Processing method and device for anchor information of live broadcast room, live broadcast system and electronic equipment

Abstract

The invention relates to a live broadcast room anchor information processing method and device, a live broadcast system, electronic equipment and a computer readable storage medium, and the processing method comprises the steps: displaying a live broadcast room cover and an AI assistant control of a corresponding live broadcast room when a user enters a home page of the live broadcast room; in response to an operation of browsing the live broadcast cover by the user, obtaining a target live broadcast room cover through the AI assistant control, and generating an anchor analysis request for analyzing an anchor of the corresponding target live broadcast room; an anchor analysis request is sent to the server through the AI assistant control, and the server associates and maps the anchor analysis request and pre-stored live broadcast room interaction information to obtain anchor information; receiving anchor information returned by the server through the AI assistant control and displaying the anchor information; therefore, the user can quickly obtain the anchor information of the live broadcasting room, the interaction cost of the user and the live broadcasting room is reduced, the interactivity of the live broadcasting room is enhanced, and the daily live volume of the live broadcasting platform is improved.

Description

Technical Field

[0001] The present application relates to the technical field of network live broadcasting, and in particular to a method, device, live broadcasting system, electronic device and computer-readable storage medium for processing anchor information in a live broadcasting room. Background Art

[0002] With the development of network technology, live streaming has been used by most Internet users. With its intuitive, immediacy and interactivity in content and form, live streaming has played an important role in promoting flexible employment, promoting economic and social development, and enriching the spiritual and cultural life of the people. Anchors can better display their talents in live broadcasts, thereby helping more anchors realize their self-worth.

[0003] In the current live broadcast platform, when users need to know the host information of the live broadcast room, they usually need to enter the live broadcast room to watch the live broadcast, browse the public screen information, view the host profile card and other multi-dimensional information to understand the host information, which makes the interactive process of users obtaining the host information complicated and inefficient. The live broadcast platform cannot timely identify the user's intention and push the appropriate host information to the user, making it difficult for users to deeply understand the host and reducing their enthusiasm for interacting with the host. Although artificial intelligence (AI) technology has been widely introduced in live broadcasts to assist in the processing of live broadcast related information to improve the efficiency of live broadcasts, the commonly used AI summary and refinement of content information mostly relies on users interacting with AI through voice and text input. The interactive cost for users to obtain the expected host information of the live broadcast room through AI is still high, which is easy to affect the interactivity of the live broadcast room and the daily active users of the live broadcast platform. Summary of the invention

[0004] Based on this, it is necessary to provide a method, device, live broadcast system, electronic device and computer-readable storage medium for processing the anchor information of the live broadcast room, so as to improve the efficiency of obtaining the anchor information of the live broadcast room through AI and reduce the interaction cost.

[0005] A method for processing anchor information in a live broadcast room, comprising:

[0006] When the user enters the homepage of the live broadcast room, the live broadcast room cover and AI assistant controls of the corresponding live broadcast room are displayed;

[0007] In response to the user's operation of browsing the live broadcast cover, the target live broadcast room cover is obtained through the AI ​​assistant control, and an anchor analysis request for analyzing the anchor of the corresponding target live broadcast room is generated;

[0008] Sending an anchor analysis request to the server through the AI ​​assistant control; wherein the server associates and maps the anchor analysis request with the pre-stored live broadcast room interaction information to obtain the anchor information;

[0009] Receive the anchor information returned by the server through the AI ​​assistant control and display it.

[0010] In one embodiment, the displaying of the live broadcast room cover and AI assistant controls of the corresponding live broadcast room includes:

[0011] In response to the user's operation of sliding up and down to browse the live broadcast cover, the live broadcast room cover located in the first line of the homepage and fully exposed is set as the preview live broadcast room cover, and the live broadcast room preview is played in real time on the preview live broadcast room cover; and the AI ​​assistant control is displayed on the upper layer of the homepage.

[0012] In one embodiment, generating an anchor analysis request for analyzing an anchor in a corresponding target live broadcast room includes:

[0013] When it is detected that the user clicks the AI ​​assistant control, the user's voice interaction input information is collected through the AI ​​assistant control and converted into interactive text;

[0014] Perform semantic recognition on the interactive text to obtain executable instructions or instruction sets; generate an anchor analysis request for analyzing the anchor of the corresponding target live broadcast room according to the instructions or instruction sets.

[0015] In one embodiment, when it is detected that a user clicks on an AI assistant control, the method further includes:

[0016] If the AI ​​assistant control does not recognize the user's voice interaction input information within the set first time, the AI ​​assistant control will be restored to the default state;

[0017] If the user inputs a piece of voice interaction input information, and the AI ​​assistant control does not recognize new voice interaction input information within the set second time, an anchor analysis request is generated to analyze the anchor of the corresponding target live broadcast room.

[0018] In one embodiment, the performing semantic recognition on the interactive text to obtain an executable instruction or instruction set includes:

[0019] Performing semantic recognition on the interactive text to obtain action keywords and analysis objects of the live broadcast analysis task, and generating executable instructions or instruction sets according to the action keywords and analysis objects;

[0020] The server associates and maps the anchor analysis request with pre-stored live broadcast room interaction information to obtain anchor information, including:

[0021] The server determines a target anchor according to the analysis object, and obtains anchor information of the target anchor by performing association mapping and matching from pre-stored live broadcast room interaction information of the target anchor according to the action keyword.

[0022] In one embodiment, in response to the user browsing the live broadcast cover, obtaining the target live broadcast room cover through the AI ​​assistant control includes:

[0023] In response to the user's operation of sliding up and down to browse the live broadcast cover, when the user stays on the homepage without taking any action for more than a third time period, the live broadcast room cover that is located in the first line of the homepage and is completely exposed is set as the target live broadcast room cover, and a flowing light stroke is added around the target live broadcast room cover.

