content is content selected from the content to be recommended based on weight information. In some embodiments, the candidate recommended content is the content to be recommended whose weight value corresponding to the weight information satisfies the second condition. The content to be recommended whose weight value corresponding to the weight information satisfies the second condition can be a preset number of content to be recommended with the highest weight value corresponding to the weight information. The content to be recommended whose weight value corresponding to the weight information satisfies the second condition can also be the content to be recommended whose weight value corresponding to the weight information is greater than the fourth threshold.

[0049] S302. When the target user is an adult, the candidate recommended content is determined as the target recommended content corresponding to the target user.

[0050] S303. When the target user is a child, filter the candidate recommendation content to obtain the target recommendation content corresponding to the target user.

[0051] Optionally, the facial image of the target user can be captured by a camera, and the target user's facial image can be recognized by a deep learning model to determine whether the target user is an adult or a child. Alternatively, the voice data of the target user can be captured by a microphone, and the target user's voice data can be recognized by a deep learning model to determine whether the target user is an adult or a child. Or, the age range of the target user can be obtained from the user's feature information, and the target user's age range can be used to determine whether the target user is an adult or a child.

[0052] In this embodiment, when the target user is determined to be an adult, the candidate recommended content is determined to be the target recommended content corresponding to the target user. When the target user is a child, adult content is filtered out from the candidate recommended content, and the filtered candidate recommended content is determined to be the target recommended content corresponding to the target user. This can recommend content that matches the age group of the target user and improve the accuracy of the recommendation results.

[0053] In some embodiments, when the content recommendation model is deployed locally, after determining the target recommendation content corresponding to the target user based on weight information and the content to be recommended in step S203 above, the method includes: receiving feedback data returned by the target user based on the target recommendation content; and updating the content recommendation model based on the feedback data. This embodiment updates the content recommendation model based on feedback data, which can improve the accuracy of the content recommendation model and thus improve the accuracy of the content recommendation results.

[0054] In this embodiment, the feedback data includes at least one of user input data and user behavior data. User input data includes one or both of ratings and comments, while user behavior data includes one or both of skip rate and viewing duration. When updating the content recommendation model based on the feedback data, the model can be retrained or its parameters can be fine-tuned. For example, the dynamic adjustment coefficients in the aforementioned weight information calculation formula can be adjusted based on the feedback data. , and Make dynamic adjustments.

[0055] In some embodiments, after determining the target recommended content corresponding to the target user based on weight information and the content to be recommended in step S203 above, the method includes: responding to the target user's playback instruction for the target content in the target recommended content, and playing the target content; pausing the playback of the target content when the target user has not watched the target content for a period exceeding a first time threshold. This embodiment, by pausing the playback of the target content when the target user has not watched the target content for a period exceeding the first time threshold, can prevent the target user from temporarily leaving and missing the exciting parts of the target content.

[0056] Optionally, it can be determined whether the target user is watching the target content based on eye-tracking technology, or through sensors configured on the terminal device itself (such as a proximity sensor), or by performing image recognition on images captured by the terminal device's camera. The first time threshold is a pre-set time threshold used to measure the duration for which the target user has not watched the target content. The first time threshold can be set according to actual needs; for example, it can be set to 3 minutes, in which case playback of the target content will be paused if the target user has not watched the target content for more than 3 minutes.

[0057] In some embodiments, after playing the target content, the process includes: receiving a first operation instruction and a second operation instruction for the target content; determining whether the first operation instruction and the second operation instruction conflict; if the first operation instruction and the second operation instruction conflict, determining the target operation instruction from the first operation instruction and the second operation instruction based on the target user's historical behavior and / or the target user's gaze focus; and executing the control operation corresponding to the target operation instruction on the target content. This embodiment determines the target operation instruction based on the target user's historical behavior and / or the target user's gaze focus when the first operation instruction and the second operation instruction conflict, which can accurately execute the corresponding control operation when user-issued instructions conflict.

[0058] In this embodiment, the first operation instruction can be any one of a voice instruction, a gesture instruction, or a remote control instruction, and the second operation instruction can also be any one of a voice instruction, a gesture instruction, or a remote control instruction. The first and second operation instructions can be of the same type, or they can be of different types. For example, the first operation instruction can be a voice instruction, and the second operation instruction can be a gesture instruction. A conflict between the first and second operation instructions means that the control operation corresponding to the first operation instruction conflicts with the control operation corresponding to the second operation instruction. For example, if the first operation instruction is a user's spoken "pause," and the second operation instruction is a "fast forward" gesture, then the first and second operation instructions conflict.

