Bullet screen processing method and device, electronic equipment and storage medium

Abstract

The invention provides a bullet screen processing method and device, electronic equipment and a storage medium. The method comprises the following steps: displaying an application interface of a simulator application in an equipment display interface; in response to a trigger operation on the bullet screen control, displaying an interactive bullet screen in the device display interface; according to the embodiment of the invention, the information transmission efficiency of carrying out human-computer interaction by utilizing the bullet screen in a cross-system collaborative interaction scene can be improved.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, and in particular to a method, apparatus, electronic device, and storage medium for processing bullet comments. Background Technology

[0002] With the increasing demand for collaborative operation of computer systems, cross-system collaborative interaction scenarios have become an important direction for technological development. In cross-system collaborative interaction scenarios, two or more operating systems can exchange data or coordinate control of applications or system services across system boundaries, thereby achieving interoperability in different operating systems. When an object in one operating system controls media data displayed in another operating system, the interface can display bullet comments related to the media data to achieve human-computer interaction, thus providing the object with an immersive experience.

[0003] However, among related technologies, the information transmission efficiency of using bullet comments for human-computer interaction in cross-system collaborative interaction scenarios is relatively low. Summary of the Invention

[0004] This disclosure provides a method, apparatus, electronic device, and storage medium for processing bullet comments, which can improve the information transmission efficiency of human-computer interaction using bullet comments in cross-system collaborative interaction scenarios.

[0005] According to one aspect of this disclosure, a method for processing bullet comments is provided, comprising: The device display interface shows the application interface of the simulator application, which displays the running screen of the target application and the bullet screen control control. The target application runs in the virtual operating system provided by the simulator application. In response to a trigger operation on the bullet screen control control, interactive bullet screens are displayed on the device display interface; The content of the interactive bullet comments is generated based on the analysis results obtained from the sentiment analysis of the running screen.

[0006] According to one aspect of this disclosure, a bullet screen processing device is provided, comprising: The first display unit is used to display the application interface of the simulator application on the device display interface. The application interface displays the running screen of the target application and the bullet screen control control. The target application runs in the virtual operating system provided by the simulator application. The second display unit is used to display interactive bullet comments on the device display interface in response to the triggering operation of the bullet comment control control. The content of the interactive bullet comments is generated based on the analysis results obtained from the sentiment analysis of the running screen.

[0007] Optionally, in one embodiment, the bullet screen processing device further includes: The third display unit is used to respond to the triggering operation of the interactive bullet screen by displaying a list of interactive options corresponding to the interactive bullet screen on the device display interface. The list of interactive options includes multiple interactive controls. The fourth display unit is used to display the bullet screen interaction effect corresponding to the target interactive control in response to the trigger operation of the target interactive control in the interactive option list.

[0008] Optionally, in one implementation, the list of interactive options includes a bullet screen like control; The fourth display unit is specifically used for: In response to the triggering operation of the bullet comment like control, the interactive bullet comment is displayed with a first preset display effect.

[0009] Optionally, in one implementation, the list of interactive options includes a bullet screen blocking control; The fourth display unit is specifically used for: In response to the triggering operation of the bullet screen blocking control, the interactive bullet screen is canceled from being displayed on the device display interface.

[0010] Optionally, in one implementation, the application interface further includes a bullet screen setting control; The bullet screen processing device also includes: The fifth display unit is used to display a list of bullet screen setting options in response to a trigger operation of the bullet screen setting control. The bullet screen setting option list includes multiple bullet screen effect setting controls. The sixth display unit is used to respond to the triggering operation of the target bullet screen effect setting control in the bullet screen setting option list, and to display the interactive bullet screen with the display effect corresponding to the target bullet screen effect setting control.

[0011] Optionally, in one implementation, the application interface further includes a recommended content control control; The bullet screen processing device also includes: The seventh display unit is used to display recommended content in the application interface in response to the triggering operation of the recommended content control control, wherein the recommended content is generated based on the running screen and the interactive bullet screen.

[0012] Optionally, in one implementation, the process of generating the bullet screen content corresponding to the interactive bullet screen includes: The system acquires the running screen within a predetermined time period and performs feature extraction based on the running screen to generate multimodal features. A preset neural network model is invoked to perform sentiment analysis based on the multimodal features to obtain the sentiment type; The interactive bullet screen content is generated based on the emotion type.

[0013] Optionally, in one embodiment, acquiring the running screen within a predetermined time period and generating multimodal features based on the running screen includes: The system acquires the running screen within a predetermined time period, and acquires screen scene information, screen object information, screen text information, and screen change information based on the running screen. Feature extraction is performed on the scene information, object information, text information, and change information to generate scene features, object features, text features, and change features.

[0014] Optionally, in one implementation, generating the bullet screen content corresponding to the interactive bullet screen based on the emotion type includes: Based on the multimodal features, the scene features corresponding to the running screen are extracted; The bullet screen content corresponding to the interactive bullet screen is generated based on the scene features and the emotion type.

[0015] Optionally, in one implementation, generating the bullet screen content corresponding to the interactive bullet screen based on the scene context features and the emotion type includes: Based on the aforementioned scene scene features, scene scene classification is performed to generate scene categories; The bullet screen content corresponding to the interactive bullet screen is generated based on the scenario category and the emotion type.

[0016] Optionally, in one implementation, generating the bullet screen content corresponding to the interactive bullet screen based on the scenario category and the emotion type includes: When the scenario category is the first scenario category, the bullet screen content database is obtained, and the bullet screen content corresponding to the interactive bullet screen is obtained from the bullet screen content database based on the emotion type; When the scenario category is the second scenario category, a preset large language generation model is invoked to generate the corresponding bullet screen content based on the emotion type.

[0017] Optionally, in one implementation, the training process of the preset large language generation model includes: Retrieve historical bullet comment data posted by the target object; Obtain the initial large language generation model, and update the parameters of the initial large language generation model based on the historical bullet screen data to generate the preset large language generation model.

[0018] Optionally, in one implementation, after obtaining the historical bullet screen data posted by the target object, the method further includes: Obtain the associated historical bullet screen data published by related objects related to the target object; The process of obtaining an initial large language generation model and updating the parameters of the initial large language generation model based on the historical bullet screen data to generate a preset large language generation model includes: Obtain the initial large language generation model, and update the parameters of the initial large language generation model based on the historical bullet screen data and the associated historical bullet screen data to generate the preset large language generation model.

[0019] Optionally, in one implementation, the training process of the preset large language generation model includes: Obtain the target object's feedback information on the interactive bullet comments; The parameters of the preset large language generation model are updated based on the interactive bullet comments and the feedback information.

[0020] Optionally, in one implementation, the interactive bullet screen display process includes: A second preset display effect is obtained based on the emotion type of the interactive bullet comments; The interactive bullet screen is rendered on the device display interface based on the second preset display effect.

[0021] Optionally, in one embodiment, the bullet screen display process includes: Obtain a third preset display effect, and add a bullet screen overlay layer to the device display interface based on the third preset display effect; The interactive bullet comments are rendered in the bullet comment overlay layer.

[0022] In the bullet screen processing method of this embodiment, the application interface of the simulator application is displayed on the device display interface. The application interface displays the running screen of the target application and the bullet screen control control. The target application runs in the virtual operating system provided by the simulator application. In response to the trigger operation of the bullet screen control control, interactive bullet screens are displayed on the device display interface. The bullet screen content corresponding to the interactive bullet screen is generated based on the analysis results obtained by performing sentiment analysis on the running screen.

[0023] Therefore, in this embodiment of the disclosure, cross-system collaborative interaction between the current device's operating system and the virtual operating system is achieved through an emulator application. When the target application's running screen is displayed in the emulator application's interface on the device's display screen, triggering the bullet screen control allows interactive bullet screens generated through sentiment analysis of the running screen to be displayed on the device's display screen. The bullet screens can be displayed more prominently on the device's display screen, rather than just on the target application's running screen. This increases the display range of the bullet screens, and the information conveyed by the bullet screens can be displayed more prominently on the interface, greatly improving the information transmission efficiency of human-computer interaction using bullet screens in cross-system collaborative interaction scenarios.

[0024] Other features and advantages of this disclosure will be set forth in the following description and will be apparent in part from the description or may be learned by practicing the disclosure. The objectives and other advantages of this disclosure may be realized and obtained by means of the structures particularly pointed out in the description, claims and drawings. Attached Figure Description

[0025] The accompanying drawings are provided to further understand the technical solutions of this disclosure and constitute a part of the specification. They are used together with the embodiments of this disclosure to explain the technical solutions of this disclosure and do not constitute a limitation on the technical solutions of this disclosure.