[0024] In one embodiment, the method for processing the live broadcast room anchor information further includes:

[0025] The guidance is displayed in a pop-up bubble manner through the AI ​​assistant control; wherein the guidance is instruction information for guiding the user to view the anchor information on the cover of the target live broadcast room;

[0026] If the user clicks on the interactive area in the guide bubble, a host analysis request is generated to analyze the host of the corresponding target live broadcast room;

[0027] If the user does not click on the guidance bubble within the fourth time period, the bubble display is cancelled.

[0028] In one embodiment, receiving the anchor information returned by the server through the AI ​​assistant control and displaying it includes:

[0029] Receive the anchor information sent by the server through the AI ​​assistant control, and use the half window to display different types of anchor information content in a structured manner;

[0030] If the user clicks on the non-half-window area, the page displaying the half-window will be closed. If the user clicks on the interactive button of the AI ​​assistant control, it will jump to execute the preset live interactive operation.

[0031] In one embodiment, the server is configured to pre-store the voice information of the live broadcast room host and digitize it; summarize the content, rhythm and / or intonation of the voice information to obtain a host analysis information data set and store it in a classified manner; receive the host analysis request of the AI ​​assistant control, and associate and map the key information carried in the host analysis request with the host analysis information data set to obtain the host information.

[0032] In one embodiment, the server is configured to collect live interaction data of the live broadcast room over a period of time in the past, input the live interaction data into a first language model to generate summary information, and input the summary information into a second language model to generate recommended speeches.

[0033] In one embodiment, the server is configured to collect chat text data from a live broadcast room, fuse the chat text data with a summary prompt template and input it into a first language model to generate training summary information and fine-tune the model, and fuse the training summary information with a speech prompt template and input it into a second language model to generate training recommended speeches and fine-tune the model.

[0034] A device for processing anchor information in a live broadcast room, comprising:

[0035] The display module is used to display the live broadcast room cover and AI assistant controls of the corresponding live broadcast room when the user enters the live broadcast room homepage;

[0036] A request module, used to respond to a user's operation of browsing a live broadcast cover, obtain a target live broadcast room cover through an AI assistant control, and generate an anchor analysis request for analyzing the anchor of the corresponding target live broadcast room;

[0037] A response module, used to send an anchor analysis request to a server through an AI assistant control; wherein the server associates and maps the anchor analysis request with pre-stored live broadcast room interaction information to obtain anchor information;

[0038] The display module is used to receive and display the anchor information returned by the server through the AI ​​assistant control.

[0039] A live broadcast system, comprising: an anchor terminal, an audience terminal and a live broadcast server; wherein the anchor terminal and the audience terminal are respectively connected to the live broadcast server via a communication network;

[0040] The anchor terminal is used to access the anchor in the live broadcast room and collect the anchor's live video stream and upload it to the live broadcast server;

[0041] The live broadcast server is used to forward the live broadcast video stream of the anchor to the audience end, and receive the analysis data request from the audience end, and use the live broadcast room anchor information processing method to generate anchor information and display it;

[0042] The audience terminal is used for audience users who access the live broadcast room, receive the live video stream of the anchor for playback, and display the anchor information.

[0043] An electronic device, comprising:

[0044] one or more processors;

[0045] Memory;

[0046] One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, and the one or more programs are configured to execute the steps of the method for processing the anchor information of the live broadcast room.

[0047] A computer-readable storage medium stores at least one instruction, at least one program, a code set or an instruction set, wherein the at least one instruction, the at least one program, the code set or the instruction set is loaded by the processor and executes the steps of the method for processing the anchor information of the live broadcast room.

[0048] The technical solution of the present application is to display the live broadcast room cover and AI assistant control of the corresponding live broadcast room when the user enters the live broadcast room homepage; when the user browses the live broadcast cover, the target live broadcast room cover is obtained through the AI ​​assistant control, and an anchor analysis request for analyzing the anchor of the corresponding target live broadcast room is generated and sent to the server. The server associates and maps the anchor analysis request with the pre-stored live broadcast room interaction information to obtain the anchor information and sends it to the AI ​​assistant control for display; this technical solution facilitates users to quickly obtain the anchor information of the live broadcast room, reduces the interaction cost between users and the live broadcast room, enhances the interactivity of the live broadcast room, and improves the daily activity of the live broadcast platform.

[0049] Furthermore, based on the large language model, summary information and recommended speeches of the live broadcast room are generated. Through the summary information, you can quickly understand the interactive topics of the current anchor and other viewers, and quickly participate in the chat by sending recommended speeches, thereby greatly improving business indicators such as user retention and usage time. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 This is a schematic diagram of an example live broadcast service application scenario;

[0051] Figure 2 is a flow chart of a method for processing live broadcast room anchor information according to an embodiment;

[0052] Figure 3 This is a schematic diagram of the homepage of a live broadcast room;

[0053] Figure 4 It is a schematic diagram of the AI ​​assistant control showing the listening state of the AI ​​assistant control;

[0054] Figure 5 This is a schematic diagram of the input status of the AI ​​assistant control;

[0055] Figure 6 This is a schematic diagram of the AI ​​assistant control's analyzing status;

[0056] Figure 7 It is a schematic diagram that the AI ​​assistant control guides the user to view;

[0057] Figure 8 It is a schematic diagram showing the analysis results in a half window;

[0058] Fig. 9 An example is a schematic diagram of the anchor information display;

[0059] Fig.10 This is an example of a user actively triggering the analysis of anchor information flow chart;