[0059] Furthermore, user history behavior is used to characterize the target user's historical operating habits. For example, if the first operation command is a spoken "pause" and the second operation command is a "fast forward" gesture, then there is a conflict between the first and second operation commands. If the target user has used gesture commands in 90% of the past situations, then the second operation command is determined to be the target operation command. Eye focus is the focus of the target user's gaze on the screen. For example, if the first operation command is a spoken "pause" and the second operation command is a "fast forward" gesture, then there is a conflict between the first and second operation commands. If the target user's eye focus is on the progress bar, then the first operation command is determined to be the target operation command.

[0060] In some embodiments, the step of determining a target operation instruction from a first operation instruction and a second operation instruction based on the target user's historical behavior and / or the target user's gaze focus specifically includes: determining whether a candidate operation instruction can be determined from the first operation instruction and the second operation instruction based on the target user's gaze focus; when a candidate operation instruction can be determined based on the target user's gaze focus, determining the candidate operation instruction determined based on the target user's gaze focus as the target operation instruction; when a candidate operation instruction cannot be determined based on the target user's gaze focus, determining a first operation probability corresponding to the first operation instruction and a second operation probability corresponding to the second operation instruction from the user's historical behavior; and determining the target operation instruction from the first operation instruction and the second operation instruction based on the first operation probability and the second operation probability.

[0061] In this embodiment, the first operation probability is used to characterize the probability that the target user executes the first operation instruction, and the second operation probability is used to characterize the probability that the target user executes the second operation instruction. In some embodiments, the step of determining the target operation instruction from the first operation instruction and the second operation instruction based on the first operation probability and the second operation probability specifically includes: when the probability value corresponding to the first operation probability is greater than the probability value corresponding to the second operation probability, the first operation instruction is determined as the target operation instruction; when the probability value corresponding to the first operation probability is less than the probability value corresponding to the second operation probability, the second operation instruction is determined as the target operation instruction.

[0062] In some embodiments, after playing the target content, the process includes: determining the target user's viewing mode based on the target user's posture information, gaze focus, and ambient noise; hiding target interface elements when the viewing mode is a first mode, and / or enabling semantic expansion functionality; enabling hybrid control commands and simplifying command operations when the viewing mode is a second mode, and / or recommending content with a duration within a first duration range from the target recommended content to the target user, thus dynamically adjusting the viewing content according to the user's viewing model.

[0063] In this embodiment, the first mode is used to characterize that the target user is focused on watching the target content. For example, when the target user is sitting upright, their gaze is focused on the center of the screen, and the ambient noise is <40dB, the user mode is determined to be the first mode, unnecessary UI elements are hidden, and semantic expansion functions are enabled (e.g., when the user says "science fiction", it automatically associates with tags such as "space" and "aliens"). The second mode is used to characterize that the target user's viewing mode is a leisure mode. For example, when the target user is lying down, their gaze is scattered, and the ambient noise is >60dB, the user mode is determined to be the second mode, gesture + voice hybrid control is enabled, and instructions are simplified (e.g., clenching a fist → unfolding = favorite, two-finger swipe = switch recommendations), and short videos with a duration of 5-15 minutes are recommended.

[0064] To better implement the content recommendation method in the embodiments of this application, based on the content recommendation method, the embodiments of this application also provide a content recommendation device, such as... Figure 4 As shown, the content recommendation device 600 includes: The information acquisition module 610 is used to acquire environmental feature information, user feature information of the target user, and content feature information of the content to be recommended. The weight determination module 620 is used to input environmental feature information, user feature information and content feature information into the content recommendation model, and output the weight information corresponding to the content to be recommended through the content recommendation model. The content recommendation module 630 is used to determine the target recommended content for a target user based on weight information and the content to be recommended.

[0065] In this embodiment, by acquiring environmental feature information, user feature information of the target user, and content feature information of the content to be recommended, the environmental feature information, user feature information, and content feature information are input into the content recommendation model. The content recommendation model outputs the weight information corresponding to the content to be recommended. Based on the weight information and the content to be recommended, the target recommended content corresponding to the target user is determined. Dynamic content recommendation can be performed by combining environmental feature information, user feature information, and content feature information, thereby improving the accuracy of the recommendation results.

[0066] In some embodiments of this application, when the content recommendation model is deployed in the cloud, the weight determination module 620 inputs environmental feature information, user feature information, and content feature information into the content recommendation model, and outputs the weight information corresponding to the content to be recommended through the content recommendation model, including: User feature information is anonymized and noise is added to obtain preprocessed user feature information; The hash values ​​corresponding to the preprocessed user feature information, environmental feature information, and content feature information are uploaded to the cloud. The hash values ​​corresponding to the preprocessed user feature information, environmental feature information, and content feature information are then input into the content recommendation model through the cloud to obtain the weight information corresponding to the content to be recommended.

