Task management method based on digital avatar and terminal equipment

Abstract

The task management method based on the digital copies comprises the steps that in response to a first operation of a user, the digital copies of the user are output, and the digital copies are generated based on a user portrait of the user; based on the historical behavior data of the user, a first task corresponding to the target date is obtained, the first task is used for representing task planning of the user on the target date, first information is output, and the first information comprises the first task; and executing the first task in response to the determined execution operation of the user. Based on the historical behavior data of the user, the task suggestion corresponding to the target date is actively generated for the user and executed, so that the digital avatar can help the user complete part of work in the digital world, and the work and life efficiency of the user is improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence (AI) technology, and in particular to a task management method and terminal device based on digital avatars. Background Art

[0002] With the rapid development of artificial intelligence, innovative applications such as digital avatars have emerged. These avatars, based on AI and virtual reality technologies, not only resemble real individuals in appearance but also enable intelligent interaction and task execution in a variety of scenarios. For example, in industries such as e-commerce, finance, and telecommunications, digital avatars can serve as virtual customer service representatives, providing users with 24 / 7 online support. While digital avatars can be widely used in various industries, current digital avatar systems still lack the ability to proactively communicate with users. Therefore, developing digital avatars with active learning and communication capabilities has become a pressing issue. Summary of the Invention

[0003] The present invention provides a task management method and terminal device based on a digital twin. By proactively generating task suggestions corresponding to target dates for users, the digital twin system can help users complete some tasks in the digital world, thereby improving their work and life efficiency.

[0004] In a first aspect, an embodiment of the present application provides a task management method based on a digital avatar, including: outputting a digital avatar of the user in response to a first operation of the user, where the digital avatar is generated based on the user's user portrait; obtaining a first task corresponding to a target date based on the user's historical behavior data, where the first task is used to characterize the user's task plan on the target date; outputting first information, where the first information includes the first task; and executing the first task in response to the user's confirmation execution operation.

[0005] In some examples, the historical behavior data is used to characterize the user's behavior over a period of historical time.

[0006] In this way, the digital twin system proactively generates the first task for the user on the target date, thereby proactively planning tasks for the user and improving the user's work and life efficiency.

[0007] In one possible implementation, before outputting the user's digital avatar, the process also includes: obtaining a user profile of the user; and generating the digital avatar based on the user profile. In this way, the user's image can be replicated based on the user's personal information and preferences.

[0008] In a possible implementation, before outputting the digital avatar of the user, it further includes: obtaining second information for characterizing the user identity; authenticating the user based on the second information. In this way, authenticating the user's identity before outputting the digital avatar of the user ensures the security of using the digital avatar.

[0009] In a possible implementation, obtaining the first task corresponding to the target date based on the user's historical behavior data includes: inputting the target date into the first model to obtain the first task, where the first model is trained based on the user's historical behavior data. In this way, since the first model is trained based on the user's historical behavior data, the first model can infer the tasks that the user may need to complete on the target date according to the user's historical behavior data. The digital avatar can remember the user's historical behavior and predict the user's current behavior according to the user's historical behavior, improving the user experience.

[0010] In a possible implementation, in response to the user's determination to execute the operation, executing the first task includes: inputting the first task into the first model to obtain at least one task to be executed and the task type corresponding to the task to be executed; based on the task type corresponding to the task to be executed, calling the task tool corresponding to the task type to execute the task to be executed. In this way, by decomposing the task to be executed by the user through the first model, the digital avatar can automatically execute the task to be executed in the digital world, greatly improving the user experience.

[0011] In a possible implementation, after outputting the first information, it further includes: in response to the user inputting a determination not to execute the operation, outputting third information for prompting the user to input a second task, where the second task is the task to be executed corresponding to the target date for the user. In this way, in the case where the user does not execute the first task, by guiding the user to input the task information that the user needs to execute, the digital avatar can help the user execute the task, improving the user's work and life efficiency.

[0012] In a possible implementation, the first task includes multiple subtasks; in response to the user inputting a determination not to execute the operation, outputting the third information includes: in response to the user's determination not to execute each subtask in the first task, outputting the third information, or in response to the user's determination not to execute at least one target subtask in the first task, outputting the third information and executing the subtasks in the first task other than the target subtask. In this way, the user can choose not to execute some subtasks in the first task. In the case where there are subtasks in the first task that do not need to be executed, by guiding the user to input the task information that the user needs to execute, the digital avatar can help the user execute the task, improving the user's work and life efficiency.

[0013] In a possible implementation, the first task includes multiple subtasks. In response to the user's determination not to perform an operation, third information is output, including: when the number of target subtasks that the user determines not to perform among the multiple subtasks is greater than or equal to the first threshold, output the third information and execute the subtasks in the first task other than the target subtasks.

[0014] In a possible implementation, the first task includes multiple subtasks; in response to the user's determination to perform an operation, the first task is executed, including: in response to the user's determination to perform an operation on each subtask in the first task, execute the subtasks in the first task; or, in response to the user's operation on at least one target subtask in the first task, execute the target subtask in the first task. In this way, the user can choose to execute some of the subtasks in the first task.

[0015] In a possible implementation, the method further includes: generating fourth information in the first time period of the target date, where the fourth information is used to prompt the user to input summary information corresponding to the target date, and the summary information is used to describe the user's behavior habits on the target date; controlling the first model to be trained in the second time period of the target date, and the first time period is earlier than the second time period. In this way, by regularly obtaining the summary information input by the user and regularly training the first model, the digital avatar can regularly learn the user's behavior habits. The accuracy of the digital avatar in generating suggestions for the user's behavior habits on a specified date is improved.

[0016] In a possible implementation, controlling the first model to be trained in the second time period of the target date includes: determining whether there is a user in front of the screen of the terminal device, and the digital avatar is running on the terminal device; when there is no user in front of the screen of the device, controlling the first model to be trained; when it is determined that there is a user in front of the screen of the device and the first model is in a training state, controlling the first model to stop training. In this way, when there is no one in front of the device screen, the first model is trained (i.e., the digital avatar is trained), avoiding training the digital avatar when the user may use the digital avatar. The user experience is improved.

[0017] In a possible implementation, controlling the first model to be trained in the second time period of the target date includes: obtaining the computing resource utilization rate of the device; when the computing resource utilization rate is less than the first threshold, controlling the first model to be trained; when the computing resource utilization rate is greater than the second threshold and the first model is in a training state, controlling the first model to stop training, and the second threshold is greater than the first threshold. In this way, when the computing resource utilization rate of the device is small, the first model is trained, reducing the operating load of the device.

[0018] In a possible implementation, controlling the training of the first model includes: obtaining fifth information, where the fifth information is the information generated by the digital avatar interacting with the user on the target date, and the fifth information includes one or more of the interaction information between the digital avatar and the user, the summary information corresponding to the target date, and the interaction information between the digital avatar and the task tool; training the first model based on the fifth information. In this way, the digital avatar can remember and learn the user's behavior habits on that day.

[0019] In a second aspect, an embodiment of the present application provides a terminal device, including:

[0020] At least one memory for storing programs;

[0021] At least one processor for executing the programs stored in the memory. When the programs stored in the memory are executed, the processor is used to execute the method described in the first aspect or any possible implementation manner of the first aspect.