[0060] Fig.11 This is an example of a flow chart of a user passively triggering the analysis of anchor information;

[0061] Fig.12 is a flow chart of generating summary information and recommending speeches according to an embodiment;

[0062] Fig.13 This is a schematic diagram of an example low-rank adaptive algorithm;

[0063] Fig.14 is a schematic diagram of the structure of a device for processing anchor information in a live broadcast room according to an embodiment;

[0064] Fig.15 This is a schematic diagram of the live broadcast system structure;

[0065] Fig.16 is a block diagram of an example electronic device. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0067] In the embodiments of the present application, the words "first", "second", etc. are used to distinguish between identical or similar items with substantially the same effects and functions. "at least one" means one or more, and "multiple" means two or more. For example, multiple objects refer to two or more objects. "Include" or "comprises" and similar words mean that the information appearing before "include" or "comprises" covers the information listed after "include" or "comprises" and its equivalents, and does not exclude other information. "And / or" mentioned in the embodiments of the present application indicates that three types of relationships may exist. The character " / " generally indicates that the objects associated before and after are in an "or" relationship.

[0068] The technical solution provided in the embodiments of the present application can be applied to Figure 1 In the application scenario of the method related to this application shown in FIG. Figure 1This is a schematic diagram of an example of a live broadcast service application scenario. The live broadcast system may include a live broadcast server, an anchor terminal and an audience terminal. The anchor terminal and the audience terminal communicate with the live broadcast server through a communication network, so that the anchor at the anchor terminal and the audience users at the audience terminal can perform real-time network live broadcast. Among them, for the anchor terminal and the audience terminal, their terminal devices can be but are not limited to various personal computers, laptops, smart phones and tablets. The live broadcast server can be implemented by an independent server or a server cluster composed of multiple servers.

[0069] In order to facilitate users to quickly obtain the host information of the live broadcast room, the technical solution of the embodiment of the present application is that when using AI-assisted live broadcast, when the user browses the live broadcast cover, the AI ​​assistant control selects the target live broadcast room cover and sends an anchor analysis request to the server. The server pre-stores the host's live broadcast room interaction information, and the AI ​​assistant control is used to conveniently obtain the host information from the server, thereby realizing convenient interaction between the user and the live broadcast room, thereby enabling the user to quickly understand the live broadcast information and enhancing the interactivity of the live broadcast room.

[0070] refer to Figure 2 As shown, Figure 2 The flowchart of a method for processing live broadcast room anchor information according to an embodiment includes the following steps:

[0071] Step S10, when the user enters the live broadcast room homepage, the live broadcast room cover and AI assistant controls of the corresponding live broadcast room are displayed.

[0072] In this step, the user enters the homepage of the live broadcast room through the live broadcast platform. The homepage of the live broadcast room displays the cover of the live broadcast room in the form of multiple small windows, and the AI ​​assistant control is displayed on the homepage. Among them, the AI ​​assistant is a live broadcast auxiliary tool configured based on AI artificial intelligence technology. It is configured to have processing capabilities for voice, text, pictures, etc. It can connect search, large models, intelligent bodies, etc., and can realize diversified search and data processing capability requirements.

[0073] In one embodiment, the method of displaying the live broadcast room cover and AI assistant control of the corresponding live broadcast room in step S10 includes:

[0074] In response to the user's operation of sliding up and down to browse the live broadcast cover, the live broadcast room cover located in the first line of the homepage and fully exposed is set as the preview live broadcast room cover, and the live broadcast room preview is played in real time on the preview live broadcast room cover; and the AI ​​assistant control is displayed on the upper layer of the homepage.

[0075] Specifically, users can browse the live broadcast cover by sliding up and down, such as Figure 3 As shown, Figure 3This is a page diagram of an example live broadcast room homepage, which includes six live broadcast room covers. The four live broadcast room covers on the homepage and fully exposed are the top four. At this time, the two live broadcast room covers in the first row (shown in the square box in the figure) can be set as preview live broadcast room covers, and then the live broadcast room preview can be played in real time on the preview live broadcast room cover, that is, the current screen content of the current live broadcast room is played, and other live broadcast room covers display cover pictures. At the same time, the AI ​​assistant control is displayed in the lower right corner of the homepage page. In the default state, the AI ​​assistant image can be a two-dimensional avatar.

[0076] Step S20, in response to the user's operation of browsing the live broadcast cover, obtain the target live broadcast room cover through the AI ​​assistant control, and generate an anchor analysis request for analyzing the anchor of the corresponding target live broadcast room.

[0077] In this step, when the user browses the live broadcast cover, the AI ​​assistant control selects the target live broadcast room cover from the live broadcast room covers on the homepage, and then generates an anchor analysis request for the anchor of the target live broadcast room to perform analysis based on the user's browsing operation.

[0078] In one embodiment, the method of generating an anchor analysis request for analyzing the anchor of the corresponding target live broadcast room in step S20 includes:

[0079] S201, when it is detected that the user clicks the AI ​​assistant control, the user's voice interaction input information is collected through the AI ​​assistant control and converted into interactive text.

[0080] S202, performing semantic recognition on the interactive text to obtain executable instructions or instruction sets; generating an anchor analysis request for analyzing the anchor of the corresponding target live broadcast room according to the instructions or instruction sets.

[0081] As an embodiment, when performing semantic recognition on interactive text, action keywords (such as analysis, understanding, etc.) and object keywords (anchor's nickname, anchor, anchor type, etc.) can be identified, and then executable instructions or instruction sets are generated based on these action keywords and analysis objects; in addition, other keywords can be set and introduced into the instructions as needed.