[0026] Figure 1 This is a system architecture diagram of the bullet screen processing method applied according to an embodiment of the present disclosure; Figure 2A This is a schematic diagram illustrating an embodiment of the present disclosure applied to a cross-system video viewing scenario; Figure 2B This is another schematic diagram illustrating an embodiment of the present disclosure applied to a cross-system video viewing scenario; Figure 2C This is another schematic diagram illustrating an embodiment of the present disclosure applied to a cross-system video viewing scenario; Figure 2D This is another schematic diagram illustrating an embodiment of the present disclosure applied to a cross-system video viewing scenario; Figure 2E This is another schematic diagram illustrating an embodiment of the present disclosure applied to a cross-system video viewing scenario; Figure 3 This is a flowchart of a bullet screen processing method according to an embodiment of the present disclosure; Figure 4 This is a schematic diagram showing the simulator settings bar according to one embodiment of the present disclosure; Figure 5 This is a schematic diagram showing an interactive option list according to one embodiment of the present disclosure; Figure 6AThis is a schematic diagram of displaying a bullet comment like control in an interactive option list according to one embodiment of the present disclosure; Figure 6B This is a schematic diagram of displaying interactive bullet comments after a "like" according to one embodiment of the present disclosure; Figure 7A This is a schematic diagram of displaying a bullet screen blocking control in an interactive options list according to one embodiment of the present disclosure; Figure 7B This is a schematic diagram illustrating the cancellation of interactive bullet comments according to one embodiment of this disclosure; Figure 8A This is a schematic diagram showing the bullet screen settings control according to one embodiment of the present disclosure; Figure 8B This is a schematic diagram showing a list of bullet screen setting options according to one embodiment of the present disclosure; Figure 8C This is a schematic diagram illustrating the display of interactive bullet comments after bullet comment settings, according to one embodiment of the present disclosure; Figure 9A This is a schematic diagram of a control for displaying recommended content according to one embodiment of the present disclosure; Figure 9B This is a schematic diagram showing recommended content according to one embodiment of the present disclosure; Figure 10 This is another flowchart of a bullet screen processing method according to an embodiment of the present disclosure; Figure 11 This is a flowchart illustrating a bullet screen rendering scene according to an embodiment of the present disclosure; Figure 12 This is a schematic diagram of the structure of a bullet screen processing device according to an embodiment of the present disclosure; Figure 13 This is a terminal structure diagram for implementing various methods according to an embodiment of the present disclosure; Figure 14 This is a server structure diagram illustrating the implementation of various methods according to an embodiment of the present disclosure. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this disclosure clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and are not intended to limit the scope of this disclosure.

[0028] Bullet comments: Commentary text that is superimposed on the screen when media content is running, using scrolling, hovering, or other methods.

[0029] Neural network models are a type of model that simulates the human brain's nervous system. They are complex network systems formed by the extensive interconnection of numerous simple processing units (called neurons), exhibiting characteristics of highly complex nonlinear dynamic learning systems.

[0030] Optical Character Recognition (OCR) is an automated technology that converts text information in images into computer-editable text. Through advanced image processing and pattern recognition algorithms, this technology can quickly and accurately recognize various printed and handwritten characters, playing a vital role in multiple fields.

[0031] Large language generation models are generative language models based on deep learning and neural networks, pre-trained on massive amounts of text, and capable of generating human-readable natural language text based on structured or unstructured inputs (such as data, knowledge, context, and prompts).

[0032] Batch rendering: A technique that submits multiple mergeable rendering requests to the image rendering module at once or in as few as possible. Its purpose is to reduce the number of rendering command round trips and rendering state switching, thereby reducing rendering resource overhead.

[0033] Dirty rectangle update: a partial redraw technique that only redraws the areas on the screen that have changed, rather than redrawing the entire screen, thereby reducing pixel fill, draw calls, and bandwidth, and improving frame rate and energy efficiency. Its core is to maintain a set of rectangles (dirty areas) representing areas that "need to be redrawn," and only submit these areas to the display system when submitting a frame.

[0034] Reinforcement learning is a machine learning method. Its fundamental framework is the Markov decision process, which allows an agent to learn optimal policies through trial and error in its interactions with the environment. The agent performs actions in the environment and receives feedback, or rewards, based on the outcomes of those actions. These reward signals guide the agent to adjust its policy to maximize long-term cumulative rewards.

[0035] In related technologies, when cross-system collaborative interaction is achieved through emulator applications, if it is desired to display bullet comments corresponding to the running screen of the target application in the virtual operating system, the bullet comments can only be displayed on the running screen. However, the visual space of the running screen is relatively small compared to the device's display interface. Therefore, the display range of the bullet comments is also small, and the target may not be able to obtain the accurate information conveyed by the bullet comments, resulting in a poor interactive experience and low information transmission efficiency for human-computer interaction using bullet comments. To address this, this disclosure provides a bullet comment processing method to improve the information transmission efficiency of human-computer interaction using bullet comments in cross-system collaborative interaction scenarios.

[0036] System architecture and scenario description of the embodiments disclosed herein Figure 1This is a system architecture diagram of a bullet screen processing method according to an embodiment of the present disclosure. It includes a terminal 140, an Internet 130, a gateway 120, a server 110, etc.

[0037] Terminal 140 can take various forms, including desktop computers, laptops, PDAs (personal digital assistants), mobile phones, vehicle terminals, dedicated terminals, tablets, smart TVs, service terminals, and robot display terminals. Furthermore, it can be a single device or a collection of multiple devices. For example, multiple devices can be connected via a local area network, sharing a single display device to work collaboratively, forming a single terminal 140. Terminal 140 can also communicate with the Internet 130 via wired or wireless means to exchange data.

[0038] Server 110 refers to a computer system that can provide certain services to terminal 140. Compared to ordinary terminal 140, server 110 has higher requirements in terms of stability, security, and performance. Server 110 can be a single high-performance computer in a network platform, a cluster of multiple high-performance computers, a portion of a single high-performance computer (e.g., a machine), or a combination of portions of multiple high-performance computers (e.g., machines).

[0039] Gateway 120, also known as an internetwork connector or protocol converter, is a computer system or device that acts as a translator between two systems using different communication protocols, data formats, languages, or even completely different architectures. It enables network interconnection at the transport layer. Gateways can also provide filtering and security functions. Messages sent from terminal 140 to server 110 are forwarded to the corresponding server 110 through gateway 120. Messages sent from server 110 to terminal 140 are also forwarded to the corresponding terminal 140 through gateway 120.

[0040] The embodiments disclosed herein can be applied in multiple scenarios, such as Figures 2A-2E The examples shown include cross-system video viewing scenarios.

[0041] like Figure 2A As shown, the object triggers the emulator application open control 211 in the device display interface 210 of the currently used operating system, thus opening the emulator application. The emulator application can be used to control applications and data in other virtual operating systems within the current operating system. For example... Figure 2B As shown, the system selection list 212 in the emulator application displays connectable virtual systems, including virtual system X1, virtual system X2, and virtual system X3. Select virtual system X1 to connect.

[0042] After connecting to the virtual system X1, as follows Figure 2CAs shown, the application interface 220 of the emulator application is opened, which displays the screen in the virtual system X1, including the application opening control 221 corresponding to the application A1 that can run in the virtual system X1.

[0043] After triggering the application to open control 221, as follows Figure 2D As shown, application A1 is opened, displaying its running screen 222 and a bullet screen control 223. After triggering the bullet screen control 223, interactive bullet screens can be generated for the content displayed in the running screen 222 and displayed on the device display interface 210. Figure 2E As shown, the interactive bullet comments displayed on the device display interface 210 include: "Tourist destination", "Beautiful scenery", "Very beautiful buildings" and "Great, great".

[0044] Therefore, it can be seen that after applying the bullet screen processing method of this disclosure embodiment, the interactive bullet screen generated for the running screen of the target application in the virtual system can ignore the boundaries between operating systems and be displayed on the device display interface. Thus, the user can obtain richer bullet screen information through a wider bullet screen display range, improving the efficiency of bullet screen information transmission, and thereby enabling the user to have a more immersive experience in cross-system video viewing scenarios.

[0045] General Description of Embodiments in this Disclosure According to one embodiment of this disclosure, a method for processing bullet comments is provided. The bullet comment processing method provided in this embodiment can be applied to, for example... Figures 2A-2E This is illustrated in scenarios such as watching videos across different systems. For example... Figure 3 The diagram shown is a flowchart of a bullet screen processing method provided in this disclosure, which can be executed by a terminal. The bullet screen processing method may include: Step 310: Display the application interface of the simulator application in the device display interface.

[0046] The device display interface can be the display interface of the device currently being used by the object. The device display interface can also be the display interface of the aforementioned terminal.

[0047] An emulator application can be an application that runs on the device's operating system and is used to create a virtual operating environment within the current operating system. Within this virtual operating environment, the operating system of another device can run. An emulator application can provide a range of tools and interfaces for interacting with other devices or operating systems. The application interface can be the screen displayed by the emulator application. For example... Figure 2C The application interface 220 shown is the application interface corresponding to the emulator application.

[0048] In addition to displaying the emulator application's interface, the device's display interface can also show the local application interface of applications currently running in the operating system. Thus, an object can simultaneously view the local applications and the applications connected to the virtual operating environment through the device's display interface.

[0049] The application interface displays the running screen of the target application and the bullet screen control controls. The target application runs in the virtual operating system provided by the emulator application.

[0050] A virtual operating system can be another operating system running within a virtual operating environment provided by an emulator application. Within this virtual operating system, applications and data on other devices' operating systems can be controlled. For example, if the current device is a desktop computer running Windows, and the virtual operating system connected via the emulator application is an Android system running on a mobile device, then the emulator application can control the mobile device's operating system from the desktop computer's display interface. This allows users to view media content on the mobile device or play games using the mobile device's operating system.