[0022] In a third aspect, an embodiment of the present application provides a computer storage medium. Instructions are stored in the computer storage medium. When the instructions are run on a computer, the computer is caused to execute the method described in the first aspect or any possible implementation manner of the first aspect.

[0023] In a fourth aspect, an embodiment of the present application provides a computer program product containing instructions. When the instructions are run on a computer, the computer is caused to execute the method described in the first aspect or any possible implementation manner of the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0025] Figure 1 It is a schematic structural diagram of a digital avatar system provided by an embodiment of the present application;

[0026] Figure 2a It is a schematic diagram of an application scenario provided by an embodiment of the present application;

[0027] Figure 2b It is another schematic diagram of an application scenario provided by an embodiment of the present application;

[0028] Figure 3 It is a schematic diagram of the operation process of a digital avatar system provided in an embodiment of the present application;

[0029] Figure 4Schematic diagram of the task processing process of a digital avatar system provided by an embodiment of the present application;

[0030] Figure 5 Schematic flow chart of a task management method based on digital avatars provided by an embodiment of the present application

[0031] Figure 6 Schematic structural diagram of a task management device provided by an embodiment of the present application. Detailed implementation manners

[0032] In order to make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described below with reference to the accompanying drawings.

[0033] In the description of the embodiments of the present application, any embodiment or design solution with "exemplary", "for example", or "for instance" should not be understood as being more preferred or having more advantages than other embodiments or design solutions. Rather, the use of words such as "exemplary", "for example", or "for instance" is intended to present relevant concepts in a specific manner.

[0034] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features. The terms "include", "comprise", "have" and their variants all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0035] Before introducing the present solution, the key terms in the embodiments of the present application will be introduced first.

[0036] (1) Digital avatar

[0037] A digital avatar refers to a virtual model created through technologies such as computer graphics, virtual reality (VR), augmented reality (AR), artificial intelligence, machine learning, and the Internet of Things (IoT), which is highly similar to a real individual or object. These virtual models not only resemble real objects in appearance but also can accurately simulate and interact in real time in terms of behavior, performance, and function.

[0038] (2) Large model

[0039] Large Models refer to machine learning models with large-scale parameters and complex computational structures. These models are typically constructed by deep neural networks and have billions or even hundreds of billions of parameters. The design purpose of large models is to improve the expressive power and prediction performance of the models, enabling them to handle more complex tasks and data. Large models have a wide range of applications in various fields, including natural language processing (NLP), computer vision (CV), speech recognition, and recommendation systems, etc. Large models learn complex patterns and features by training on massive amounts of data and have stronger generalization capabilities, allowing them to make accurate predictions on unseen data.

[0040] (3) Incremental Training

[0041] Incremental training is a machine learning method that allows the model to be gradually updated as new data is added based on existing knowledge, without having to retrain the entire model from scratch. The core of incremental training lies in the model's ability to continuously learn new knowledge and remain stable in the face of complex and changing environmental variations.

[0042] (4) Checkpoint Training

[0043] Checkpoint training is a technique for saving the intermediate state of the model during training so that training can be resumed from the most recent saved point after a training interruption. Its main purpose is to prevent the loss of training progress due to accidental interruptions (such as program crashes, system restarts, or manual stopping of training).

[0044] (5) Intermediate State of Model Training

[0045] The intermediate state of model training refers to the state of the model at a certain moment during the model training process, which can include the following key parts:

[0046] Model weights refer to the set of model parameters that determine the prediction ability of the model. During training, the model continuously adjusts the weights to minimize the loss function, thereby better fitting the training data.

[0047] Optimizer state includes the internal state accumulated by the optimizer during training, such as momentum, the square of the gradient, etc. This state information helps the optimizer update the model weights more effectively in subsequent training.

[0048] Epoch number indicates the number of complete training cycles that the model has completed. One training cycle refers to the process of the model performing one forward propagation and one backward propagation on the entire training dataset.

[0049] The current loss refers to the loss value of the model at a certain moment during the training process. The loss value is used to measure the difference between the model's predicted value and the true value, and is the basis for optimizing the model's weights.

[0050] Other auxiliary information, such as the state of the learning rate scheduler, the state of the random number generator (used for data augmentation and batch sampling), the index of the training data, etc.

[0051] By saving and loading these intermediate states, the training can be paused and resumed at any time during the training process, ensuring that the training progress is not lost. At the same time, the training can also be continued on different devices or in different environments.

[0052] (6) OpenCV

[0053] OpenCV is an open-source computer vision and machine learning software library, which is widely used in image processing, computer vision, deep learning and other fields.

[0054] (7) General large models

[0055] General large models refer to those large models that can be used in multiple fields and tasks. They use deep learning algorithms with large computing power, massive open data and huge amounts of parameters to train on large-scale unlabeled data to find features and discover patterns, and then form a powerful generalization ability to "draw inferences from one instance".

[0056] (8) Historical behavior data

[0057] Historical behavior data refers to various behavior information of users in the past period. For example, historical behavior data can include users' browsing behaviors (such as browsing records, stay times, click paths on web pages), purchase behaviors (such as users' purchase records, purchased product categories), interaction behaviors (such as likes, comments, sharing behaviors of users on social platforms), etc. Users' historical behavior data reflects users' preferences, habits and decision-making patterns. Exemplarily, users' historical behavior data can also include historical behaviors in the work aspect. Historical behavior data in the work aspect reflects users' work patterns.

[0058] Exemplarily, users' historical behavior data can be as follows:

[0059] Monday, 9:00 - 10:00, have a team meeting;

[0060] Wednesday, 14:00 - 17:00, organize team building;

[0061] Friday, 16:00 - 17:00, write a weekly work summary.

[0062] Next, this solution will be introduced.

[0063] Digital humans generated based on large model technology can be applied in multiple scenarios. For example, in the entertainment field, digital humans can serve as virtual idols for music performances, live interactions, brand endorsements, etc. In the service field, digital humans can act as online customer service to handle user inquiries and complaints. In the education field, digital humans can serve as virtual teachers to provide personalized teaching guidance and answer questions. However, when digital humans are used in these fields, they usually only support passively implementing the tasks proposed by users and do not support actively planning tasks for users, etc.

[0064] The embodiment of the present application provides a digital avatar management method, which supports the digital avatar to actively interact with users in real time to help users complete most of the work in the digital world, greatly improving the work and life efficiency of users.

[0065] Exemplarily, Figure 1 shows a schematic structural diagram of a digital avatar system provided by an embodiment of the present application. As Figure 1 shown, the digital avatar system 100 may include: a control module 110, a first model 120, and an information library 130.

[0066] The control module 130 is used to control the digital avatar to interact with the user in real time. For example, the control module 130 can obtain the instruction information of the user and call the first model 120 or other task processing tools based on the obtained user instruction information to implement the intention related to the user instruction. Among them, the task processing tools that the control module 130 can call may include; a second model, an Agent tool, a third model, an application program (APP), etc., where the second model may be a general large model and the third model may be an industry-specific large model.

[0067] The first model 120 can be used to generate information related to the user's behavior habits. For example, the first model 120 can generate the digital avatar image of the user according to the user's photo information and personal portrait information. Exemplarily, the first model 120 may be a large model or a large language model (LLM).