[0082] Furthermore, in order to optimize the processing logic of the AI ​​assistant control, when it is detected that the user clicks the AI ​​assistant control, if the AI ​​assistant control does not recognize the user's voice interaction input information within the set first time, the AI ​​assistant control will be restored to the default state; if the user enters a voice interaction input information, and the AI ​​assistant control does not recognize new voice interaction input information within the set second time, an anchor analysis request is generated to analyze the anchor of the corresponding target live broadcast room.

[0083] Exemplarily, the user clicks on the AI ​​assistant control, such as Figure 4 As shown, Figure 4 This is a schematic diagram of the AI ​​Assistant control in the listening state; the AI ​​Assistant control expands to the left to enter the listening state, and displays the copy: "You speak, I listen." If no new voice is recognized within 5 seconds, it will automatically collapse to the right and return to the default state. The user can also actively click the close button to end the current process. When the user is inputting voice, such as Figure 5 As shown, Figure 5 This is a schematic diagram of the AI ​​assistant control input state; the AI ​​assistant control expands from the listening state to the left and changes to the input state. When the AI ​​assistant control receives the voice information input by the user and cannot recognize new voice information within 3 seconds, the AI ​​assistant enters the analyzing state. Figure 6 As shown, Figure 6 This is a diagram of the AI ​​assistant control in the analysis state; the bubble text displays: "XX (AI assistant nickname) is helping you analyze". After the analysis is completed, the bubble disappears and the analysis result half window is pulled up.

[0084] As in the scheme of the above-mentioned embodiment, a scheme is provided for users to actively interact and analyze anchor information. The user clicks on the AI ​​assistant control to turn on the voice input interaction function. The AI ​​assistant control collects the user's voice interaction input information. The AI ​​assistant control converts the voice interaction input information into interactive text, and then performs semantic recognition based on the interactive text to obtain the user's intention, and then generates executable instructions or instruction sets. The AI ​​assistant control generates an anchor analysis request for analyzing the anchor in the target live broadcast room according to the instructions or instruction sets, and then requests the server to analyze and summarize the anchor information in the live broadcast room.

[0085] In one embodiment, in response to the user browsing the live broadcast cover, obtaining the target live broadcast room cover through the AI ​​assistant control in step S20 may include:

[0086] In response to the user's operation of sliding up and down to browse the live broadcast cover, when the user stays on the homepage without taking any action for more than a third time period, the live broadcast room cover that is located in the first line of the homepage and is completely exposed is set as the target live broadcast room cover, and a flowing light stroke is added around the target live broadcast room cover.

[0087] In this step, when the user slides up and down to browse the live broadcast cover, if the user stays on the homepage without taking any action for a certain period of time, the AI ​​assistant control will automatically set the live broadcast room cover that is located in the first row of the homepage and is fully exposed as the target live broadcast room cover, and add a flowing light stroke around the target live broadcast room cover.

[0088] Furthermore, in order to guide and instruct users to better use this function, guidance instructions are displayed in the form of pop-up bubbles through the AI ​​assistant control. The guidance instructions are instruction information to guide users to view the anchor information on the cover of the target live broadcast room; if the user clicks on the interactive area in the guidance bubble, an anchor analysis request is generated to analyze the anchor of the corresponding target live broadcast room; if the user does not click on the guidance bubble within the fourth time period, the bubble display is cancelled.

[0089] For example, the user passively triggers the analysis process. When the user stays on the homepage for more than 5 seconds without taking any action, the live broadcast cover that is ranked first in the page and fully exposed will be displayed with a streamer stroke, and the AI ​​assistant will actively push a message to guide the user to view it, such as Figure 7 As shown, Figure 7 This is a schematic diagram of the AI ​​assistant control guiding users to view. The push content includes: guiding copy "This is the current preview of the anchor information, come and have a look", close button, 5s countdown copy, anchor live preview cover, and details button. The bubble disappears when any of the following conditions are met: for example: the user does not operate after 5s, manually clicks the close button, or the analysis cover does not meet the conditions of being in the first row of the sort and fully exposed. If the user clicks the information push content area / details button, the analysis half window will be opened.

[0090] As in the scheme of the above-mentioned embodiment, a host information analysis task function is provided which is passively triggered by the user. The AI ​​assistant control determines the user's intention and selects the target live broadcast room cover according to the current user's operation of browsing the live broadcast room cover. The user can implement interactive operations by clicking the interactive area in the guide bubble, or cancel subsequent operations by countdown.

[0091] Step S30, sending an anchor analysis request to the server through the AI ​​assistant control; wherein the server associates and maps the anchor analysis request with the pre-stored live broadcast room interaction information to obtain the anchor information.

[0092] In this step, the AI ​​assistant control requests the server for the anchor information of the live broadcast room. After receiving the anchor analysis request, the server associates and maps the pre-stored live broadcast room interaction information with the executable instructions or instruction sets in the request to obtain the anchor information.

[0093] In one embodiment, when the server performs association mapping to obtain anchor information, it can determine the target anchor according to the analysis object, and perform association mapping and matching from the live broadcast room interaction information pre-stored by the target anchor according to the action keywords to obtain the anchor information of the target anchor.

[0094] Specifically, the server associates and matches the keywords with the background data set, retrieves the current host and historical voice information to analyze the voice content, rhythm, intonation, etc. The object keyword can be used to determine which host needs to be analyzed, and the action keyword can be used to know the host information the user wants to obtain.