[0051] In one implementation, connecting to a virtual operating system via an emulator application utilizes a virtual disk image (VDI) file corresponding to the virtual operating system, which stores all the data within the virtual operating system. After opening the emulator application, the user selects the virtual operating system they wish to connect to, for example... Figure 2B As shown, select virtual system X1 from among multiple virtual operating systems. The emulator application can load and run the VDI file corresponding to this virtual operating system, thereby opening the virtual operating system within the emulator application.

[0052] The target application can be an application running within a virtual operating system. For example, a video application or a game application within a virtual operating system. The running screen can be the screen currently being played in the target application, for example... Figure 2D The running screen 222 in the text is the running screen corresponding to application A1.

[0053] The bullet screen control is a control in the emulator application's interface used to control whether bullet screens are turned on or off. When the bullet screen control is triggered, the bullet screen is turned on; when the bullet screen control is triggered again, the bullet screen is turned off.

[0054] In one implementation, the bullet screen control is displayed independently in the emulator application's interface, such as... Figure 2D As shown, the bullet screen control 223 is displayed in the lower right corner of the application interface 220.

[0055] In another implementation, the application interface includes a simulator settings bar, which includes bullet screen control controls.

[0056] The emulator settings panel includes various functional controls for configuring the emulator, including controls for controlling the bullet screen (or comment section). The emulator settings panel can be located at the top of the emulator application's interface or on the sidebar. The emulator sidebar can also be hidden and displayed by default controls within the emulator application's interface. For example... Figure 4 As shown, the application interface 210 displays the running screen 222 and the emulator settings bar 410. The emulator settings bar 410 is located on the side of the application interface 210 and includes: emulator volume setting control 411, emulator screen recording control 412, emulator screen setting control 413, and bullet screen control control 414.

[0057] Step 320: In response to the trigger operation of the bullet screen control, display the interactive bullet screen on the device display interface.

[0058] The content of interactive bullet comments is generated based on the analysis results obtained from sentiment analysis of the running screen.

[0059] In one implementation, the process of generating the content corresponding to the interactive bullet comments includes: The system acquires the running screen within a predetermined time period and extracts features based on the running screen to generate multimodal features. A pre-defined neural network model is invoked to perform sentiment analysis based on multimodal features to obtain the sentiment type; The interactive bullet comments are generated based on the emotion type.

[0060] The scheduled time period can be a predetermined time before the current time point, such as 30 seconds, 1 minute, or 5 minutes before the current time point.

[0061] In one implementation, when an object triggers the bullet screen control, in response to a bullet screen receiving request, running frames can be acquired at a predetermined frame rate within a predetermined time period to form a sequence of running frame images. The predetermined frame rate can be the frequency at which static frames are extracted from a continuous running frame segment; for example, the predetermined frame rate is 15fps. Based on the predetermined frame rate, an image sequence consisting of multiple static running frames can be constructed.

[0062] In one implementation, acquiring runtime frames at a predetermined frame rate and constructing a sequence of runtime frame images can be configured through a frame buffer interface embedded in the emulator application. The frame buffer can be used to temporarily store runtime frames of the target application in the virtual operating system, and the frame buffer interface can directly read and write to the runtime frames. Therefore, after determining the predetermined frame rate, runtime frames can be acquired and image sequences constructed according to the predetermined frame rate through the frame buffer interface.

[0063] In one implementation, during the process of acquiring the running screen and constructing the running screen image sequence based on the frame buffer interface, an intelligent throttling algorithm can be used to finely control the data processing traffic and computing resources, thereby avoiding the waste of computing resources.

[0064] In one implementation, the running screen can be preprocessed before feature extraction. The preprocessing process may include adjusting the screen size and color space. Adjusting the screen size ensures the uniformity of the input data dimensions. Adjusting the color space makes the screen features more prominent, which helps improve the accuracy of feature extraction.

[0065] In one implementation, feature extraction based on the running screen generates multimodal features, including: The running screen is reconstructed to generate a screen pyramid; Feature extraction is performed based on the image pyramid to generate multimodal features.

[0066] In one implementation, the image pyramid generated after reconstructing the running image can be a Gaussian pyramid based on the running image.

[0067] A Gaussian pyramid is a set of running images with progressively decreasing resolution. The first layer in a Gaussian pyramid can be the original running image, and each subsequent layer can be generated by Gaussian blurring and downsampling the previous layer. This process is repeated until a set of running images with decreasing resolution is obtained.

[0068] Feature extraction based on the image pyramid can extract key points in different image scale spaces of the image pyramid, thereby improving the robustness of feature extraction to changes in image resolution.

[0069] In one implementation, acquiring the running screen within a predetermined time period and extracting features based on the running screen to generate multimodal features includes: Acquire the running screen within a predetermined time period, and obtain screen scene information, screen object information, screen text information, and screen change information based on the running screen; Feature extraction is performed on scene information, object information, text information, and scene change information to generate scene features, object features, text features, and scene change features.

[0070] Scene information can indicate the environment in which the screen is running, such as indoor scenes, forest scenes, desert scenes, and urban scenes.

[0071] In one implementation, obtaining scene information based on the running screen can involve calling a preset neural network model and inputting the running screen into the preset neural network model. The preset neural network model performs scene classification based on the image features in the running screen to obtain the scene information corresponding to the running screen. For example, the preset neural network model is a convolutional neural network model.

[0072] The object information in the screen can indicate the information of the objects contained in the running screen, such as cars, tools and people appearing in the screen.

[0073] In one implementation, obtaining object information from the running screen can be achieved by calling a preset neural network model and inputting the running screen into the preset neural network model. The preset neural network model performs object detection based on the image features in the running screen to obtain the object information in the running screen. For example, the preset neural network model is a YOLO network.

[0074] On-screen text information indicates the text information contained in the running screen, such as captions or dialogue information on the screen.

[0075] In one implementation, obtaining text information from a running screen can be achieved by calling a preset text extraction model and inputting the running screen into the preset text extraction model. The preset text extraction model extracts text based on image features in the running screen to obtain the text information in the running screen. For example, the preset text extraction model is a deep learning model built based on optical character recognition technology.

[0076] Screen change information can indicate changes in the content of the running screen, such as a character in the screen changing from sitting to standing, or a character in the screen changing from laughing to crying.

[0077] In one implementation, image change information can be extracted based on frame difference technology. In frame difference technology, the range of motion in a moving image is examined by calculating the difference between two adjacent frames, and then image change information is extracted from it.

[0078] After acquiring scene information, object information, text information, and change information, feature extraction can be performed to obtain scene features, object features, and text features.

[0079] In one implementation, feature extraction of scene information, object information, text information, and scene change information can be performed using a pre-defined neural network model, such as a convolutional neural network model. Each type of information can correspond to a pre-defined neural network model for feature extraction. For example, scene information can be input into the corresponding pre-defined neural network model to obtain scene features; object information can be input into the corresponding pre-defined neural network model to obtain object features.

[0080] Based on the scene features, object features, text features, and scene change features extracted from the running screen, the sentiment type of the running screen can be analyzed in a multimodal manner from both the visual and textual levels, thus improving the accuracy of sentiment type analysis.

[0081] In one implementation, the process of extracting multimodal features from the running screen can be carried out by using a circular buffer to manage the running screen frame sequence. The circular buffer has the characteristics of high efficiency, stability and low latency in data processing. Therefore, using a circular buffer to manage the running screen frame sequence is beneficial to improving the efficiency and stability of bullet screen content generation and greatly reducing the system load.

[0082] After generating multimodal features, a preset neural network model can be called to perform sentiment analysis based on the multimodal features to obtain the sentiment type.

[0083] In one implementation, the preset neural network model can be a neural network model built on the Transformer architecture, used to perform multi-objective classification of the emotions conveyed by the running screen based on multimodal features, and to determine the emotion type corresponding to the running screen from multiple candidate emotion types.

[0084] In one implementation, multiple candidate emotion types may include happiness, trust, fear, surprise, sadness, disgust, anger, and anticipation.

[0085] In one implementation, the preset neural network model can be based on the probability distributions corresponding to multiple candidate emotion types input by multimodal feature input. For example, the output of the preset neural network model is {happiness: 0.1, trust: 0.08, fear: 0.2, surprise: 0.15, sadness: 0.12, disgust: 0.04, anger: 0.23, anticipation: 0.08}.

[0086] In one implementation, the emotion type corresponding to the running screen can be the one with the highest probability value among multiple candidate emotion types output by a preset neural network model. In the example above, the emotion type corresponding to the running screen is "anger".

[0087] Because emotions are complex, a single scene may contain multiple emotions. Therefore, in one implementation, determining the emotion type corresponding to the running scene from the probability distribution of candidate emotion types output by a preset neural network model includes: Multiple candidate sentiment types are sorted based on probability values, and a predetermined number of candidate sentiment types are obtained. The emotion type corresponding to the running screen is determined by the probability value of the candidate emotion type that exceeds the predetermined threshold among the first predetermined number of candidate emotion types.

[0088] For example, the result of ranking multiple candidate emotion types based on the probability distribution of candidate emotion types output by the preset neural network model in the above example is: anger, fear, surprise, sadness, happiness, trust, expectation, and disgust.

[0089] The predetermined number can be the number of emotion types determined from multiple candidate emotion types to correspond to the running screen. For example, if the predetermined number is 3, then the top 3 candidate emotion types can be obtained based on the sorting results, namely: anger, fear, and surprise.