[0068] The information repository 130 is used to store user data information and the information of the interaction between the digital avatar system and the user. For example, the information repository 130 can store the user's personal portrait information, the user's diary / weekly journal information, browsing logs, the interaction information between the control module 130 and external systems, etc. Among them, the user's diary information refers to the summary information of the user's daily work, and the weekly journal information refers to the summary information of the user's weekly work; the browsing log refers to the data set of the user's browsing behavior on the Internet, which details the websites, pages visited by the user during a specific period and the relevant browsing behavior information; the interaction information between the control module and external systems refers to the interaction information between the control module 120 and the application program when the control module responds to the obtained user instruction information and calls the application program related to the user instruction. In a possible example, the information stored in the information repository 130 can be stored in the form of a database or a file.

[0069] It can be understood that the above digital avatar system and the task processing tools required to be called during the process of the digital avatar system executing tasks can be deployed on the terminal device as shown in Figure 2a The user's digital avatar image is displayed on the display of the terminal device, and an interaction box is displayed beside the digital avatar image. Among them, based on the digital avatar system, the digital avatar can interact with the user through the interaction box.

[0070] In one example, as shown in Figure 2b While ensuring the security of user information, the information repository in the digital avatar system can also be deployed in the cloud. By deploying the information repository in the cloud, the occupancy of storage resources on the terminal device can be reduced.

[0071] In another possible example, the installation file of the digital avatar system can also be stored in a USB flash drive. When the user needs to use the digital avatar, the user can insert the USB flash drive into the terminal device, so that the terminal device can load and start the digital avatar system. Among them, all the information generated each time the digital avatar system is started on the terminal device needs to be saved to the USB flash drive. By storing the installation file of the digital avatar system in the USB flash drive and starting the digital avatar system through the USB flash drive, the use of the digital avatar is not restricted by the venue.

[0072] It can be understood that the digital avatar system creates digital avatars through a series of technical means (such as 3D modeling, deep learning, natural language processing, etc.) to enable them to simulate the behaviors and interactions of real people, and the digital avatar system is also used to provide a platform for the operation and interaction of digital avatars. In other words, after importing the user information of User 1 into the digital avatar system, the digital avatar system containing the user information of User 1 can be called the digital avatar corresponding to User 1. When the user information of User 1 in the digital avatar system is cleared and the user information of User 2 is imported, then the current digital avatar system can be called the digital avatar corresponding to User 2. <http: / / www.example.com / <http: / / www.example.com /

[0073] Next, based on <http: / / www.example.com / Figure 1 the digital avatar system shown, the operation process of the digital avatar system will be introduced. It can be understood that the digital avatar system can run on a device or terminal device with computing power and processing power. For the sake of convenience of description, the following takes the digital avatar system running on a terminal device as an example for introduction. Exemplarily, <http: / / www.example.com / Figure 3 shows a schematic diagram of the operation process of the digital avatar system. Refer to <http: / / www.example.com / Figure 3 , the operation process of the digital avatar system can be divided into 3 stages, namely: (1) initialization stage, (2) operation stage, (3) optimization stage. <http: / / www.example.com / <http: / / www.example.com /

[0074] (1) Initialization stage <http: / / www.example.com / <http: / / www.example.com /

[0075] The initialization stage refers to the stage when the user starts the digital avatar system and generates the digital avatar image corresponding to the user. In the initialization stage, the user needs to start the digital avatar system on the terminal device. For example, the user can double-click the shortcut icon corresponding to the digital avatar system on the display interface corresponding to the terminal device to start the digital avatar system. In a possible example, starting the digital avatar system may include three stages: loading the program file, initializing the program, and displaying the interface. Among them, in the stage of loading the program file, in response to the user's double-click operation on the shortcut icon corresponding to the digital avatar system, the operating system on the terminal device will read the program file (such as,.exe file) from the installation path corresponding to the digital avatar system. In the stage of initializing the program, the digital avatar system will load the necessary resources (such as configuration files, data files, etc.) during the operation of the digital avatar system to start the digital avatar system. In the stage of displaying the interface, after the digital avatar system is started, a window or interface can be displayed on the display interface of the terminal device, and the user can interact with the digital avatar system through the displayed window or interface. <http: / / www.example.com / <http: / / www.example.com /

[0076] During the startup process of the digital avatar system, the digital avatar system can also authenticate the user. If the authentication is successful, the digital avatar system can generate the user's digital avatar image and interactive dialog box, and display the generated digital avatar image and interactive dialog box on the corresponding display interface of the digital avatar system. Among them, the interactive dialog box and the digital avatar image can be displayed in one display interface, and the interactive dialog box can be displayed next to the digital avatar image. In a possible example, as Figure 3 shown, the process of the digital avatar system verifying the user may include the following steps:

[0077] In step S1, the control module 110 in the digital avatar system acquires the user's face image. For example, the control module 110 can call the camera on the terminal device to take a picture of the user to acquire the user's face image.

[0078] In step S2, the control module 110 can authenticate the user based on the acquired user face image and the face image information stored in the information library 130. Exemplarily, after the control module 110 acquires the user's face image, the control module 110 can extract the face features in the face image. Then, the control module 110 compares the extracted face features with the face features stored in the information library 130 and outputs the comparison result. Among them, when performing face feature matching, the similarity score of the feature points can be calculated, and the similarity score obtained is used to determine whether the face features collected by the camera match the face features stored in the information library. If the face features collected by the camera match the face features stored in the information library 130 successfully, a prompt of "authentication successful" can be output. If the face features collected by the camera do not match the face features stored in the information library 130, a prompt of "authentication failed" can be output.

[0079] In some embodiments, when the terminal device does not have a camera or other image acquisition function, the control module 110 can also verify the user's identity through the account and password input by the user. For example, after the control module 110 acquires the account and password 1 input by the user, the control module 110 acquires the password 2 corresponding to the account from the information library 130 based on the account input by the user. Then, the control module 110 matches the password 1 and the password 2. If the password 1 and the password 2 are the same, the user authentication is considered successful; otherwise, the authentication fails.

[0080] In step S3, when the user authentication is passed, based on the user photo information and personal portrait information stored in the information repository 130, the control module 110 may call the first model 120 to generate the digital avatar image of the user. Exemplarily, after the user authentication is passed, the control module 110 obtains the user's photo information and personal portrait information from the information repository 130. Then, the control module 110 inputs the obtained user photo information and personal portrait information into the first model 120, and generates the digital avatar image of the user through the first model 120. In a possible example, when the user authentication is passed, the control module 110 may first obtain the digital avatar image corresponding to the user from the information repository 130. If the obtaining fails (that is, the digital avatar image corresponding to the user is not saved in the information repository), the control module 110 may call the first model 120 to generate the digital avatar image of the user.

[0081] In step S4, after generating the digital avatar image of the user through the first model 120, the control module 110 may also store the generated digital avatar image in the information repository 130, so that after the digital avatar system is started next time, there is no need to regenerate the digital avatar image of the user. In some embodiments, when the digital avatar system is used for the first time, the control module 110 calls the first model 120 to generate the digital avatar image of the user and saves it in the information repository 130. After that, every time the digital avatar system is started, the control module 110 obtains the digital avatar image of the user from the information repository 130. When the control module 110 detects that at least one of the user photo information and personal portrait information in the information repository 130 is updated, the control module 110 may call the large model 130 to regenerate the digital avatar image based on the updated user photo information and / or personal portrait information.