[0095] In one embodiment, in order to realize the association matching function, the server can perform corresponding operations, pre-store the voice information of the host in the live broadcast room and digitize it; summarize the content, rhythm, and intonation of the voice information to obtain the host analysis information data set and store it in categories. Exemplarily, the background server can build a sound collection module, a speech recognition module, and a semantic analysis module, which are respectively used for sound collection, speech recognition, and semantic analysis functions; build an emotional speech database, pre-store the live interactive information of the host in the live broadcast room, and digitize and pre-process the information. When receiving the host analysis request of the AI ​​assistant control, the server associates and maps the key information carried in the host analysis request with the host analysis information data set to obtain the host information.

[0096] As in the solution of the above-mentioned embodiment, the server can obtain the anchor summary analysis information of the live broadcast room by retrieving the current and historical voice information of the anchor to analyze the voice content, rhythm, intonation, etc., so that users can quickly and conveniently understand the anchor information of the live broadcast room.

[0097] Step S40: Receive the anchor information returned by the server through the AI ​​assistant control and display it.

[0098] In this step, the AI ​​assistant control receives the anchor information returned by the server, and then displays it through the corresponding window page. Users can enter the live broadcast room to watch the live broadcast based on the anchor information they know.

[0099] In one embodiment, the method of receiving and displaying the anchor information returned by the server through the AI ​​assistant control in step S40 includes:

[0100] S401, receiving the anchor information sent by the server through the AI ​​assistant control, and using a half window to perform structured display of different types of anchor information content.

[0101] Specifically, the AI ​​assistant control receives the anchor information sent by the server, then opens a half window, and combines different types of anchor information content for structured display of different structures, so that users can easily obtain the required anchor information.

[0102] S402: If the user clicks on the non-half-window area, the page displaying the half-window is closed; if the user clicks on the interactive button of the AI ​​assistant control, the user jumps to execute the preset live interactive operation.

[0103] As in the solution of the above-mentioned embodiment, the AI ​​assistant control can be used to conveniently summarize and analyze the live interactive information of the anchor in the live broadcast room, as well as present it visually in real time, so as to identify the user's intention and push appropriate anchor information to the user, so that the user can more actively understand the anchor in depth and increase the enthusiasm for interacting with the anchor.

[0104] For example, when the analysis is completed, the system automatically pulls up a half window to display the content, such as Figure 8 As shown, Figure 8 It is a schematic diagram of the half-window display analysis results; it mainly includes:

[0105] (1) Description: The following is the analysis result of {host nickname}

[0106] (2) Live broadcast status area: includes live broadcast cover preview area, host nickname, broadcast topic, and follow button; when a user clicks on the live broadcast cover preview area, the window is half closed and the corresponding live broadcast room is entered. If the user clicks on the follow button, the host is directly followed. If the user has already followed the host, the follow button text is changed to "Already followed". If the analysis result is multiple hosts, this area is displayed in a horizontal arrangement, and horizontal sliding is supported to switch between views.

[0107] like Fig. 9 As shown, Fig. 9 An example is a schematic diagram of the anchor information display; if the first anchor is currently displayed, swiping to the right has no response, and if the last anchor is currently displayed, swiping to the left has no response. When the swipe successfully switches the live broadcast status area, the anchor information overview area and the current chat area at the bottom are replaced in real time.

[0108] (3) Host information overview: The AI ​​assistant control retrieves the host’s profile card, historical gameplay of frequently used functions, and voice recognition of the host’s current and historical voice content and emotions, and displays them in a structured form in the form of a table: basic information, host characteristics, host’s favorite functions, and host’s hobbies.

[0109] (4) Currently in the chat area: The AI ​​assistant control retrieves and summarizes the voice information of the current live broadcast room within the past 10 minutes, and refines it into a short summary (the short summary must contain verbs + nouns), which is presented in a flowing barrage from right to left.

[0110] Based on the method for processing the live broadcast room anchor information provided by this application, for example, Fig.10 As shown, Fig.10This is an example of a flowchart in which a user actively triggers the analysis of anchor information. When the user clicks on the AI ​​assistant control and the AI ​​assistant enters the listening state, the user enters information through voice interaction. The voice recognition module converts the voice into text, and then the semantic recognition module translates the text into executable instructions or instruction sets, which are transmitted to the server and associated with the pre-stored information of the background anchor. If the AI ​​assistant control does not recognize the voice information within 3 seconds, the AI ​​assistant returns to the default state. If the user inputs a voice and the AI ​​assistant control does not recognize the new voice information within 3 seconds, an anchor analysis request is automatically generated. The AI ​​assistant control loads the anchor information from the server and displays it in a half-window format. The user can view the analysis results. When the user exits the viewing, the page returns to the homepage.

[0111] Based on the method for processing the live broadcast room anchor information provided by this application, for example, Fig.11 As shown, Fig.11 This is an example of a flowchart of user passively triggering the analysis of anchor information. When the user slides to browse the live broadcast cover, the live broadcast room cover that is fully exposed in the first row automatically plays the live broadcast preview. When the user stays on the homepage without operating for more than 5 seconds, a streamer stroke is added around the live broadcast room cover that previews the live broadcast. At the same time, the AI ​​assistant guides the user to view the anchor analysis content of the currently playing video cover through a pop-up bubble; an anchor analysis request is automatically generated, and the AI ​​assistant control loads the anchor information from the server and displays it in a half-window format. If the user clicks on the interactive area in the guide bubble, an anchor analysis request is generated to analyze the anchor in the corresponding target live broadcast room; if the user does not operate the guide bubble, the bubble disappears automatically after 5 seconds.

[0112] In one embodiment, when a new user enters the live broadcast room, he or she is often unable to immediately participate in the current topic chat because he or she is not clear about the chat topic between the host and other audience users and does not understand the current live broadcast content.