[0090] However, only when the probability values ​​of one or more candidate emotion types are high can they be the emotion conveyed by the running screen. If a candidate emotion type ranks high in the ranking results but has a low probability value, then this candidate emotion type may not be the emotion conveyed by the running screen. Therefore, a predetermined threshold can be set to eliminate candidate emotion types with high rankings but low probability values, and candidate emotion types with both high rankings and high probability values ​​can be used as the emotion types corresponding to the running screen. For example, based on the above example, with a predetermined threshold of 0.18, among anger (0.23), fear (0.2), and surprise (0.15), the probability values ​​corresponding to anger and fear exceed the predetermined threshold. Therefore, anger and fear are determined to be the emotion types corresponding to the running screen.

[0091] In one implementation, a preset neural network model is invoked to perform sentiment analysis based on multimodal features to obtain sentiment types, including: Obtain historical sentiment types; The system calls a preset neural network model to perform sentiment analysis based on multimodal features and historical sentiment types to obtain the sentiment type.

[0092] Historical sentiment types can be derived from the sentiment types corresponding to the preceding frames in the current frame. By combining the multimodal features of the current frame with historical sentiment types to determine the current frame's sentiment type, reliable contextual information can be provided for sentiment analysis, thereby improving its accuracy. For example, if the historical sentiment type is happiness or anticipation, and the current frame's sentiment type is fear or disgust, the likelihood of a change in emotion is low; therefore, the current frame's sentiment type is more likely to be surprise or happiness. Thus, a pre-defined neural network model can perform sentiment analysis based on multimodal features and historical sentiment types to determine the current frame's sentiment type.

[0093] After performing sentiment analysis on multimodal features, interactive bullet comments can be generated based on sentiment types. The bullet comment content can match the atmosphere corresponding to the sentiment type of the current running screen. For example, when the sentiment type is "happy," the bullet comment content might be phrases like "So happy!", "Great!", or "Awesome!" that convey joy; when the sentiment type is "fear," the bullet comment content might be phrases like "Terrible!" or "I can't watch anymore!" that convey terror. Bullet comment content that matches the sentiment type of the running screen helps users build a sense of interactive experience and makes them more immersive.

[0094] In one implementation, generating bullet screen content corresponding to interactive bullet screens based on emotion type includes: Extract scene features corresponding to the running screen based on multimodal features; The interactive bullet comments are generated based on the scene context and emotional type.

[0095] The scene context features corresponding to the running screen can indicate the scene-related characteristics in the running screen. The scene can be a general description that combines events and emotions in the running screen, indicating the events that occur in the running screen and the emotions conveyed.

[0096] Extracting scene context features corresponding to the running screen based on multimodal features can extract features from the running screen that can represent events and emotions in the running screen. For example, the screen text information and screen change information in the multimodal features.

[0097] In one implementation, extracting scene features corresponding to the running screen based on multimodal features can be achieved by calling a preset neural network model. The preset neural network model can learn the contribution of features from different modalities to scene analysis and extract the features with higher contributions as scene features.

[0098] After obtaining the scene context features, the interactive bullet screen content can be generated based on the scene context features and emotional type.

[0099] In one implementation, generating interactive bullet screen content based on scene context features and emotional type includes: Based on the scene scene features, classify the scene scene and generate scene categories; The interactive bullet comments are generated based on the context category and emotion type.

[0100] Scenario categories can be summaries of the events and emotions conveyed in a running scene. For example, scenario categories are divided into simple scenarios and complex scenarios. In simple scenarios, the events and emotions conveyed in the running scene may be relatively simple, such as winning a game or being hit. In complex scenarios, the events and emotions conveyed in the running scene may be more complex, such as events that evoke mixed feelings of joy and sorrow, or events involving many people with different emotions.

[0101] In one implementation, scene classification based on scene features can be performed by calling a preset neural network model. The scene features are input into the preset neural network model for scene classification, generating corresponding scene categories.

[0102] Generating bullet comments based on scenario type and emotion type allows the bullet comments to simultaneously match the scenario and emotion of the running screen.

[0103] In one implementation, the interactive bullet screen content is generated based on the context category and emotion type, including: When the scenario category is the first scenario category, the bullet screen content database is retrieved, and the bullet screen content corresponding to the interactive bullet screen is retrieved from the bullet screen content database based on the sentiment type; When the scenario category is the second scenario category, the preset large language generation model is invoked to generate the corresponding bullet screen content based on the emotion type.

[0104] Scenario categories can be divided into primary scenario category and secondary scenario category. Primary scenario category indicates scenarios with simpler events and emotions conveyed in the running screen; secondary scenario category indicates scenarios with more complex events and emotions conveyed in the running screen.

[0105] For the first scenario category's running screen, since the events and emotions conveyed are relatively simple, statements that can convey the corresponding emotional type can be used as bullet screen content. For example, when the emotional type is happiness, the bullet screen content can be simple statements that convey a happy emotion, such as "hahaha" or "so happy." The bullet screen content database can store the bullet screen content of simple statements corresponding to each emotional type. After obtaining the bullet screen content database, the statements corresponding to the emotional type can be retrieved from it as the bullet screen content for interactive bullet screens.

[0106] For the second scenario category's running screen, the events and emotions conveyed are more complex, and there may be multiple emotion types. To accurately represent these complex emotions in the bullet screen content, a bullet screen content database alone is insufficient. Therefore, a pre-defined large language generation model can be invoked to generate bullet screen content corresponding to the emotion type.

[0107] The pre-trained large language generation model can generate corresponding sentences as bullet screen content based on the emotional type of the input. For example, since emotional types include happiness and sadness, the pre-trained large language model can generate bullet screen content similar to "both happy and sad" for these two seemingly contradictory emotional types.

[0108] In one implementation, the training process of the pre-defined large language generation model includes: Retrieve historical bullet comment data posted by the target object; Obtain the initial large language generation model, and update the parameters of the initial large language generation model based on historical bullet screen data to generate the preset large language generation model.

[0109] The target object can be an object that is using an emulator application to connect to a virtual operating system. Historical bullet screen data can be bullet screens posted by the target object within a predetermined time period prior to the current time in the target application or other related applications.

[0110] When a target application posts a bullet comment in its own application or other related applications, the bullet comment text can be stored. When training a pre-defined large language generation model, historical bullet comment data from the target object within a predetermined time period can be obtained to fine-tune the large language generation model.

[0111] Note that you must obtain the target's consent before acquiring their historical bullet comment data. Furthermore, the collection, use, and processing of historical bullet comment data will comply with relevant laws, regulations, and standards. When obtaining the target's consent, you can do so through pop-up windows or redirection to a confirmation page to obtain their individual permission or consent.

[0112] The initial large language generation model can be based on a large language generation model pre-trained on massive amounts of text data. By training the initial large language generation model using historical bullet screen data, it can learn the language habits of the target audience, enabling it to generate bullet screen content that aligns with the target audience's ideas.

[0113] In another implementation, after obtaining the historical bullet screen data published by the target object, the method further includes: obtaining the associated historical bullet screen data published by related objects associated with the target object.

[0114] Related objects can be other objects that interact with the target object in the target application or other related applications. Interaction events can include the target object liking, commenting, or saving content related to the related objects. Because there have been interaction events between the target object and related objects, the bullet comments generated using the language habits of the related objects, while differing from the target object's language habits, can create a more immersive experience for the target object in a group discussion.

[0115] Obtaining the associated historical bullet comment data published by related objects can retrieve the target object's interaction events within a predetermined time period and determine the associated objects based on these events. This involves retrieving the associated historical bullet comment data published by related objects from the stored bullet comment content.

[0116] Note that prior consent from the associated parties must be obtained before acquiring their historical bullet comment data. Furthermore, the collection, use, and processing of this data will comply with relevant laws, regulations, and standards. When obtaining consent from associated parties, individual permission or consent can be obtained through pop-up windows or redirection to a confirmation page.

[0117] Based on this, an initial large language generation model is obtained, and the parameters of the initial large language generation model are updated based on historical bullet screen data to generate a preset large language generation model. This includes: obtaining the initial large language generation model, updating the parameters of the initial large language generation model based on historical bullet screen data and associated historical bullet screen data to generate a preset large language generation model.

[0118] By training an initial large language generation model using historical bullet screen data and related historical bullet screen data, a preset large language generation model can be generated. This allows the bullet screen content generated by the preset large language model to be more diverse while conforming to the language habits of the target audience, enabling the target audience to obtain an immersive experience of community discussion.

[0119] Training a pre-defined large language generation model based on historical bullet comment data posted by the target audience can make the bullet comments generated by the pre-defined large language generation model more in line with the target audience's language habits, making the target audience more likely to identify with the bullet comment content and improving the accuracy of bullet comment content generation.

[0120] For the first scenario category of runtime content, retrieving bullet comment content from a bullet comment content database can improve the efficiency of bullet comment generation. For the second scenario category of runtime content, using a preset large language model to generate bullet comment content can make the bullet comment content more closely match the runtime screen, thus improving the accuracy of bullet comment generation.

[0121] Classifying the scene scenarios and generating bullet screen content based on scenario categories and emotional types can make the influence of scene scenarios on bullet screen content generation clearer. Different bullet screen content can be generated for different types of scene scenarios, which helps to improve the accuracy of bullet screen content generation.