[0082] (2) Running stage

[0083] In the running stage, the digital avatar system may obtain the date information on the terminal device, and actively generate suggestions for the user's behavior habits on that date based on the obtained date information. Then, the digital avatar system performs tasks based on the generated suggestions for the user's behavior habits.

[0084] In a possible example, as Figure 3 shown, the process of the digital avatar system generating suggestions for the user's behavior habits and performing tasks based on the suggestions for the user's behavior habits may include the following steps:

[0085] In step S5, the control module 110 obtains the user's facial expression information and generates a greeting based on the user's facial expression information. For example, the control module 110 can call the camera on the terminal device to obtain the user's face image, and obtain the user's facial expression information based on the user's face image. For example, the control module 110 can use OpenCV and a deep learning model for face detection and expression prediction. After the control module 110 obtains the user's facial expression information, the control module can output a greeting in the interaction dialog box according to the obtained facial expression information. Exemplarily, if the facial expression information of the user obtained by the control module 110 is "sad", the control module 110 can output the greeting "You seem a bit sad today. Is there anything I can do to help?" in the interaction dialog box.

[0086] In a possible example, in the configuration file of the digital avatar system, greetings corresponding to different facial expressions can be preset. For example, when the "sad" expression is detected, the digital avatar system will display the greeting: "You seem a bit sad today. Is there anything I can do to help?"; when the "happy" expression is detected, the digital avatar system will display: "You're in a great mood today! Is there anything I can do for you?"

[0087] In step S6, the control module 110 obtains the target date. Exemplarily, the control module 110 can use the date displayed on the terminal device as the target date. Specifically, the control module 110 can obtain the date displayed on the terminal device through the operating system of the terminal device. The obtained target date can include: year, month, day, and day of the week X.

[0088] In step S7, the first model 120 generates user behavior habit suggestions corresponding to the target date based on the obtained target date (which can also be referred to as the first task hereinafter). After the control module 110 obtains the target date, it inputs the obtained target date into the first model 120. The first model 120 can output the user behavior habit suggestions corresponding to this date according to the input target date. Among them, the behavior habit suggestions output by the first model 120 can be obtained based on the user's historical behavior habits, or can be inferred based on the user's historical behavior habits. Exemplarily, the user's historical behavior data shows that the user writes meeting minutes every Tuesday and plays badminton every Thursday. When the target date input to the first model 120 is "March 5th, Wednesday", the large model analyzes the user's historical behavior and determines that there is no task arrangement on Wednesday. However, the user did not write the meeting minutes on March 4th (Tuesday). At this time, the first model 120 can output the user behavior habit suggestion: "Write the meeting minutes for March 3rd". When the target date input to the first model 120 is "March 6th, Thursday", the large model analyzes the user's historical behavior and outputs the user behavior habit suggestion: "Play badminton".

[0089] It can be understood that the historical behavior data of the user can be used to train the first model 120 so that the first model 120 can output the behavior habit suggestions corresponding to the input date information according to the input date information. For example, the user's historical behavior data includes: sending work emails on Monday, playing badminton on Wednesday, writing a weekly work summary on Friday, etc. After training the first model 120 based on the user's historical behavior data, if the date received by the large model is: "March 9th, Friday", then the user behavior habit suggestion output by the first model 120 is: "Write a weekly work summary".

[0090] In a possible example, when the first model 120 generates the user's behavior habit suggestions corresponding to the target date according to the input target date, the user's behavior habit suggestions corresponding to the target date can include multiple sub-habits, where each sub-habit can be regarded as a sub-task.

[0091] In step S8, the control module 110 determines whether it has obtained the user behavior habit suggestions. If it has obtained the user behavior habit suggestions, it executes step S10; otherwise, it executes step S9. That is, after the control module 110 sends the target date to the first model 120, the control module 110 also needs to determine whether the first model 120 has generated the user behavior habit suggestions corresponding to the target date. For example, if the control module 110 does not receive the user behavior habit suggestions sent by the first model 120 within a preset time period, then the control module 110 can consider that the first model 120 has not output the user behavior habit suggestions corresponding to the target date.

[0092] In step S9, the control module 110 obtains the task information input by the user and executes the task. After the control module 110 determines that the first model 120 does not output suggestions on user behavior habits, the control module 110 can output interactive information to obtain the task information that the user needs to complete. The control module 110 can input the obtained task information into the first model 120 to determine the task type to which the obtained task information belongs through the first model 120. After the control module 110 obtains the task type to which the task information input by the user belongs, based on the obtained task type, by querying the pre-stored first relationship table, it determines the first task tool that needs to be called. The first relationship table is used to represent the corresponding relationship between the task type and the task tool. The first relationship table can be stored in the control module 110 or in the information repository 130. After the control module 110 determines the first task tool that needs to be called, it calls the first task tool based on the task information input by the user to perform task processing and outputs the processing result.

[0093] In a possible example, the task tools associated with the control module 110 may include: one or more of a general large model, an industry-specific large model, an Agent, and an application APP.

[0094] In some embodiments, if the control module 110 is not connected to the first task tool, or an abnormality occurs in the connection between the control module and the first task tool, resulting in the control module 110 being unable to call the first task tool, the control module 110 can process the task by calling the general large model based on the task information input by the user and output the processing result. If the control module 110 is not connected to the large model, or an abnormality occurs in the connection between the control module 110 and the large model, the control module 110 can process the task by calling the first model 120 based on the task information input by the user and output the processing result.

[0095] Next, a specific example is used to illustrate the task processing process of the control module 110. As Figure 4 shown, when the control module 110 does not obtain suggestions on user behavior habits, the control module 110 conducts new information interaction with the user in the interactive box and obtains the task information input by the user as "Help me order a cup of latte coffee". The control module 110 sends the obtained task information to the first model 120 to determine, through the first model 120, that the task type corresponding to this task information is the takeaway service. After the control module 110 obtains that the task type to which the task information input by the user belongs is the takeaway service, by searching the first relationship table, it determines that the task tool corresponding to the takeaway service is Application 1 (APP1). Then, the control module 110 can call APP1 to place an order and display the ordering result in the interactive box.

[0096] In a possible example, as Figure 4 shown, the first relationship table contains the corresponding relationships between multiple task types and task processing tools. The task types and task processing tools can be stored in the form of key-value pairs. For example, the task type is used as the key, and the task processing tool is used as the value. It can be understood that in the key-value data structure, the key is usually unique, but the value can be of any type, including lists, sets, dictionaries, etc. Therefore, one key can correspond to multiple values. For example, in the first relationship table, the document service can correspond to multiple document processing APPs (APP2, APP3).

[0097] In step S10, it is determined whether to execute the user behavior habit suggestion. If so, step S11 is executed; otherwise, step S9 is executed. When the control module 110 obtains the user behavior habit suggestion, the control module 110 can output a prompt message in the interaction box to prompt the user whether to execute the user behavior habit suggestion. When the user refuses to execute the user behavior habit suggestion, the control module 110 can prompt the user to input the task information to be executed, and based on the task information input by the user, execute the relevant task.