[0113] Based on this, this embodiment provides a technical solution for automatically generating a live broadcast room summary and recommended speeches based on a large language model and displaying them to audience users, so that new users can quickly understand the interactive topics between the current anchor and other audience users through the summary; specifically, the server can be configured to collect live broadcast interaction data of the live broadcast room over a period of time in the past, input the live broadcast interaction data into the first large language model to generate summary information, and input the summary information into the second large language model to generate recommended speeches.

[0114] As an example, refer to Fig.12 As shown, Fig.12 The flowchart of generating summary information and recommended speech of an embodiment may include the following:

[0115] S501, collecting chat text data in the live broadcast room.

[0116] The live interactive data in the live broadcast room mainly includes the host's voice signal and the text information entered by the audience on the public screen of the live broadcast room. In order to obtain the host's clean speaking content, the original voice signal can be separated from the human voice to remove the background music, and then the voice signal can be converted into text data using speech recognition technology.

[0117] For example, if it is a live broadcast scene, the host in the live broadcast room and the live broadcast host connected to the microphone will be distinguished. For the text information entered by the audience on the public screen, text processing is performed, including filtering out non-text symbols such as emoticons, downsampling excessive speeches, etc., so as to obtain high-quality public screen text information.

[0118] For example, assuming that the public screen speech set G has a total of m speeches, and the maximum number of samples is n, then the sampled speech set H is:

[0119]

[0120] Among them, RS n (G) is to randomly sample n speeches from G. When m≤n, the entire speech set G is selected; when m>n, n speeches are randomly sampled from G; the chat text data of the host and the audience are obtained, and then the chat text data of the host and the audience are aligned according to the timestamp; then the data is manually annotated, and the summary and recommended speeches are manually filled in for each group of host and audience text data.

[0121] Preferably, in order to significantly improve the efficiency of annotation, a pre-trained large language model can be used to automatically generate pre-annotated summaries and recommended speeches, and then the annotation results can be manually corrected.

[0122] S502, fusing the chat text data with a summary prompt template and inputting it into a first language model to generate training summary information and fine-tune the model; fusing the training summary information with a speech prompt template and inputting it into a second language model to generate training recommended speech and fine-tune the model.

[0123] For example, high-quality summary prompt templates and speech prompt templates are obtained through continuous experiments and optimization of prompt input; the chat text data is filled into the summary prompt template and input into the first language model to obtain the corresponding summary information, and the summary information is filled into the speech prompt template and input into the second language model to obtain the corresponding recommended speech; in actual operation, the chat text data is directly connected to the prompt template as the input of the large language model. Each summary contains two versions: a detailed description and an outline description, which can be used for thumbnail display and expanded display of the live room summary respectively.

[0124] refer to Fig.13 As shown, Fig.13It is a schematic diagram of an example low-rank adaptation algorithm. The low-rank adaptation (LoRA) algorithm is used to fine-tune the first and second large language models. First, a bypass composed of a dimensionality reduction matrix A and a dimensionality increase matrix B is added to the pre-trained model weight matrix W. The product BA of the two matrices in the bypass has the same dimension as W. During training, the pre-trained weight W is fixed, and only A and B are trained. A is randomly initialized as a Gaussian distribution, and B is initialized as 0. During output, the product BA is superimposed with the W parameters. During inference, the input is multiplied by the superimposed parameters as the output. Assume the pre-trained weight matrix The update can be expressed as:

[0125] W0 + ΔW = W0 + BA

[0126] Where The rank r << min(d, k). During training, W0 remains fixed. The number of parameters (d×r + r×k) of the matrix BA that needs to be fine-tuned is much smaller than the number of parameters (d×k) of the pre-trained weight W0, achieving that only a very small number of parameters need to be trained to complete the fine-tuning of the model. During forward inference, W0 and ΔW will be multiplied by the same input x and finally added together:

[0127] h = W0x + ΔWx = W0x + BAx

[0128] The first and second large language models have the same model structure and pre-trained weights, and are fine-tuned using different data for LoRA training. Among them, during the training phase, the first large language model uses the chat text data of the host and the audience's speeches as input, and the manually annotated summary as the target output to train the LoRA weights. The second large language model uses the manually annotated summary and the recommended speech as input and target output respectively to train the LoRA weights. After training, the first and second large language models have different LoRA weights. During the inference phase, the chat text data is input into the first large language model to generate summary information, and then the generated summary information is input into the second large language model to generate recommended speeches, which can continuously generate summary information and recommended speeches, with higher timeliness.

[0129] According to the solution of the above embodiment, new users can quickly understand the interaction topics of the current host and other audiences through the summary information, and participate in the chat by clicking to send recommended speeches, thereby significantly improving business indicators such as user retention and usage duration.

[0130] The following describes an embodiment of the device for processing the host information in the live broadcast room.

[0131] As Fig.14 shown Fig.14 It is a schematic diagram of the structure of the device for processing the host information in the live broadcast room according to an embodiment, including:

[0132] Display module 10, used to display the live broadcast room cover and AI assistant controls of the corresponding live broadcast room when the user enters the live broadcast room homepage;

[0133] The request module 20 is used to obtain the target live broadcast room cover through the AI ​​assistant control in response to the user's operation of browsing the live broadcast cover, and generate an anchor analysis request for analyzing the anchor of the corresponding target live broadcast room;

[0134] The response module 30 is used to send an anchor analysis request to the server through the AI ​​assistant control; wherein the server associates and maps the anchor analysis request with the pre-stored live broadcast room interaction information to obtain the anchor information;

[0135] The display module 40 is used to receive and display the anchor information returned by the server through the AI ​​assistant control.