[0122] Generating bullet comments based on scene context and emotion type can obtain bullet comment content according to the scene context, making the bullet comment content more matching the scene context and improving the accuracy of bullet comment content generation.

[0123] The generation of interactive bullet comments based on sentiment analysis of the multimodal features corresponding to the running screen can accurately analyze the sentiment type corresponding to the running screen from multiple dimensions, thereby improving the accuracy of generating interactive bullet comments based on sentiment type.

[0124] After generating the corresponding bullet screen content, the bullet screen content can be rendered on the device's display interface to form interactive bullet screens. For example... Figure 2E As shown, interactive bullet comments include phrases like "Beautiful scenery," "Beautiful buildings," "Great job," and "Tourist destination." Although these interactive bullet comments are generated based on the target application's screen, they can ignore the boundaries between different operating systems and be displayed on a broader device display interface.

[0125] After the bullet screen control is triggered, it can intelligently detect the current system performance indicators, dynamically set data processing parameters suitable for the current hardware configuration, load rendering resources, and ensure the efficiency and effect of interactive bullet screen rendering.

[0126] In one implementation, the bullet screen display process includes: Obtain the third preset display effect, and add a bullet screen overlay layer on the device display interface based on the third preset display effect; Render interactive bullet comments in the bullet comment overlay layer.

[0127] The third preset display effect can be a pre-set bullet screen rendering effect. For example, the bullet screen is displayed at the top with red text.

[0128] A bullet screen overlay can be a transparent layer on the device's display interface, and each interactive bullet screen can correspond to one bullet screen overlay. After obtaining the third preset display effect, bullet screen overlays can be added to the device's display interface based on the third preset display effect. For example, if the third preset display effect specifies that the bullet screen is displayed at the top, then a bullet screen overlay can be added to the order section on the device's display interface; if the font color is red, then the font color can be preset to red on the bullet screen overlay.

[0129] In one implementation, when adding a bullet screen overlay layer to the device display interface, if there are multiple bullet screen overlay layers corresponding to multiple interactive bullet screens, the multiple bullet screen overlay layers can be added adaptively according to the size of the device display interface and the positional relationship between the multiple bullet screen overlay layers while determining the addition position based on the third preset display effect, thereby avoiding overlapping rendering of multiple interactive bullet screens and improving the rationality of the rendering layout of multiple interactive bullet screens.

[0130] In one implementation, when there are multiple bullet screen overlays corresponding to multiple interactive bullet screens, the display effect can be set independently for each bullet screen overlay. For example, bullet screen overlay D1 can be displayed scrolling at the top, while bullet screen overlay D2 can be displayed hovering on the side. Therefore, when adding bullet screen overlays to the device display interface, the target display effect corresponding to the bullet screen overlay can be obtained.

[0131] After adding a bullet screen overlay to the device display interface, the corresponding interactive bullet screen can be rendered in the bullet screen overlay.

[0132] In one implementation, rendering interactive bullet comments on a bullet comment overlay includes: obtaining a preset rendering effect and rendering the interactive bullet comments on the bullet comment overlay based on the preset rendering effect. The preset rendering effect can be a preset rendering animation effect in the system, such as integrated particle effects and dynamic transition animation effects for interactive bullet comments.

[0133] In one implementation, the rendering process of interactive bullet comments can be optimized using batch rendering and dirty rectangle updates through the DirectX / OpenGL underlying interface, ensuring smooth rendering while rendering complex visual effects.

[0134] During the display of bullet comments, interactive bullet comments are rendered on a transparent overlay. Interactive bullet comments can be set independently on the transparent overlay, which improves the accuracy of bullet comment display.

[0135] In one of the aforementioned embodiments, the content of the interactive bullet comments is generated based on the emotional type of the running screen. Therefore, in one embodiment, the bullet comment display process includes: The second preset display effect is obtained based on the emotion type of the interactive bullet comments; The interactive bullet screen is rendered on the device display interface based on the second preset display effect.

[0136] The second preset display effect can be the display effect corresponding to the emotion type. Different emotion types can correspond to different display effects. For example, the text color of the display effect corresponding to the interactive bullet screen with the emotion type of "happiness" can be bright yellow; the text color of the display effect corresponding to the interactive bullet screen with the emotion type of "sadness" can be blue; and the text color of the display effect corresponding to the interactive bullet screen with the emotion type of "fear" can be red.

[0137] The second preset display effect corresponding to different emotion types can be preset by the object.

[0138] After obtaining the second preset display effect, interactive bullet comments can be rendered on the device display interface based on the second preset display effect.

[0139] Using different display effects to render interactive bullet comments for different emotional types allows the audience to more intuitively feel the emotions conveyed by the bullet comments, which helps improve the information transmission efficiency of interactive bullet comments and further enhances the audience's immersive experience in interactive bullet comments.

[0140] In one implementation, after displaying interactive bullet comments on the device display interface in response to a trigger operation of the bullet comment control, the method further includes: In response to the triggering of interactive bullet comments, a list of interactive options corresponding to the interactive bullet comments is displayed on the device display interface; In response to a trigger operation on a target interactive control in the interactive options list, display the corresponding bullet screen interactive effect for the target interactive control.

[0141] The trigger for interactive bullet comments is clicking on them. The interaction options list includes multiple interactive controls, which can be used for interaction between objects and bullet comments. The interaction options list can be a floating window displayed on the device's screen. For example... Figure 5 As shown, after the interactive bullet screen "Great, great" is triggered, the corresponding interactive option list 510 is displayed, which includes interactive control S1, interactive control S2 and interactive control S3.

[0142] After triggering the target interactive control in the interactive options list, interactive bullet comments can be displayed using the corresponding bullet comment interaction effect of the target interactive control.

[0143] In one implementation, the interactive option list includes a bullet comment like control. Therefore, in response to a trigger operation on a target interactive control in the interactive option list, displaying the bullet comment interaction effect corresponding to the target interactive control includes: in response to a trigger operation on the bullet comment like control, displaying the interactive bullet comment with a first preset display effect.

[0144] The "like" action in bullet comments indicates that an audience likes a specific interactive bullet comment. The first preset display effect can be the effect displayed after a "like" action is performed on the interactive bullet comment. For example, adding an underline to a liked interactive bullet comment.

[0145] For example Figure 6A As shown, the interactive option list 510 includes a bullet comment like control 611. After clicking the bullet comment like control 611, as shown... Figure 6BAs shown, the interactive bullet comment "Great, great" is underlined.

[0146] Displaying liked interactive comments with the first preset display effect allows users to see liked interactive comments more intuitively, improving the interactive experience and information transmission efficiency of interactive comments.

[0147] In another implementation, the interactive options list includes a bullet screen blocking control. Therefore, in response to a triggering operation on a target interactive control in the interactive options list, displaying the bullet screen interaction effect corresponding to the target interactive control includes: in response to a triggering operation on the bullet screen blocking control, canceling the display of interactive bullet screens on the device display interface.

[0148] The "disable bullet comments" function indicates to an object that it dislikes a particular bullet comment. Therefore, when an object triggers a bullet comment screen action, the bullet comment can be undisplayed on the device's display interface.

[0149] For example Figure 7A As shown, the interactive option list 510 includes a bullet screen blocking control 711. After clicking the bullet screen blocking control 711, as... Figure 7B As shown, the interactive bullet comment "Great, great" disappeared.

[0150] Unblocking interactive bullet comments that the object doesn't want to see can prevent the appearance of bullet comments that the object doesn't want to see, which helps to improve the object's immersive experience.

[0151] Based on the list of interactive options, objects can interact with interactive bullet comments, which helps to enhance the immersive experience that interactive bullet comments bring to objects.

[0152] In one of the aforementioned implementations, the bullet screen content can be generated based on a preset large language generation model.

[0153] Based on this, in one implementation, the training process of the pre-defined large language generation model includes: Obtain the target object's feedback information to the interactive bullet comments; The parameters of the preset large language generation model are updated based on interactive bullet comments and feedback information.

[0154] Feedback information can indicate the target audience's attitude towards the interactive bullet comments (danmu). In one implementation, the feedback information can be obtained based on the interaction between the target audience and the interactive bullet comments. For example, in the aforementioned implementation, the interactive option list corresponding to the interactive bullet comments includes a bullet comment like control and a bullet comment block control. Triggering the bullet comment like control indicates that the target audience likes the interactive bullet comment; triggering the bullet comment block control indicates that the target audience dislikes the interactive bullet comment.

[0155] Therefore, based on the interactive actions of the target object, feedback information can be divided into positive feedback information and negative feedback information. Positive feedback information indicates that the target object likes the interactive bullet screen, while negative feedback information indicates that the target object dislikes the interactive bullet screen.

[0156] After obtaining the target object's feedback information to the interactive bullet comments, the preset large language generation model can be trained based on the interactive bullet comments and feedback information.

[0157] In one implementation, training samples can be constructed based on interactive bullet comments and feedback information. Positive training samples are constructed based on interactive bullet comments corresponding to positive feedback information; negative training samples are constructed based on interactive bullet comments corresponding to negative feedback information. A pre-defined large language generation model performs reinforcement learning based on the positive and negative training samples, making the generated bullet comment content approach the style preferred by the target audience and avoid styles disliked by the target audience.