[0098] In a possible example, when the behavior habit suggestion corresponding to the target date for the user contains multiple sub-habits, the user can select some of the sub-habits from the multiple sub-habits to execute. When the user selects at least one sub-habit in the behavior habit suggestion corresponding to the target date to execute, it can be considered that the user selects to execute the behavior habit suggestion corresponding to the target date, that is, step S11 needs to be executed; otherwise, step S9 is executed.

[0099] In a possible example, when the behavior habit suggestion corresponding to the target date for the user contains multiple sub-habits, only when the user selects to execute all the sub-habits is it considered that the user selects to execute the behavior habit suggestion corresponding to the target date, that is, step S11 is executed. When the user selects not to execute at least one sub-habit in the behavior habit suggestion corresponding to the target date, step S9 is executed. Among them, during the execution of step S9, the control module 110 also needs to execute the remaining sub-habits that the user selects to execute in the behavior habit suggestion.

[0100] In a possible example, when there are multiple sub-habits included in the behavior habit suggestion corresponding to the target date, and the number of sub-habits that the user chooses not to perform is greater than or equal to a preset threshold (for example, the first threshold), it can be considered that the user chooses not to perform the behavior habit suggestion corresponding to the target date, that is, step S9 is executed. During the execution of step S9, the control module 110 also needs to execute the remaining sub-habits that the user chooses to perform in the behavior habit suggestion. When the number of sub-habits that the user chooses not to perform is less than the preset threshold, it can be considered that the user executes the behavior habit suggestion corresponding to the target date, that is, step S11 is executed.

[0101] In step S11, the user behavior habit suggestion is executed. In the case where the user determines that the obtained user behavior habit suggestion needs to be executed, the control module 110 can call the first model 120 to decompose the task corresponding to the user behavior habit suggestion, obtaining at least one task and the task type corresponding to the task. For example, if the user behavior habit suggestion is "play badminton", the first model 120 decomposes "play badminton" and can obtain that playing badminton can consist of two tasks. Task 1 is "book a badminton court ticket", and task 2 is "buy water". Among them, the category corresponding to task 1 is "leisure and entertainment category", and the category corresponding to task 2 is "takeaway service category".

[0102] After the control module 110 obtains the tasks and task types corresponding to the user behavior habit suggestion, it can determine the task tools that each task needs to call based on the obtained task types. Then, the control module 110 calls the task tools corresponding to each task to execute the tasks and displays the task execution results in the interactive dialog box so that the user can modify or confirm the task execution results. The process of the control module 110 calling the task tools to execute the tasks can refer to step S9 and will not be elaborated here.

[0103] It can be understood that during the operation stage of the digital avatar system, the digital avatar system can interact with the user and actively help the user complete some work in the digital world through steps S5 - S11. Among them, during the operation stage of the digital avatar system, the digital avatar system can also remind the user to summarize the work of a day or a week. For example, the digital avatar system can remind the user to summarize at a fixed time period, and its implementation process can refer to steps S12 - S15. There is no execution sequence between steps S12 - S15 and steps S5 - S11. For example, steps S12 - S15 can occur between steps S5 - S11.

[0104] In step S12, the control module 110 determines whether it is in the first time period. If so, step S13 is executed. The control module 110 periodically obtains the time information displayed on the terminal device to determine whether the currently obtained time is in the first time period. For example, the control module can obtain the time information displayed on the terminal device every 20 minutes. The first time period can be from 17:30 to 19:00 every day.

[0105] In step S13, the control module 110 outputs a prompt message to prompt the user to input a diary or a weekly journal. When the currently obtained time is in the first time period, the control module 110 can output a prompt message in the interaction box to prompt the user to perform a daily summary or a weekly summary. Exemplarily, a daily summary can be a summary of the user's work for one day. For example, what work has been done and what work has not been completed, etc.; a weekly summary can be a summary of the work for one week.

[0106] After the user receives the prompt message output by the control module 110, the user can input feedback information in the input box. Among them, the feedback information can be a diary or a weekly journal. When the user does not need to perform a daily summary or a weekly summary, the feedback information input by the user can be used to indicate that the user does not need to perform a summary.

[0107] In some embodiments, if the feedback information input by the user is used to indicate that the user does not need to perform a summary, the user can directly input the feedback information in the interaction dialog box. When the user needs to perform a daily summary or a weekly summary, the user can choose to upload the summary content in the form of a document to the interaction dialog box, or directly input the summary content in the dialog box.

[0108] In step S14, the control module 110 determines whether it has received the feedback information input by the user. If it has received the feedback information input by the user, step S15 is executed; otherwise, step S13 is executed. After the control module 110 determines that it has received the feedback information input by the user, the control module 110 can stop outputting the prompt message.

[0109] In step S15, the control module 110 saves the feedback information input by the user. After the control module 110 receives the feedback information input by the user, the control module 110 can save the feedback information input by the user to the information library 130.

[0110] It can be understood that during the operation stage of the digital avatar system, the information (including text information and file information) that the digital avatar system interacts with the user, and the information that the digital avatar system interacts with the task processing need to be stored in the information library 130 in real time, so as to facilitate training the first model 120 in the digital avatar system during a specified time period, so that the digital avatar system can remember and learn the user's behavior.

[0111] (3) Optimization stage

[0112] During the optimization phase, the first model 120 in the digital twin system is trained to memorize and learn user behavior and update knowledge. In some examples, training of the first model 120 can be performed during the user's rest periods (e.g., from 7:00 PM to midnight and from 12:00 AM to 6:00 AM daily).

[0113] When controlling the training of the first model 120, the control module 110 primarily uses two methods. In the first method, the control module 110 determines whether to trigger the training of the first model 120 by determining whether a user is in front of the terminal device screen. This avoids training the first model 120 while the user is using the terminal device. The implementation process of the first method can be referred to steps S16-S22.

[0114] In step S16, the control module 110 determines whether the terminal device is connected to a power source. If the terminal device is connected to a power source, step S17 is executed, otherwise step S16 is continued. The control module 110 can detect whether the terminal device is connected to a power source at a fixed time interval (for example, every 10 minutes). In one possible example, the control module 110 can determine whether the terminal device is connected to a power source by monitoring the operating status signal of the terminal device, such as the speed signal of the motor, the data update signal of the sensor, etc. If the operating status signal of the terminal device changes normally, it means that the terminal device is connected to a power source and is in working condition; if the operating status signal of the terminal device is abnormal or unchanged, it means that the terminal device may not be connected to a power source.

[0115] In step S17, the control module 110 determines whether a user is in front of the terminal device screen. If so, the process proceeds to step S19; otherwise, the process proceeds to step S18. The control module 110 may use a camera on the terminal device to determine whether a user is in front of the terminal device screen. For example, the terminal device 110 may use the camera on the terminal device to take photos at regular intervals. The control module 110 then performs face detection on the photos captured by the camera. If a face is detected, it is determined that a user is in front of the terminal device screen; otherwise, it is determined that no user is in front of the terminal device screen.