[0136] The processing device for the live broadcast room anchor information of this embodiment can execute a method for processing the live broadcast room anchor information provided by the embodiment of the present application. The implementation principle is similar. The actions performed by each module in the processing device for the live broadcast room anchor information in each embodiment of the present application correspond to the steps in the method for processing the live broadcast room anchor information in each embodiment of the present application. For the detailed functional description of each module in the processing device for the live broadcast room anchor information, please refer to the description in the corresponding method for processing the live broadcast room anchor information shown in the previous text, which will not be repeated here.

[0137] An embodiment of the live broadcast system is described below.

[0138] The live broadcast system of this embodiment refers to Fig.15 As shown, Fig.15This is a schematic diagram of the structure of an example live broadcast system. The system framework may include an anchor terminal, a live broadcast server and an audience terminal. The audience terminal and the anchor terminal may access the live broadcast platform through a client installed on an electronic device. For example, the anchor terminal and the audience terminal may be computer devices such as a PDA, a smart phone, a tablet computer, a desktop computer or a laptop computer, etc., but this is not limited. They may also be software modules of an application. The live broadcast server may be implemented using an independent server or a server cluster consisting of multiple servers. Among them, the anchor end is connected to the anchor user of the live broadcast room, collects the anchor's audio and video and uploads it to the live broadcast server, and the anchor interacts with the audience through the anchor end; the live broadcast server receives the audio and video data uploaded by the anchor end, forwards the live audio and video data to the audience end, and at the same time receives the analysis request for the anchor of the live broadcast room sent by the audience end, and uses the processing method of the anchor information of the live broadcast room provided in the aforementioned embodiment to analyze the anchor information accordingly; the audience end is connected to the audience user of the live broadcast room, plays the live audio and video content, and the audience interacts with the anchor through the audience end. After the user on the audience end enters the homepage of the live broadcast room, in response to the user's operation of sliding up and down to browse the cover of the live broadcast room, the anchor information analysis request of the target live broadcast room cover is sent to the live broadcast server, the anchor information is loaded from the live broadcast server and UI interaction information is generated for display on the client interface.

[0139] Embodiments of electronic devices and computer-readable storage media are described below.

[0140] The present application provides a technical solution for an electronic device to implement functions related to a method for processing anchor information in a live broadcast room. The electronic device of this embodiment includes one or more processors, a memory; one or more applications, wherein the one or more applications are stored in the memory and configured to be executed by one or more processors, and the one or more programs are configured for the steps of the method for processing anchor information in a live broadcast room of any embodiment.

[0141] like Fig.16 As shown, Fig.16 1 is a block diagram of an exemplary electronic device; the electronic device may be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, etc. The electronic device 100 may include one or more of the following components: a processing component 102, a memory 104, a power component 106, a multimedia component 108, an audio component 109, an input / output (I / O) interface 112, a sensor component 114, and a communication component 116.

[0142] The processing component 102 generally controls the overall operation of the electronic device 100, such as operations associated with display, telephone calls, data communications, camera operations, and recording operations.

[0143] The memory 104 is configured to store various types of data to support operations in the electronic device 100, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0144] The power supply component 106 provides power to various components of the electronic device 100 .

[0145] The multimedia component 109 includes a screen that provides an output interface between the electronic device 100 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). In some embodiments, the multimedia component 108 includes a front camera and / or a rear camera.

[0146] The audio component 109 is configured to output and / or input audio signals.

[0147] I / O interface 112 provides an interface between processing component 102 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to, a home button, a volume button, a start button, and a lock button.

[0148] The sensor component 114 includes one or more sensors for providing various aspects of status assessment for the electronic device 100. The sensor component 114 may include a proximity sensor configured to detect the presence of a nearby object without any physical contact.

[0149] The communication component 116 is configured to facilitate wired or wireless communication between the electronic device 100 and other devices. The electronic device 100 can access a wireless network based on a communication standard, such as WiFi, an operator network (such as 2G, 3G, 4G or 5G), or a combination thereof.

[0150] The present application provides a technical solution of a computer-readable storage medium, which is used to implement functions related to a method for processing live broadcast room anchor information. The computer-readable storage medium stores at least one instruction, at least one program, code set or instruction set, and the at least one instruction, at least one program, code set or instruction set is loaded by a processor and executed by the method for processing live broadcast room anchor information of any embodiment.

[0151] In an exemplary embodiment, the computer-readable storage medium may be a non-temporary computer-readable storage medium including instructions, such as a memory including instructions. For example, the non-temporary computer-readable storage medium may be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like.

[0152] The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A method for processing anchor information in a live broadcast room, characterized in that: include: When the user enters the homepage of the live broadcast room, the live broadcast room cover and AI assistant controls of the corresponding live broadcast room are displayed; In response to the user's operation of browsing the live broadcast cover, the target live broadcast room cover is obtained through the AI ​​assistant control, and an anchor analysis request for analyzing the anchor of the corresponding target live broadcast room is generated; Sending an anchor analysis request to the server through the AI ​​assistant control; wherein the server associates and maps the anchor analysis request with the pre-stored live broadcast room interaction information to obtain the anchor information; Receive the anchor information returned by the server through the AI ​​assistant control and display it.

2. The method for processing live broadcast room anchor information according to claim 1, characterized in that: The live broadcast room cover and AI assistant controls for displaying the corresponding live broadcast room include: In response to the user's operation of sliding up and down to browse the live broadcast cover, the live broadcast room cover located in the first line of the homepage and fully exposed is set as the preview live broadcast room cover, and the live broadcast room preview is played in real time on the preview live broadcast room cover; and the AI ​​assistant control is displayed on the upper layer of the homepage.