[0158] Training a pre-defined large language generation model based on feedback from the target audience can adjust the model's tendency to generate bullet comments, thereby personalizing bullet comment generation and enhancing the immersive experience provided by interactive bullet comments to the target audience.

[0159] In one implementation, the application interface also includes a bullet screen settings control. Triggering the bullet screen settings control allows for adjustments to the display effects of interactive bullet screens.

[0160] Based on this, in response to the triggering operation of the bullet screen control, after displaying the interactive bullet screen on the device display interface, the following also includes: In response to a triggering action on the bullet screen settings control, display a list of bullet screen settings options; In response to a trigger operation on the target bullet screen effect setting control in the bullet screen setting option list, the interactive bullet screen is displayed with the display effect corresponding to the target bullet screen effect setting control.

[0161] The bullet screen settings list includes multiple bullet screen effect setting controls. These controls indicate various different bullet screen display effects, such as bullet screen display at the top, bullet screen display at the side, bullet screen full screen display, bullet screen display with a small number of bullet screens, and bullet screen scrolling display.

[0162] The bullet comment settings list can also be displayed as a floating window in the application interface. After clicking the target bullet comment effect setting control in the bullet comment settings list, the interactive bullet comment will be displayed according to the display effect corresponding to the target bullet comment effect setting control.

[0163] For example Figure 8A As shown, the application interface includes a bullet screen settings control 810. After triggering the bullet screen settings control 810, as... Figure 8BAs shown, a list of bullet screen setting options 820 is displayed, including a bullet screen top display control 821, a bullet screen side display control 822, a bullet screen full-screen display control 823, a bullet screen partial display control 824, a bullet screen text setting control 825, and a bullet screen operation setting control 826. Among these, triggering the bullet screen text setting control 825 and the bullet screen operation setting control 826 allows for more refined settings of the bullet screen text display effect and operation mode. After the object triggers the bullet screen top display control 821, as shown... Figure 8C As shown, the interactive bullet comments are displayed at the top of the device's display interface.

[0164] The bullet screen settings list allows users to customize various display effects for interactive bullet screens, enhancing the personalization of the bullet screen display.

[0165] In one implementation, the application interface also includes a recommended content control; In response to the triggering operation of the bullet screen control, after displaying the interactive bullet screen on the device display interface, it also includes: In response to a trigger action on the recommended content control, the recommended content is displayed in the application interface.

[0166] The recommended content control can be used to enable the display of recommended content in the application interface. Recommended content is generated based on the running screen and interactive comments.

[0167] When an object controls a virtual operating system through an emulator application, recommended content can be generated based on the target application's screen and interactive comments. This recommended content can be generated based on the object's emotional state. For example, if the target application's screen displays a travel video, and the interactive comments include travel-related remarks, the application interface can recommend travel-related applications or other content to the object, thereby increasing the diversity of how the object uses the emulator application to operate the virtual operating system.

[0168] Therefore, once the recommended content control is triggered, the recommended content can be displayed in the application interface. For example... Figure 9A As shown, after triggering the recommended content control control 910, as follows: Figure 9B As shown, the application interface displays a recommended content display box 920, which shows recommended content generated based on the running screen and interactive bullet comments, including content M1, content M2, application N1 and application N2.

[0169] In one implementation, the recommended content is determined based on the running screen and interactive bullet comments, including: Feature extraction is performed on the running screen and interactive bullet comments to generate image feature sequences and text features; The system invokes a pre-defined neural network model to perform content matching based on image feature sequences and text features to determine recommended content.

[0170] Feature extraction from the running screen and interactive bullet comments can be performed using a preset neural network model to generate image feature sequences corresponding to the running screen and text features corresponding to the interactive bullet comments.

[0171] The preset neural network model can search the candidate content library for content that matches the image feature sequence and text features as recommended content.

[0172] In summary, this embodiment of the disclosure achieves cross-system collaborative interaction between the current device's operating system and a virtual operating system through an emulator application. When the target application's running screen is displayed in the emulator application's interface on the device's display screen, triggering the bullet screen control allows interactive bullet screens generated through sentiment analysis of the running screen to be displayed on the device's display screen. The bullet screens can be displayed more prominently on the device's display screen, rather than just on the target application's running screen. This increases the display area of ​​the bullet screens, allowing the information conveyed by the bullet screens to be displayed more clearly on the interface, greatly improving the information transmission efficiency of human-computer interaction using bullet screens in cross-system collaborative interaction scenarios.

[0173] This disclosure provides a detailed description of embodiments in conjunction with specific application scenarios. like Figure 10 The diagram illustrates the specific process of applying the bullet screen processing method provided in this disclosure to a bullet screen rendering and display scenario during cross-system operation. This method can be executed by an electronic device, which can be... Figure 1 The terminal 140 or server 110 shown. In this embodiment, the method is described using the terminal 140 as an example. (In conjunction with...) Figure 11 The methods for handling bullet comments include: Step 1001: In response to the launch command for the emulator application in the terminal, run the emulator application in the terminal.

[0174] An emulator application can run within an operating system and is used to create a target virtual operating system within the current operating system. Within this target virtual operating system, applications that are not supported by the current operating system but are supported by the virtual operating system can run. Emulator applications can provide a range of tools and interfaces for interacting with other devices or operating systems.

[0175] When the emulator application receives a run request for the target virtual operating system, it runs the target virtual operating system.

[0176] The target virtual operating system can be another operating system running within the virtual operating environment provided by the emulator application. Within the target virtual operating system, applications and data in the operating system of another device can be controlled. For example, the current device is a desktop computer running Windows, while the virtual operating system connected through the emulator application is the Android system installed on a mobile device.

[0177] Step 1002: In response to the triggering of the target application in the target virtual operating system, display the running screen of the target application in the application interface of the emulator application.

[0178] The target application can be an application that runs within a virtual operating system. Examples include video applications or game applications within a virtual operating system.

[0179] After triggering the target application's open control within the virtual operating system, the target application's running screen can be displayed. This running screen can be the screen displayed after the target application is opened.

[0180] Step 1003: In response to the triggering operation of the bullet screen control in the simulator application, generate an interactive bullet screen display request.

[0181] The bullet screen control is a control in the emulator application's interface used to control whether bullet screens are turned on or off. When the bullet screen control is triggered, the bullet screen is turned on; when the bullet screen control is triggered again, the bullet screen is turned off.

[0182] When the bullet screen control is triggered in the emulator application, an interactive bullet screen display request can be generated. Based on the interactive bullet screen display request, interactive bullet screens can be generated and rendered on the device's display interface.

[0183] Step 1004: The terminal initializes the bullet screen system based on the interactive bullet screen display request.

[0184] Upon receiving a request to display interactive bullet comments, the system can intelligently detect current system performance metrics and dynamically set data processing parameters suitable for the current hardware configuration, loading rendering resources to ensure efficient and effective interactive bullet comment rendering. A multi-layered caching mechanism can also be established to ensure that the interactive bullet comment generation function is quickly ready while other non-essential components provide support as needed.

[0185] After configuring the system settings, the bullet screen system initialization is complete.

[0186] Step 1005: The terminal obtains the running screen of the target application during the predetermined time period.

[0187] The predetermined time period can be a pre-defined period of time prior to the current time. Within this predetermined time period, running frames are captured at a predetermined frame rate to form a sequence of running frame images. The predetermined frame rate can be the frequency at which static frames are extracted from a continuous segment of running frames; for example, the predetermined frame rate is 15fps. Based on the predetermined frame rate, an image sequence consisting of multiple static running frames can be constructed.

[0188] After acquiring the running screen, it can be preprocessed. The preprocessing process can include adjusting the screen size and color space.

[0189] Step 1006: The terminal extracts features from the running screen to obtain multimodal features.

[0190] Multimodal features can include scene features, object features, text features, and image change features.

[0191] Scene features can indicate the environment information of the running screen; object features can indicate the object information contained in the running screen; text features indicate the text information contained in the running screen; screen change features can indicate the content changes in the running screen.

[0192] Extracting scene features, object features, text features, and screen change features based on the running screen can be done using a pre-defined neural network model, such as a convolutional neural network model. The running screen is input into the pre-defined neural network model corresponding to each modality feature to extract the corresponding features.

[0193] Step 1007: The terminal calls a preset neural network model to perform sentiment analysis based on multimodal features and obtains the sentiment type.

[0194] The preset neural network model can be a neural network model built on a Transformer architecture, used to perform multi-objective classification of the emotions conveyed by the running screen based on multimodal features, and to determine the corresponding emotion type of the running screen from multiple candidate emotion types. Multiple candidate emotion types can include happiness, trust, fear, surprise, sadness, disgust, anger, and anticipation, etc.

[0195] Step 1008: The terminal calls the preset large language generation model to generate interactive bullet screen content based on emotion type and obtains the bullet screen display effect.

[0196] The pre-trained large language generation model can generate corresponding sentences as bullet screen content based on the sentiment type of the input.

[0197] Different emotional types can correspond to different display effects for bullet comments. For example, the text color of the display effect for interactive bullet comments with the emotional type of "happiness" can be yellow; the text color of the display effect for interactive bullet comments with the emotional type of "sadness" can be blue. The bullet comment display effects corresponding to different emotional types can be preset by the object. The corresponding bullet comment display effect can be obtained based on the emotional type.