[0116] In step S18, the control module 110 controls the first model 120 to be trained to learn the user's behavior and knowledge. When the control module 110 detects that there is no user in front of the terminal device screen, the control module 110 can control the first model 120 to be trained. For example, the control module 110 can control the large model to perform incremental training. Among them, when the control module 110 controls the first model 120 to perform incremental training, the control module 110 can obtain all the information stored on the current day from the information repository 130 and input the obtained information into the first model 120 to trigger the first model 120 to perform incremental training.

[0117] In step S19, the control module 110 determines whether the first model 120 is in a training state. If it is in a training state, step S20 is executed; otherwise, step S18 is executed. If the control module 110 detects that there is a user in front of the terminal device screen, it indicates that the user is still working. At this time, the digital avatar system can continue to provide services to the user and it is not suitable for training. Therefore, if the first model 120 is in a training state, the control module 110 needs to control the first model 120 to stop training.

[0118] In step S20, the control module 110 controls the first model 120 to stop training. In a possible example, when the control module 110 controls the first model 120 to stop training, the control module 110 needs to save the intermediate state of the training of the first model 120 to the information repository 130 to facilitate restoring the training state of the first model 120 when needed. That is, when the first model 120 resumes training, the first model 120 only needs to continue training from the intermediate position instead of starting the entire training process again. By saving and loading the intermediate state of the training of the first model 120, the control module 110 can pause and resume the training of the first model 120 at any time during the training process, ensuring that the training progress is not lost and can also continue training on different devices or in different environments.

[0119] In step S21, the control module 110 determines whether the time when there is no user in front of the terminal device screen is greater than a preset value. If it is greater, step S22 is executed; otherwise, step S21 is executed. After the first model 120 stops training, the control module 110 also needs to periodically detect whether there is a user in front of the terminal device screen. When the time when the control module 110 detects that there is no user in front of the terminal device screen is greater than the preset value, the control module 110 can consider that the user has left, avoiding the situation where the user only briefly leaves the terminal device.

[0120] In step S22, the control module 110 controls the first model 120 to start breakpoint training. When the control module 110 determines that there is no user in front of the terminal device screen, the control module 110 can control the first model 120 to start breakpoint training.

[0121] In the second method, the control module 110 determines whether to trigger the training of the first model 120 by judging the utilization rate of the computing resources of the terminal device (for example, the central processing unit (CPU), graphics processing unit (GPU), and memory of the terminal device), avoiding training the first model 120 when the terminal device has a high load. The implementation process of the second method can refer to steps S23 - S27.

[0122] In step S23, the control module 110 determines whether the terminal device is connected to a power source. If the terminal device is connected to a power source, step S24 is executed; otherwise, step S23 is continued. Among them, the implementation process of step S23 is the same as that of step S16 and will not be elaborated here.

[0123] In step S24, the control module 110 determines whether the utilization rate of the computing resources on the terminal device is less than the lowest threshold of the normal load range. If so, step S25 is executed; otherwise, step S26 is executed.

[0124] In step S25, the control module 110 controls the first model 120 to be trained. The control module 110 can obtain the utilization rate of the computing resources on the terminal device through the tools (such as the task manager) built into the terminal device, and determine whether to control the first model 120 to be trained according to the utilization rate of the computing resources on the terminal device. In a possible example, the normal load range of the computing resources on the terminal device can be preset to 30% - 80%. When the control module 110 detects that the utilization rate of the computing resources on the terminal device is less than 30%, it indicates that the terminal device is in an idle stage. At this time, the control module 110 can control the first model 120 to be trained. Among them, the implementation process of step S25 is the same as that of step S18 and will not be elaborated here.

[0125] In step S26, the control module 110 determines whether the utilization rate of the computing resources on the terminal device is greater than the highest threshold of the normal load range. If so, step S27 is executed; otherwise, step S24 is executed. When the control module 110 detects that the utilization rate of the computing resources of the terminal device exceeds the highest threshold of the normal load range, it means that the terminal device is in a high-load operation state. In this case, it is not suitable to train the first model 120.

[0126] In step S27, it is determined whether the large model is in a training state. If so, step S28 is executed; otherwise, step S24 is executed.

[0127] In step S28, the control module 110 controls the first model 120 to stop training. The implementation process of step S28 is the same as that of step S20 and will not be elaborated here.

[0128] In a possible example, the initialization phase, the running phase, and the optimization phase can be carried out at different time periods of a day. For example, it can be preset that the digital avatar system implements the methods of the initialization phase and the running phase during the user's active time period (for example, 6:00 to 18:00 on weekdays), and implements the method of the optimization phase during the user's rest time period (19:00 to 24:00 and 00:00 to 06:00 every day).

[0129] In the embodiments of the present application, after the digital avatar system is started, the digital avatar system can actively interact with the user and generate suggestions on the user's behavior habits, so that the digital avatar system can help the user complete some work in the digital world and improve the user's work and life efficiency. And the digital avatar system can remember and learn the user's behavior according to the interaction records with the user, improving the user's usage experience.

[0130] Next, based on the digital avatar system described above, a task management method based on digital avatar provided by the embodiments of the present application will be introduced.

[0131] Exemplarily, Figure 5 The flowchart of a digital avatar control method provided by the embodiments of the present application is shown. It can be understood that this method can be executed by any device, equipment, platform with computing and processing capabilities. Exemplarily, this method can be executed by a terminal device, where the device can be implemented by software and / or hardware. For the convenience of description, the following takes the digital avatar system running on the terminal device as an example for introduction. As Figure 5 shown, the digital avatar control method includes:

[0132] Step 501, in response to the user's first operation, output the user's digital avatar, where the digital avatar is generated based on the user's user profile.

[0133] In this embodiment, the control module starts the digital avatar corresponding to the user in response to the operation of starting the digital avatar issued by the user (i.e., the first operation). It can be understood that after the user information is imported into the digital avatar system, the digital avatar system containing the user information can be called the digital avatar corresponding to the user. Among them, the user information can include the user profile.

[0134] In a possible example, outputting the user's digital avatar includes outputting the digital avatar image of the user.

[0135] In a possible embodiment, before outputting the user's digital avatar, the user's identity can also be verified. For example, the control module can obtain second information, where the second information can be used to represent the user's identity. In a possible example, the second information can include the user's biometric information, such as a face image, iris, fingerprint, etc. The control module can obtain the user's biometric information through external devices connected to the terminal device (such as a camera, fingerprint collector, etc.). After obtaining the user's biometric information, the control module matches the obtained user's biometric information with the user's biometric information stored in the information database. If the matching is successful, it is determined that the user's identity verification is passed. In another possible example, the second information can also be credential information. For example, the account and password information for logging in to the digital avatar system. After the digital avatar system is started, the control module can output a prompt message on the display interface of the terminal device to prompt the user to enter the account and password for logging in to the digital avatar system. After obtaining the account and password entered by the user, the control module obtains the password information corresponding to the account from the information database according to the obtained account information. Then, the control module matches the password entered by the user with the password obtained from the information database. If the matching is successful, it is determined that the user's identity verification is passed.

[0136] When the user's identity authentication is passed, the control module can control the first model to generate the digital avatar image corresponding to the user and display the generated digital avatar image on the screen of the terminal device. In a possible example, when the user's identity authentication is passed, the control module can also generate an interactive dialog box, which can be used to implement the interaction between the digital avatar and the user, and the interactive dialog box can be displayed next to the digital avatar image.

