Live broadcast data processing method, device and system, electronic equipment and storage medium

Abstract

The invention discloses a live broadcast data processing method. The method comprises the steps of determining a live broadcast service opening strategy used for determining a target live broadcast service opening opportunity; based on the live broadcast service opening strategy, determining an opening opportunity of the target live broadcast service; when the starting opportunity of the target live broadcast service arrives, generating a live broadcast service starting instruction for instructing a cloud to start the target live broadcast service to a user side; sending the live broadcast service starting instruction to the cloud; determining a target commodity shelving strategy, wherein the target commodity shelving strategy is a strategy used for representing shelving of the target commodity in the target live broadcast room; generating a target commodity shelving instruction for indicating shelving of the target commodity in the target live broadcast room based on the target commodity shelving strategy; and sending the target commodity shelving instruction to the cloud. According to the scheme provided by the invention, the broadcasting time and the commodity shelving strategy can be generated and automatically triggered, automatic broadcasting and commodity shelving are realized, and the development of automatic live broadcasting is promoted.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a live data processing method, device, system, electronic device and storage medium. Background Art

[0002] With the rapid development of Internet technology, live broadcasting has become an important way of information dissemination and entertainment. For example, merchants use live broadcasting to promote and display products and interact with the audience in real time, thereby attracting buyers and facilitating orders. With the traditional live broadcasting method, merchants complete the entire live broadcast by operating the start of the broadcast, product introduction, product listing, and closing the broadcast. In the traditional live broadcasting process, each link requires real-time operation and participation by merchants or anchors, and over-reliance on manual operation, which increases the manpower cost invested by merchants in live broadcasting services.

[0003] Therefore, how to reduce the live broadcast process’s reliance on manual operations and reduce merchants’ manpower costs for live broadcasts is a problem that needs to be solved at present. Summary of the invention

[0004] The present application provides a live broadcast data processing method, device, system electronic equipment and computer-readable storage medium, which are used to reduce the dependence of the live broadcast process on manual operation, reduce labor costs, and help improve the degree of automation of live broadcast.

[0005] In a first aspect, the present application provides a method for processing live broadcast data, including: determining a live broadcast service start-up strategy for determining a target live broadcast service start-up timing, wherein the target live broadcast service is a digital human live broadcast service for a target merchant provided to a user terminal; based on the live broadcast service start-up strategy, determining a start-up timing of the target live broadcast service; when the start-up timing of the target live broadcast service arrives, generating a live broadcast service start-up instruction for instructing a cloud to start the target live broadcast service to a user terminal; sending the live broadcast service start-up instruction to the cloud; determining a target product listing strategy, wherein the target product listing strategy is a strategy for indicating listing a target product in a target live broadcast room; generating a live broadcast service listing instruction for instructing a cloud to list a target product in a target live broadcast room based on the target product listing strategy. The target product listing instruction for listing the target product in the target live broadcast room is displayed; the target product listing instruction is sent to the cloud; wherein, the target product listing strategy is determined, including: obtaining the target product listing strategy configured by the target merchant, or generating the target product listing strategy; the target product listing strategy is generated, including: generating listing trigger data for the target product according to the target live broadcast material data, and generating the target product listing strategy according to the listing trigger data for the target product, wherein the listing trigger data for the target product is the data output in the target live broadcast room for triggering the listing of the target product in the target live broadcast room.

[0006] Optionally, the method further includes: determining live broadcast material data for the target live broadcast service as target live broadcast material data; and sending the target live broadcast material data to the cloud.

[0007] Optionally, determining the live broadcast material data for the target live broadcast service includes: obtaining the target live broadcast material data configured by the target merchant; or generating the target live broadcast material data.

[0008] Optionally, it also includes: obtaining a material data request message issued by the cloud for requesting to obtain the target live broadcast material data; sending the target live broadcast material data to the cloud includes: sending the target live broadcast material data to the cloud in response to the material data request message.

[0009] Optionally, it also includes: obtaining an anchor interaction information acquisition request message issued by the cloud for requesting anchor interaction information for target user interaction information, wherein the target user interaction information is interaction information issued by the user end for the target live broadcast service; obtaining anchor interaction information for the target user interaction information; and returning the anchor interaction information for the target user interaction information to the cloud in response to the anchor interaction information acquisition request message.