[0198] Step 1009: The terminal renders interactive bullet comments on the device display interface based on the bullet comment display effect.

[0199] When rendering interactive bullet comments on the device's display interface, a bullet comment overlay can be added. This overlay can be a transparent layer on the display interface, and each interactive bullet comment can correspond to one overlay. The position of the overlay can be determined based on a preset bullet comment display effect. For example, if the preset effect is to display bullet comments at the top, then the overlay can be added at the top of the display interface.

[0200] After adding a bullet screen overlay, interactive bullet screens can be rendered on the overlay based on the display effect corresponding to the emotion type.

[0201] Step 1010: The terminal obtains object feedback information based on the interactive option list corresponding to the interactive bullet screen.

[0202] When interactive comments are displayed on the device's screen, triggering the interactive comments will display a list of interactive options. This list can include comments like and comments block controls.

[0203] When an object triggers the comment like control, the interactive comment can be re-rendered with the display effect corresponding to the comment like control, so that the interactive comment is displayed on the device display interface with the effect after liking, for example, changing the comment text color to yellow.

[0204] Once the object triggers the bullet screen blocking control, the interactive bullet screen can also be instantly canceled in the device's display interface.

[0205] When an object triggers the comment like control, it indicates that the object likes the interactive comment and generates positive feedback; when an object triggers the comment block control, it indicates that the object dislikes the interactive comment and generates negative feedback.

[0206] Step 1011: The terminal adjusts the parameters of the preset large language generation model based on the object feedback information.

[0207] After obtaining the object's feedback information, training samples can be generated based on the object's feedback information and the corresponding interactive comments. Interactive comments corresponding to positive interaction information can generate positive training samples; interactive comments corresponding to negative interaction information can generate negative training samples.

[0208] By using positive and negative training samples, a pre-defined large language generation model can be trained to make the generated bullet screen content closer to the style preferred by the target audience and away from the style disliked by the target audience.

[0209] In summary, the bullet screen processing method of this disclosure can achieve boundless bullet screen rendering in cross-operating system interaction scenarios. By ignoring the barriers between operating systems, interactive bullet screens can be displayed in a wider space, greatly improving the information transmission efficiency of interactive bullet screens.

[0210] Description of apparatus and devices according to embodiments of this disclosure It is understood that although the steps in the above flowcharts are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated in this embodiment, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the above flowcharts may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps.

[0211] It should be noted that in various specific embodiments of this application, when processing is required based on data related to the characteristics of the target content, such as target content attribute information or attribute information sets, permission or consent from the target content provider will be obtained first. Furthermore, the collection, use, and processing of this data will comply with relevant laws, regulations, and standards. In addition, when embodiments of this application require obtaining target content attribute information, separate permission or consent from the target content provider will be obtained through pop-ups or redirection to a confirmation page. Only after obtaining the separate permission or consent from the target content provider will the necessary target content-related data for the normal operation of the embodiments of this application be obtained.

[0212] Figure 12 This is a schematic diagram of the structure of a bullet screen processing device 1200 provided in an embodiment of the present disclosure. The bullet screen processing device 1200 includes: The first display unit 1210 is used to display the application interface of the simulator application in the device display interface. The application interface displays the running screen of the target application and the bullet screen control control. The target application runs in the virtual operating system provided by the simulator application. The second display unit 1220 is used to display interactive bullet comments on the device display interface in response to the triggering operation of the bullet comment control control. The interactive bullet comments are generated based on the sentiment analysis results obtained from the running screen.

[0213] Optionally, in one embodiment, the bullet screen processing device 1200 further includes: The third display unit (not shown) is used to display a list of interactive options corresponding to the interactive bullet screen on the device display interface in response to the triggering operation of the interactive bullet screen. The list of interactive options includes multiple interactive controls. The fourth display unit (not shown) is used to display the bullet screen interaction effect corresponding to the target interactive control in response to the trigger operation of the target interactive control in the interactive option list.

[0214] Optionally, in one implementation, the list of interactive options includes a bullet screen like control; The fourth display unit (not shown) is specifically used for: In response to the triggering of the danmaku like control, the interactive danmaku is displayed with the first preset display effect.

[0215] Optionally, in one implementation, the interactive options list includes a bullet screen blocking control; The fourth display unit (not shown) is specifically used for: In response to the triggering of the bullet screen blocking control, the interactive bullet screen is canceled from being displayed on the device's display interface.

[0216] Optionally, in one implementation, the application interface also includes a bullet screen setting control; The bullet screen processing device 1200 also includes: The fifth display unit (not shown) is used to display a list of bullet screen setting options in response to a trigger operation on the bullet screen setting control. The bullet screen setting option list includes multiple bullet screen effect setting controls. The sixth display unit (not shown) is used to display interactive bullet comments in response to a trigger operation on the target bullet comment effect setting control in the bullet comment setting option list, with the display effect corresponding to the target bullet comment effect setting control.

[0217] Optionally, in one implementation, the application interface also includes a recommended content control control; The bullet screen processing device 1200 also includes: The seventh display unit (not shown) is used to display recommended content in the application interface in response to the triggering operation of the recommended content control control. The recommended content is generated based on the running screen and interactive bullet comments.

[0218] Optionally, in one implementation, the process of generating the bullet screen content corresponding to the interactive bullet screen includes: The system acquires the running screen within a predetermined time period and extracts features based on the running screen to generate multimodal features. A pre-defined neural network model is invoked to perform sentiment analysis based on multimodal features to obtain the sentiment type; The interactive bullet comments are generated based on the emotion type.

[0219] Optionally, in one implementation, acquiring the running screen within a predetermined time period and generating multimodal features based on the running screen includes: Acquire the running screen within a predetermined time period, and obtain screen scene information, screen object information, screen text information, and screen change information based on the running screen; Feature extraction is performed on scene information, object information, text information, and scene change information to generate scene features, object features, text features, and scene change features.

[0220] Optionally, in one implementation, generating bullet screen content corresponding to interactive bullet screens based on emotion type includes: Extract scene features corresponding to the running screen based on multimodal features; The interactive bullet comments are generated based on the scene context and emotional type.

[0221] Optionally, in one implementation, generating bullet screen content corresponding to interactive bullet screens based on scene context features and emotion type includes: Based on the scene scene features, classify the scene scene and generate scene categories; The interactive bullet comments are generated based on the context category and emotion type.

[0222] Optionally, in one implementation, generating bullet screen content corresponding to the interactive bullet screen based on the scenario category and emotion type includes: When the scenario category is the first scenario category, the bullet screen content database is retrieved, and the bullet screen content corresponding to the interactive bullet screen is retrieved from the bullet screen content database based on the sentiment type; When the scenario category is the second scenario category, the preset large language generation model is invoked to generate the corresponding bullet screen content based on the emotion type.

[0223] Optionally, in one implementation, the training process of the pre-defined large language generation model includes: Retrieve historical bullet comment data posted by the target object; Obtain the initial large language generation model, and update the parameters of the initial large language generation model based on historical bullet screen data to generate the preset large language generation model.

[0224] Optionally, in one implementation, after obtaining the historical bullet comment data posted by the target object, the method further includes: Retrieve historical bullet comment data posted by related objects associated with the target object; Obtain the initial large language generation model, and update the parameters of the initial large language generation model based on historical bullet screen data to generate a preset large language generation model, including: Obtain the initial large language generation model, and update the parameters of the initial large language generation model based on historical bullet screen data and related historical bullet screen data to generate the preset large language generation model.

[0225] Optionally, in one implementation, the training process of the pre-defined large language generation model includes: Obtain the target object's feedback information to the interactive bullet comments; The parameters of the preset large language generation model are updated based on interactive bullet comments and feedback information.

[0226] Optionally, in one implementation, the process of displaying interactive bullet comments includes: The second preset display effect is obtained based on the emotion type of the interactive bullet comments; The interactive bullet screen is rendered on the device display interface based on the second preset display effect.

[0227] Optionally, in one implementation, the bullet screen display process includes: Obtain the third preset display effect, and add a bullet screen overlay layer on the device display interface based on the third preset display effect; Render interactive bullet comments in the bullet comment overlay layer.

[0228] This disclosure also provides an electronic device, which includes a memory and a processor. The memory stores a computer program; when the processor executes the computer program, it implements the bullet screen processing method of this disclosure. The electronic device can be a terminal 140 or a server 110.

[0229] Reference Figure 13 , Figure 13To implement the structural block diagram of a portion of the terminal 140 according to an embodiment of this disclosure, the terminal includes: a radio frequency (RF) circuit 1310, a memory 1315, an input unit 1330, a display unit 1340, a sensor 1350, an audio circuit 1360, a wireless fidelity (WiFi) module 1370, a processor 1380, and a power supply 1390, among other components. Those skilled in the art will understand that... Figure 13 The terminal 140 structure shown does not constitute a limitation on a mobile phone or computer, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0230] The RF circuit 1310 can be used to receive and transmit signals during information transmission or calls. In particular, it receives downlink information from the base station and processes it with the processor 1380; in addition, it transmits uplink data to the base station.

[0231] The memory 1315 can be used to store software programs and modules. The processor 1380 executes various functional applications of the content terminal and generates text summaries by running the software programs and modules stored in the memory 1315.