[0137] Step 502: Based on the user's historical behavior data, obtain the first task corresponding to the target date. The first task is used to represent the user's task plan on the target date. The historical behavior data is used to represent the user's behavior within a certain historical period.

[0138] In this embodiment, the first task may refer to the user behavior habit suggestions mentioned above. After the digital avatar is started, the digital avatar can also obtain the target date. In a possible example, the digital avatar can obtain the target date through the terminal device, that is, the digital avatar can obtain the date information displayed on the terminal device. For example, the date information obtained by the control module from the terminal device may be XX year XX month XX day, Monday. In some possible examples, the control module can obtain the date information displayed on the terminal device through the internal clock module of the terminal device. For example, the control module can obtain the date information displayed on the terminal device through the internal clock circuit provided by the control module itself. Among them, the internal clock circuit starts timing after the control module is powered on. The control module can read the time information recorded by the internal clock circuit through specific registers or instructions and convert it into a date format. For example, for the STM32 series of single-chip microcontrollers, its internal clock system can provide a time reference, and the control module can obtain the current date and time by reading the values of relevant registers.

[0139] In a possible example, after the control module obtains the target date displayed on the terminal device, the control module can input the obtained target date into the first model so that the first model outputs the first task corresponding to the target date. Among them, the first model is trained based on the historical behavior data of the user.

[0140] Step 503, output the first information, and the first information includes the first task.

[0141] In this embodiment, the control module inputs the obtained target date into the first model, and the first model can output the first task corresponding to the target date (that is, the user behavior habit suggestions mentioned above). After the control module obtains the first task corresponding to the target date, the control module can also display the obtained first task in the interaction dialog box to facilitate the user to confirm whether the digital avatar system needs to execute the first task. Then, the control module can receive the user's processing opinion on the first task.

[0142] Step 504, in response to the user's determination to execute the operation, execute the first task.

[0143] In this embodiment, after the control module outputs the first task, the control module can receive the user's operation on the first task. Among them, the user's operation on the first task can be to execute or not to execute. The control module can execute the first task in response to the user's determination to execute the operation.

[0144] In a possible example, the first task may include multiple subtasks. In response to the user's determination to perform an operation, performing the first task includes: in response to the user's determination to perform an operation for each subtask in the first task, performing the subtasks in the first task; or, in response to the user's operation to perform at least one target subtask in the first task, performing the target subtasks in the first task. In this way, the user can select to perform some of the subtasks in the first task.

[0145] In a possible implementation, the first task may include multiple subtasks. When the number of target subtasks that the user determines not to perform among the multiple subtasks is greater than or equal to the first threshold, the user's operation can be considered as a determination not to perform the operation. At this time, in response to the user's determination not to perform the operation, the third information can be output, and the subtasks other than the target subtasks in the subtasks are performed. When the number of target subtasks that the user determines not to perform among the multiple subtasks is less than the first threshold, the user's operation can be considered as a determination to perform the operation. At this time, in response to the user's determination to perform the operation, the subtasks other than the target subtasks in the first task are performed. In a possible example, when the control module executes the first task, the control module can call the first model to decompose the first task to obtain at least one task and the task type corresponding to the task. Then, the control module can call the corresponding task tool according to the task type corresponding to each task to execute the task. The process of the control module executing the first task can refer to step S11 and will not be elaborated here. In a possible example, when the user refuses to execute the first task, the user can output the third information in the interactive dialog box to obtain the task that the user needs to execute. The first task includes multiple subtasks; in response to the determination not to perform the operation input by the user, outputting the third information includes: in response to the user's determination not to perform the operation for each subtask in the first task, outputting the third information, or, in response to the user's determination not to perform at least one target subtask in the first task, outputting the third information and performing the subtasks other than the target subtasks in the first task. The specific implementation process of the control performing the task according to the task information input by the user can refer to step S9 and will not be elaborated here.

[0146] In a possible example, when the control module inputs the target date into the first model, but the first model does not output the user behavior habit suggestion corresponding to the target date, the control module can output interactive information in the interactive dialog box to obtain the task that the user needs to execute. The specific implementation process of the control module performing the task according to the task information input by the user can refer to step S9 and will not be elaborated here.

[0147] In a possible example, after outputting the user's digital avatar in response to the first operation issued by the user, the control module can also periodically obtain the time displayed on the terminal device. When the time displayed on the terminal device is within the first time period of the target date, the control module can generate the fourth information. The fourth information is used to prompt the user to input the summary information corresponding to the target date, where the summary information is used to describe the user's behavior habits on the target date. When the time displayed on the terminal device is within the second time period of the target date, the control module can control the first model to be trained, where the first time period is earlier than the second time period.

[0148] In a possible example, when the control module controls the first model to be trained, the control module first determines whether there is a user on the terminal device. When it is determined that there is no user in front of the terminal device screen, the control module can control the first model to be trained. When the control module determines that there is a user in front of the terminal device screen and the first model is in a training state, the control module can control the first model to stop training. When there is a user in front of the terminal device screen, it indicates that the digital avatar may be serving the user. Therefore, in order not to affect the user experience, the control module will trigger the first model to be trained only when it detects that there is no user in front of the terminal device screen, and the control module can trigger the first model to stop training when it detects that there is a user in front of the terminal device screen.

[0149] In another possible example, the control module can determine whether to control the first model to be trained according to the computing resource utilization rate of the terminal device. For example, when the computing resource utilization rate of the terminal device is less than the first threshold, the control module controls the first model to be trained. When the computing resource utilization rate of the terminal device is greater than the second threshold and the first model is in a training state, the control module controls the first model to stop training, and the second threshold is greater than the first threshold. The utilization rate of the computing resources on the terminal device being less than the first threshold indicates that the terminal device is in an idle stage, and the control module can control the first model to be trained. When the computing resources on the terminal device are greater than the second threshold, it indicates that the terminal device is in a busy stage at this time, and the control module can control the first model to stop training to reduce the computing load of the terminal device.

[0150] In a possible example, when the control module controls the first model to be trained, it can use the information stored in the information library for the target date (i.e., the fifth information) as training data to train the first model, so that the first model can remember and learn the user's behavior. Among them, the data stored in the information library for the target date can include one or more of the following: the interaction information between the digital avatar and the user, the summary information corresponding to the target date, and the interaction information between the digital avatar and the task tool.

[0151] In a possible example, when the first model is being trained, the first model can perform incremental training or breakpoint training.

[0152] In the embodiments of the present application, after the digital avatar is started, the digital avatar can actively interact with the user and generate suggestions for the user's behavior habits, so that the digital avatar can help the user complete some work in the digital world, improving the user's work and life efficiency. In addition, the digital avatar can remember and learn the user's behavior based on the interaction records with the user, improving the user's usage experience.

[0153] It can be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not mean the order of execution. The execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application. In addition, the various embodiments described above and the technical features in each embodiment can be combined according to the actual situation, and the combined solution is still within the protection scope of the present application.