[0067] In some embodiments of this application, when the content recommendation model is deployed locally, before the weight determination module 620 inputs environmental feature information, user feature information, and content feature information into the content recommendation model and outputs the weight information corresponding to the content to be recommended through the content recommendation model, the weight determination module 620 is further configured to: The first network model is trained based on the training data to obtain the second network model; Upload the first model parameters of the second network model to the cloud, and receive the second model parameters returned by the cloud based on the first model parameters; The model parameters of the second network model are updated based on the parameters of the second model to obtain the content recommendation model.

[0068] In some embodiments of this application, when the content recommendation model is deployed locally, after the content recommendation module 630 determines the target recommendation content corresponding to the target user based on weight information and the content to be recommended, the content recommendation module 630 is further configured to: Receive feedback data from target users based on target recommended content; The content recommendation model is updated based on feedback data.

[0069] In some embodiments of this application, the content recommendation module 630 determines the target recommended content corresponding to the target user based on weight information and the content to be recommended, including: Based on weight information and the content to be recommended, candidate recommended content is determined; When the target user is an adult, the candidate recommendation content will be determined as the target recommendation content corresponding to the target user; When the target user is a child, the candidate recommendations are filtered to obtain the target recommendations corresponding to the target user.

[0070] In some embodiments of this application, after the content recommendation module 630 determines the target recommended content corresponding to the target user based on weight information and the content to be recommended, the content recommendation module 630 is further configured to: Responding to the target user's playback command for the target content in the target recommended content, play the target content; If the target user does not watch the target content for more than a first time threshold, the playback of the target content will be paused.

[0071] In some embodiments of this application, after the content recommendation module 630 plays the target content, the content recommendation module 630 is further configured to: Upon receiving a first operation instruction and a second operation instruction targeting the target content, determine whether there is a conflict between the first operation instruction and the second operation instruction; When there is a conflict between the first operation instruction and the second operation instruction, the target operation instruction is determined from the first operation instruction and the second operation instruction based on the target user's user history behavior and / or the target user's gaze focus. Execute the control operation corresponding to the target operation instruction on the target content.

[0072] This application embodiment also provides a terminal device that integrates any of the content recommendation devices provided in this application embodiment. The terminal device includes: One or more processors; Memory; and One or more applications, wherein the applications are stored in memory and configured to be executed by a processor from the steps of the content recommendation method in any of the embodiments described above.

[0073] This application also provides a terminal device that integrates any of the content recommendation devices provided in this application. For example... Figure 5 As shown, it illustrates a structural schematic diagram of the terminal device involved in the embodiments of this application. Specifically: The terminal device may include components such as a processor 801 with one or more processing cores, a memory 802 with one or more computer-readable storage media, a power supply 803, and an input unit 804. Those skilled in the art will understand that... Figure 5 The terminal device structure shown does not constitute a limitation on the terminal device and may include more or fewer components than shown, or combine certain components, or have different component arrangements. Wherein: The processor 801 is the control center of the terminal device. It connects various parts of the terminal device via various interfaces and lines, and performs various functions and processes data by running or executing software programs and / or modules stored in the memory 802, and by calling data stored in the memory 802, thereby providing overall monitoring of the terminal device. Optionally, the processor 801 may include one or more processing cores; preferably, the processor 801 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the modem processor may not be integrated into the processor 801.

[0074] The memory 802 can be used to store software programs and modules. The processor 801 executes various functional applications and data processing by running the software programs and modules stored in the memory 802. The memory 802 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the terminal device, etc. In addition, the memory 802 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, the memory 802 may also include a memory controller to provide the processor 801 with access to the memory 802.

[0075] The terminal device also includes a power supply 803 that supplies power to the various components. Preferably, the power supply 803 can be logically connected to the processor 801 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system. The power supply 803 may also include one or more DC or AC power supplies, recharging systems, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components.

[0076] The terminal device may also include an input unit 804, which can be used to receive input digital or character information, and generate keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function control.

[0077] Although not shown, the terminal device may also include a display unit, etc., which will not be described in detail here. Specifically, in this embodiment, the processor 801 in the terminal device loads the executable files corresponding to the processes of one or more applications into the memory 802 according to the following instructions, and the processor 801 runs the applications stored in the memory 802 to realize various functions, as follows: Acquire environmental characteristic information, user characteristic information of target users, and content characteristic information of the content to be recommended; Environmental feature information, user feature information, and content feature information are input into the content recommendation model, and the content recommendation model outputs the weight information corresponding to the content to be recommended. Based on weight information and the content to be recommended, the target recommended content for the target user is determined.

[0078] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be performed by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor.