[0232] The input unit 1330 can be used to receive input numeric or character information, and to generate key signal inputs related to the settings and function control of the content terminal. Specifically, the input unit 1330 may include a touch panel 1331 and other input devices 1332.

[0233] The display unit 1340 can be used to display input or provided information, as well as various menus of the content terminal. The display unit 1340 may include a display panel 1341.

[0234] Audio circuitry 1360, speaker 1361, and microphone 1362 provide an audio interface.

[0235] In this embodiment, the processor 1380 included in the terminal 140 can execute the bullet screen processing method of the previous embodiment.

[0236] The embodiments of the present invention can be applied to various scenarios, including but not limited to cross-system data processing and cross-system media data exchange.

[0237] Figure 14This is a partial structural block diagram of a server 110 implementing an embodiment of the present disclosure. The server 110 can vary significantly due to different configurations or performance characteristics, and may include one or more central processing units (CPUs) 1422 (e.g., one or more processors) and memory 1432, and one or more storage media 1430 (e.g., one or more mass storage devices) for storing application programs 1442 or data 1444. The memory 1432 and storage media 1430 may be temporary or persistent storage. The program stored in the storage media 1430 may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the server. Furthermore, the CPU 1422 may be configured to communicate with the storage media 1430 and execute the series of instruction operations stored in the storage media 1430 on the server.

[0238] Server 110 may also include one or more power supplies 1426, one or more wired or wireless network interfaces 1450, one or more input / output interfaces 1458, and / or one or more operating systems 1441, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0239] The central processing unit 1422 in server 110 can be used to execute the bullet screen processing method of the present disclosure embodiments.

[0240] This disclosure also provides a storage medium for storing a computer program for executing the bullet screen processing methods of the foregoing embodiments.

[0241] This disclosure also provides a computer program product, which includes a computer program. The processor of an electronic device reads and executes the computer program, causing the electronic device to perform the above-described bullet screen processing method.

[0242] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in this disclosure and the foregoing drawings are used to distinguish similar terms and are not necessarily used to describe a particular order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this disclosure described herein can be implemented, for example, in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “including,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatuses.

[0243] It should be understood that in this disclosure, "at least one item" refers to one or more items, and "more than one item" refers to two or more items. "And / or" is used to describe the relationship between related content, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related content are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0244] It should be understood that in the description of the embodiments of this disclosure, "multiple" means two or more, "greater than", "less than", "exceeding" etc. are understood to exclude the number itself, and "above", "below", "within" etc. are understood to include the number itself.

[0245] In this disclosure, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0246] In the several embodiments provided in this disclosure, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0247] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0248] Furthermore, the functional units in the various embodiments of this disclosure can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0249] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this disclosure. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0250] It should also be understood that the various implementation methods provided in this disclosure can be combined arbitrarily to achieve different technical effects. The above is a detailed description of the embodiments of this disclosure. However, this disclosure is not limited to the above embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of this disclosure. All such equivalent modifications or substitutions are included within the scope defined by the claims of this disclosure.

Claims

1. A method for processing bullet comments, characterized in that, include: The device display interface shows the application interface of the simulator application, which displays the running screen of the target application and the bullet screen control control. The target application runs in the virtual operating system provided by the simulator application. In response to a trigger operation on the bullet screen control control, interactive bullet screens are displayed on the device display interface; The content of the interactive bullet comments is generated based on the analysis results obtained from the sentiment analysis of the running screen.

2. The method according to claim 1, characterized in that, After displaying interactive bullet comments on the device display interface in response to a trigger operation of the bullet comment control, the method further includes: In response to the triggering operation of the interactive bullet screen, a list of interactive options corresponding to the interactive bullet screen is displayed on the device display interface, and the list of interactive options includes multiple interactive controls; In response to a trigger operation on a target interactive control in the list of interactive options, the bullet screen interactive effect corresponding to the target interactive control is displayed.

3. The method according to claim 2, characterized in that, The list of interactive options includes a bullet screen like control; In response to a trigger operation on a target interactive control in the interactive option list, the corresponding bullet screen interactive effect of the target interactive control is displayed, including: In response to the triggering operation of the bullet comment like control, the interactive bullet comment is displayed with a first preset display effect.

4. The method according to claim 2, characterized in that, The list of interactive options includes a bullet screen blocking control; In response to a trigger operation on a target interactive control in the interactive option list, the corresponding bullet screen interactive effect of the target interactive control is displayed, including: In response to the triggering operation of the bullet screen blocking control, the interactive bullet screen is canceled from being displayed on the device display interface.

5. The method according to claim 1, characterized in that, The application interface also includes bullet screen settings controls; After displaying interactive bullet comments on the device display interface in response to a trigger operation of the bullet comment control, the method further includes: In response to a trigger operation on the bullet screen setting control, a bullet screen setting option list is displayed, which includes multiple bullet screen effect setting controls; In response to a trigger operation on a target bullet screen effect setting control in the bullet screen setting option list, the interactive bullet screen is displayed with the display effect corresponding to the target bullet screen effect setting control.

6. The method according to claim 1, characterized in that, The application interface also includes a recommendation content control control; After displaying interactive bullet comments on the device display interface in response to a trigger operation of the bullet comment control, the method further includes: In response to a trigger operation on the recommended content control control, recommended content is displayed in the application interface, wherein the recommended content is generated based on the running screen and the interactive bullet comments.

7. The method according to claim 1, characterized in that, The generation process of the bullet screen content corresponding to the interactive bullet screen includes: The system acquires the running screen within a predetermined time period and performs feature extraction based on the running screen to generate multimodal features. A preset neural network model is invoked to perform sentiment analysis based on the multimodal features to obtain the sentiment type; The interactive bullet screen content is generated based on the emotion type.

8. The method according to claim 7, characterized in that, The step of acquiring the running screen within a predetermined time period and extracting features based on the running screen to generate multimodal features includes: The system acquires the running screen within a predetermined time period, and acquires screen scene information, screen object information, screen text information, and screen change information based on the running screen. Feature extraction is performed on the scene information, object information, text information, and change information to generate scene features, object features, text features, and change features.

9. The method according to claim 7, characterized in that, The step of generating the bullet screen content corresponding to the interactive bullet screen based on the emotion type includes: Based on the multimodal features, the scene features corresponding to the running screen are extracted; The bullet screen content corresponding to the interactive bullet screen is generated based on the scene features and the emotion type.

10. The method according to claim 9, characterized in that, The step of generating the bullet screen content corresponding to the interactive bullet screen based on the scene features and the emotion type includes: Based on the aforementioned scene scene features, scene scene classification is performed to generate scene categories; The bullet screen content corresponding to the interactive bullet screen is generated based on the scenario category and the emotion type.

11. The method according to claim 10, characterized in that, The step of generating the bullet screen content corresponding to the interactive bullet screen based on the scenario category and the emotion type includes: When the scenario category is the first scenario category, the bullet screen content database is obtained, and the bullet screen content corresponding to the interactive bullet screen is obtained from the bullet screen content database based on the emotion type; When the scenario category is the second scenario category, a preset large language generation model is invoked to generate the corresponding bullet screen content based on the emotion type.

12. The method according to claim 11, characterized in that, The training process of the preset large language generation model includes: Retrieve historical bullet comment data posted by the target object; Obtain the initial large language generation model, and update the parameters of the initial large language generation model based on the historical bullet screen data to generate the preset large language generation model.

13. The method according to claim 12, characterized in that, After obtaining the historical bullet screen data posted by the target object, the process also includes: Obtain the associated historical bullet screen data published by related objects related to the target object; The process of obtaining an initial large language generation model and updating the parameters of the initial large language generation model based on the historical bullet screen data to generate a preset large language generation model includes: Obtain the initial large language generation model, and update the parameters of the initial large language generation model based on the historical bullet screen data and the associated historical bullet screen data to generate the preset large language generation model.

14. The method according to claim 11, characterized in that, The training process of the preset large language generation model includes: Obtain the target object's feedback information on the interactive bullet comments; The parameters of the preset large language generation model are updated based on the interactive bullet comments and the feedback information.

15. The method according to claim 7, characterized in that, The interactive bullet screen display process includes: A second preset display effect is obtained based on the emotion type of the interactive bullet comments; The interactive bullet screen is rendered on the device display interface based on the second preset display effect.

16. The method according to claim 1, characterized in that, The bullet screen display process includes: Obtain a third preset display effect, and add a bullet screen overlay layer to the device display interface based on the third preset display effect; The interactive bullet comments are rendered in the bullet comment overlay layer.

17. A bullet screen processing device, characterized in that, include: The first display unit is used to display the application interface of the simulator application on the device display interface. The application interface displays the running screen of the target application and the bullet screen control control. The target application runs in the virtual operating system provided by the simulator application. The second display unit is used to display interactive bullet comments on the device display interface in response to the triggering operation of the bullet comment control control. The content of the interactive bullet comments is generated based on the analysis results obtained from the sentiment analysis of the running screen.

18. An electronic device comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the bullet screen processing method according to any one of claims 1 to 16.

19. A storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the bullet screen processing method according to any one of claims 1 to 16.

20. A computer program product comprising a computer program that is read and executed by a processor of an electronic device, causing the electronic device to perform the bullet screen processing method according to any one of claims 1 to 16.

Images