[0154] Exemplarily, the embodiments of the present application also provide a task management device. As Figure 6 shown, the task management device 600 includes: an interaction module 610, a processing module 620, and a first model 630. Among them, the processing module 620 outputs the user's digital avatar in response to the user's first operation, and the digital avatar is generated based on the user's user profile; the processing module 620 is further configured to obtain a first task corresponding to the target date based on the user's historical behavior data, and the first task is used to represent the user's task plan on the target date; the interaction module 610 is configured to output a first message, and the first message includes the first task; the processing module 620 is further configured to execute the first task in response to the user's determined execution operation.

[0155] In some embodiments, before the processing module 620 outputs the user's digital avatar, the interaction module 610 is further configured to obtain a second message, and the second message is used to represent the user's identity; the processing module 620 is further configured to authenticate the user based on the second message.

[0156] In some embodiments, the processing module 620 obtains the first task corresponding to the target date based on the user's historical behavior data, including: inputting the target date into the first model 630 to obtain the first task, and the first model 630 is trained based on the user's historical behavior data.

[0157] In some embodiments, the processing module 620 executes the first task in response to the user's determined execution operation, including: inputting the first task into the first model 630 to obtain at least one task to be executed and the task type corresponding to the task to be executed; based on the task type corresponding to the task to be executed, calling a task tool corresponding to the task type to execute the task to be executed.

[0158] In some embodiments, after the interaction module 610 outputs the first information, the interaction module 610 is further configured to output third information in response to a user input of determining not to perform an operation, where the third information is used to prompt the user to input a second task, and the second task is a to-be-executed task corresponding to the target date for the user.

[0159] In some embodiments, in the first time period of the target date, the processing module 620 is further configured to generate fourth information, where the fourth information is used to prompt the user to input summary information corresponding to the target date, and the summary information is used to describe the user's behavior habits on the target date; in the second time period of the target date, the processing module 620 is further configured to control the first model 630 to be trained, and the first time period is earlier than the second time period.

[0160] In some embodiments, in the second time period of the target date, the processing module 620 controls the first model 630 to be trained, including: determining whether there is a user in front of the screen of the terminal device, where a digital avatar runs on the terminal device; in the case that there is no user in front of the screen of the device, controlling the first model 630 to be trained; in the case that it is determined that there is a user in front of the screen of the device and the first model 630 is in a training state, controlling the first model 630 to stop training.

[0161] In some embodiments, in the second time period of the target date, the processing module 620 controls the first model 630 to be trained, including: obtaining the computing resource utilization rate of the device; in the case that the computing resource utilization rate is less than a first threshold, controlling the first model 630 to be trained; in the case that the computing resource utilization rate is greater than a second threshold and the first model 630 is in a training state, controlling the first model 630 to stop training, and the second threshold is greater than the first threshold.

[0162] In some embodiments, the processing module 620 controls the first model 630 to be trained, including: obtaining fifth information, where the fifth information is information generated by the digital avatar interacting with the user on the target date, and the fifth information includes one or more of the interaction information between the digital avatar and the user, the summary information corresponding to the target date, and the interaction information between the digital avatar and the task tool; training the first model 630 based on the fifth information.

[0163] Based on the method in the above embodiments, an embodiment of the present application provides a computer-readable storage medium storing a computer program, and when the computer program runs on a processor, the processor is caused to execute the method in the above embodiments.

[0164] Based on the method in the above embodiments, an embodiment of the present application provides a computer program product, characterized in that when the computer program product runs on a processor, it causes the processor to execute the method in the above embodiments.

[0165] Based on the method in the above embodiments, an embodiment of the present application provides a terminal device, which includes a main board and a chip. Among them, the chip is integrated on the main board, and the chip includes at least one memory for storing programs; at least one processor for executing the programs stored in the memory, and when the programs stored in the memory are executed, the processor is used to execute the method in the above embodiments.

[0166] The method steps in the embodiments of the present application can be implemented in a hardware manner or by a processor executing software instructions. The software instructions can be composed of corresponding software modules, and the software modules can be stored in a random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, mobile hard disks, CD-ROMs, or any other form of storage medium well-known in the art. An exemplary storage medium is coupled to the processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an ASIC.

[0167] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center in a wired manner (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wirelessly (such as infrared, wireless, microwave, etc.). The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more integrated available media. The available medium may be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid state disk (SSD)), etc.

[0168] It can be understood that the various numerical numbers involved in the embodiments of the present application are only for the convenience of description and are not used to limit the scope of the embodiments of the present application.

Claims

1. A task management method based on digital avatars, characterized in that, Including: In response to a first operation of the user, output a digital avatar of the user, which is generated based on the user profile of the user; Based on the historical behavior data of the user, obtain a first task corresponding to a target date, where the first task is used to represent the task plan of the user on the target date; Output first information, where the first information includes the first task; In response to the user's determination to execute the operation, execute the first task.

2. The method according to claim 1, characterized in that, Before outputting the digital avatar of the user, it further includes: Obtain second information, where the second information is used to represent the user identity; Authenticate the user based on the second information.

3. The method according to claim 1 or 2, characterized in that, The obtaining the first task corresponding to the target date based on the historical behavior data of the user includes: Input the target date into a first model to obtain a first task, where the first model is trained based on the historical behavior data of the user.

4. The method according to claim 1, characterized in that, The responding to the user's determination to execute the operation and executing the first task includes: Input the first task into the first model to obtain at least one task to be executed and the task type corresponding to the task to be executed; Based on the task type corresponding to the task to be executed, call a task tool corresponding to the task type to execute the task to be executed.

5. The method according to claim 1, wherein After outputting the first information, it further includes: In response to the user's input of a determination not to execute the operation, output third information, where the third information is used to prompt the user to input a second task, and the second task is the task to be executed corresponding to the user on the target date.

6. The method according to any one of claims 1-5, characterized in that, The method further includes: In a first time period of the target date, generate fourth information, where the fourth information is used to prompt the user to input summary information corresponding to the target date, and the summary information is used to describe the user's behavior habits on the target date; In a second time period of the target date, control the first model to be trained, where the first time period is earlier than the second time period.

7. The method according to claim 6, wherein The controlling the first model to be trained in the second time period of the target date includes: Determine whether there is a user in front of the screen of the terminal device, where the digital avatar runs on the terminal device; When there is no user in front of the screen of the device, control the first model to be trained; When it is determined that there is a user in front of the screen of the device and the first model is in a training state, control the first model to stop training.

8. The method according to claim 6, wherein The controlling the first model to be trained in the second time period of the target date includes: Obtain the computing resource utilization rate of the device; When the computing resource utilization rate is less than a first threshold, control the first model to be trained; When the computing resource utilization rate is greater than a second threshold and the first model is in a training state, control the first model to stop training, where the second threshold is greater than the first threshold.

9. The method according to any one of claims 7 or 8, characterized in that The controlling the first model to be trained includes: Obtain the fifth information, where the fifth information is the information generated by the digital avatar interacting with the user on the target date, and the fifth information includes one or more of the following: the interaction information between the digital avatar and the user, the summary information corresponding to the target date, and the interaction information between the digital avatar and the task tool; Train the first model based on the fifth information.

10. A terminal device, characterized in that, The terminal device includes: At least one memory for storing programs; At least one processor for executing the programs stored in the memory. When the programs stored in the memory are executed, the processor is used to execute the method according to any one of claims 1-9.

Images