[0079] Therefore, embodiments of this application provide a computer-readable storage medium, which may include: read-only memory (ROM), random access memory (RAM), a magnetic disk, or an optical disk, etc. A computer program is stored thereon, and the computer program is loaded by a processor to execute the steps in any of the content recommendation methods provided in embodiments of this application. For example, the computer program loaded by the processor can execute the following steps: Acquire environmental characteristic information, user characteristic information of target users, and content characteristic information of the content to be recommended; Environmental feature information, user feature information, and content feature information are input into the content recommendation model, and the content recommendation model outputs the weight information corresponding to the content to be recommended. Based on weight information and the content to be recommended, the target recommended content for the target user is determined.

[0080] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the detailed descriptions of other embodiments above, which will not be repeated here.

[0081] In practice, each of the above units or structures can be implemented as an independent entity or can be arbitrarily combined to be implemented as the same or several entities. For the specific implementation of each of the above units or structures, please refer to the previous method embodiments, which will not be repeated here.

[0082] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0083] The foregoing has provided a detailed description of a content recommendation method, apparatus, device, and computer-readable storage medium provided in the embodiments of this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A content recommendation method, characterized in that, include: Acquire environmental characteristic information, user characteristic information of target users, and content characteristic information of the content to be recommended; The environmental feature information, user feature information, and content feature information are input into the content recommendation model, and the weight information corresponding to the content to be recommended is output through the content recommendation model. Based on the weight information and the content to be recommended, the target recommended content for the target user is determined.

2. The content recommendation method according to claim 1, characterized in that, When the content recommendation model is deployed in the cloud, the process of inputting the environmental feature information, the user feature information, and the content feature information into the content recommendation model, and outputting the weight information corresponding to the content to be recommended through the content recommendation model, includes: The user feature information is anonymized and noise is added to obtain preprocessed user feature information; The hash values ​​corresponding to the preprocessed user feature information, environmental feature information, and content feature information are uploaded to the cloud. The hash values ​​corresponding to the preprocessed user feature information, environmental feature information, and content feature information are then input into the content recommendation model through the cloud to obtain the weight information corresponding to the content to be recommended.

3. The content recommendation method according to claim 1, characterized in that, When the content recommendation model is deployed locally, before inputting the environmental feature information, user feature information, and content feature information into the content recommendation model, and outputting the weight information corresponding to the content to be recommended through the content recommendation model, the process includes: The first network model is trained based on the training data to obtain the second network model; Upload the first model parameters of the second network model to the cloud, and receive the second model parameters returned by the cloud based on the first model parameters; The model parameters of the second network model are updated based on the second model parameters to obtain the content recommendation model.

4. The content recommendation method according to claim 1, characterized in that, When the content recommendation model is deployed locally, after determining the target recommendation content corresponding to the target user based on the weight information and the content to be recommended, the process includes: Receive feedback data returned by the target user based on the target recommended content; The content recommendation model is updated based on the feedback data.

5. The content recommendation method according to claim 1, characterized in that, The step of determining the target recommended content for the target user based on the weight information and the content to be recommended includes: Based on the weight information and the content to be recommended, candidate recommended content is determined; When the target user is an adult, the candidate recommendation content is determined as the target recommendation content corresponding to the target user; When the target user is a child, the candidate recommended content is filtered to obtain the target recommended content corresponding to the target user.

6. The content recommendation method according to claim 1, characterized in that, After determining the target recommended content for the target user based on the weight information and the content to be recommended, the process includes: In response to the target user's playback command for target content in the target recommended content, the target content is played. When the target user has not watched the target content for a period of time exceeding a first time threshold, the playback of the target content is paused.

7. The content recommendation method according to claim 6, characterized in that, After playing the target content, the process includes: Upon receiving a first operation instruction and a second operation instruction for the target content, determine whether there is a conflict between the first operation instruction and the second operation instruction; When there is a conflict between the first operation instruction and the second operation instruction, the target operation instruction is determined from the first operation instruction and the second operation instruction based on the target user's user history behavior and / or the target user's gaze focus. The control operation corresponding to the target operation instruction is executed on the target content.

8. A content recommendation device, characterized in that, include: The information acquisition module is used to acquire environmental feature information, user feature information of the target user, and content feature information of the content to be recommended; The weight determination module is used to input the environmental feature information, the user feature information and the content feature information into the content recommendation model, and output the weight information corresponding to the content to be recommended through the content recommendation model. The content recommendation module is used to determine the target recommended content for the target user based on the weight information and the content to be recommended.

9. A terminal device, characterized in that, The terminal device includes: One or more processors; Memory; and One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the processor to implement the content recommendation method of any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores a computer program, which is loaded by a processor to perform the steps of the content recommendation method according to any one of claims 1 to 7.

Images