3. The method for processing live broadcast room anchor information according to claim 1, characterized in that: Generate an anchor analysis request for analyzing the anchor of the corresponding target live broadcast room, including: When it is detected that the user clicks the AI ​​assistant control, the user's voice interaction input information is collected through the AI ​​assistant control and converted into interactive text; Perform semantic recognition on the interactive text to obtain executable instructions or instruction sets; generate an anchor analysis request for analyzing the anchor of the corresponding target live broadcast room according to the instructions or instruction sets.

4. The method for processing the live broadcast room anchor information according to claim 3, characterized in that: When it is detected that the user clicks on the AI ​​assistant control, it also includes: If the AI ​​assistant control does not recognize the user's voice interaction input information within the set first time, the AI ​​assistant control will be restored to the default state; If the user inputs a piece of voice interaction input information, and the AI ​​assistant control does not recognize new voice interaction input information within the set second time, an anchor analysis request is generated to analyze the anchor of the corresponding target live broadcast room.

5. The method for processing live broadcast room anchor information according to claim 3, characterized in that: The performing semantic recognition on the interactive text to obtain an executable instruction or instruction set includes: Performing semantic recognition on the interactive text to obtain action keywords and analysis objects of the live broadcast analysis task, and generating executable instructions or instruction sets according to the action keywords and analysis objects; The server associates and maps the anchor analysis request with pre-stored live broadcast room interaction information to obtain anchor information, including: The server determines a target anchor according to the analysis object, and obtains anchor information of the target anchor by performing association mapping and matching from pre-stored live broadcast room interaction information of the target anchor according to the action keyword.

6. The method for processing live broadcast room anchor information according to claim 1, characterized in that: In response to the user browsing the live broadcast cover, the target live broadcast room cover is obtained through the AI ​​assistant control, including: In response to the user's operation of sliding up and down to browse the live broadcast cover, when the user stays on the homepage without taking any action for more than a third time period, the live broadcast room cover that is located in the first line of the homepage and is completely exposed is set as the target live broadcast room cover, and a flowing light stroke is added around the target live broadcast room cover.

7. The method for processing the live broadcast room anchor information according to claim 6, characterized in that: Also includes: The guidance is displayed in a pop-up bubble manner through the AI ​​assistant control; wherein the guidance is instruction information for guiding the user to view the anchor information on the cover of the target live broadcast room; If the user clicks on the interactive area in the guide bubble, a host analysis request is generated to analyze the host of the corresponding target live broadcast room; If the user does not click on the guidance bubble within the fourth time period, the bubble display is cancelled.

8. The method for processing live broadcast room anchor information according to claim 1, characterized in that: The AI ​​assistant control receives the anchor information returned by the server and displays it, including: Receive the anchor information sent by the server through the AI ​​assistant control, and use the half window to display different types of anchor information content in a structured manner; If the user clicks on the non-half-window area, the page displaying the half-window will be closed. If the user clicks on the interactive button of the AI ​​assistant control, it will jump to execute the preset live interactive operation.

9. The method for processing the live broadcast room anchor information according to claim 8, characterized in that: The server is configured to pre-store the voice information of the anchor in the live broadcast room and digitize it; summarize the content, rhythm and / or intonation of the voice information to obtain an anchor analysis information data set and store it in categories; receive the anchor analysis request of the AI ​​assistant control, and associate and map the key information carried in the anchor analysis request with the anchor analysis information data set to obtain the anchor information.

10. The method for processing the live broadcast room anchor information according to any one of claims 1 to 9, characterized in that: The server is configured to collect live interaction data of the live broadcast room over a period of time in the past, input the live interaction data into a first language model to generate summary information, and input the summary information into a second language model to generate recommended speeches.

11. The method for processing the live broadcast room anchor information according to claim 9, characterized in that: The server is configured to collect chat text data from the live broadcast room, fuse the chat text data with a summary prompt template and input it into a first language model to generate training summary information and fine-tune the model, and fuse the training summary information with a speech prompt template and input it into a second language model to generate training recommended speeches and fine-tune the model.

12. A device for processing anchor information in a live broadcast room, characterized in that: include: The display module is used to display the live broadcast room cover and AI assistant controls of the corresponding live broadcast room when the user enters the live broadcast room homepage; A request module, used to respond to a user's operation of browsing a live broadcast cover, obtain a target live broadcast room cover through an AI assistant control, and generate an anchor analysis request for analyzing the anchor of the corresponding target live broadcast room; A response module, used to send an anchor analysis request to a server through an AI assistant control; wherein the server associates and maps the anchor analysis request with pre-stored live broadcast room interaction information to obtain anchor information; The display module is used to receive and display the anchor information returned by the server through the AI ​​assistant control.

13. A live broadcast system, characterized in that: include: An anchor terminal, an audience terminal and a live broadcast server; wherein the anchor terminal and the audience terminal are respectively connected to the live broadcast server via a communication network; The anchor terminal is used to access the anchor in the live broadcast room and collect the anchor's live video stream and upload it to the live broadcast server; The live broadcast server is used to forward the live broadcast video stream of the anchor to the audience end, and receive the analysis data request from the audience end, and use the method for processing the anchor information of the live broadcast room according to any one of claims 1 to 11 to generate the anchor information and display it; The audience terminal is used for audience users who access the live broadcast room, receive the live video stream of the anchor for playback, and display the anchor information.

14. An electronic device, characterized in that: The electronic device comprises: one or more processors; Memory; One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, and the one or more programs are configured to execute the steps of the method for processing the live broadcast room anchor information as described in any one of claims 1-11.

15. A computer-readable storage medium, characterized in that: The storage medium stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set are loaded by the processor and execute the steps of the method for processing the live broadcast room anchor information as described in any one of claims 1-8.

Images