[0010] Optionally, generating the target product listing strategy based on the listing trigger data for the target product includes: generating a listing trigger data detection task, and obtaining the listing trigger data detection task identifier, wherein the listing trigger data detection task is used to indicate whether the data output in the target live broadcast room contains the listing trigger data for the target product; generating a target product listing instruction generation task, and obtaining a target product listing instruction generation task identifier; generating a target product listing instruction sending task, and obtaining a target product listing instruction sending task identifier; recording the listing trigger data detection task identifier, the target product listing instruction generation task identifier, and the target product listing instruction sending task identifier into the task data item of the product listing strategy template to generate the target product listing strategy.

[0011] In a second aspect, the present application provides a method for processing live broadcast data, including: obtaining a live broadcast service start instruction for instructing to start a target live broadcast service for a user terminal, the target live broadcast service is a digital human live broadcast service for a target merchant, the live broadcast service start instruction is generated based on a live broadcast service start strategy, the live broadcast service start strategy is a strategy for determining the timing of starting the target live broadcast service, the live broadcast service start strategy is pre-configured by the target merchant or generated by a live broadcast service end that provides the digital human live broadcast service; in response to the live broadcast service start instruction, generating a digital human live broadcast data stream for the target merchant as a target live broadcast data stream; and converting the target live broadcast data stream into a digital human live broadcast data stream. Push to the user end; obtain a target commodity listing instruction for indicating to list the target commodity in the target live broadcast room corresponding to the target live broadcast data stream, the target commodity is the target commodity that the target merchant can provide to the user, the target commodity listing instruction is generated based on the target commodity merchant policy, the target commodity listing policy is a policy for indicating listing the target commodity in the target live broadcast room, the target commodity listing policy is pre-configured by the target merchant or generated by the live broadcast service end; obtain the live broadcast data of the target commodity in the target live broadcast room as the target commodity live broadcast data; add the target commodity live broadcast data to the target live broadcast data stream.

[0012] Optionally, it also includes: obtaining live broadcast material data for the target live broadcast service provided by the live broadcast server as target live broadcast material data, wherein the target live broadcast material data is pre-set by the target merchant or generated by the live broadcast server; the digital human live broadcast data stream for the target merchant includes: generating a digital human live broadcast data stream for the target merchant based on the target live broadcast material data.

[0013] Optionally, it also includes: sending a material data request message to the live broadcast server to request obtaining the target live broadcast material data; obtaining the live broadcast material data of the digital human live broadcast service of the target merchant provided by the live broadcast server includes: obtaining the target live broadcast material data provided by the live broadcast server in response to the material data request message.

[0014] Optionally, it also includes: obtaining interactive information sent by the user end for the target live broadcast service as target user interactive information; generating response information for the target user interactive information according to the interactive strategy for the target live broadcast service as target anchor interactive information, wherein the interactive strategy is pre-configured by the target merchant or generated by the live broadcast service end; and sending the target anchor interactive information to the user end.

[0015] Optionally, generating response information for the target user interaction information based on the interaction strategy for the target live broadcast service includes: determining whether interaction information matching the target user interaction information exists in pre-set user interaction information, and using the interaction information matching the target user interaction information in the pre-set user interaction information as matching interaction information; if so, obtaining the anchor interaction information corresponding to the matching interaction information as matching anchor interaction information; and generating response information for the target user interaction information based on the matching anchor interaction information.

[0016] Optionally, generating response information for the target user interaction information according to the interaction strategy for the target live broadcast service includes: sending an anchor interaction information acquisition request message to the live broadcast service end for requesting anchor interaction information for the target user interaction information; obtaining anchor interaction information for the target user interaction information provided by the live broadcast service end in response to the anchor interaction information acquisition request message; and generating response information for the target user interaction information according to the anchor interaction information for the target user interaction information.

[0017] In the third aspect, the present application provides a live broadcast data processing device, including: a live broadcast service startup strategy determination unit, used to determine the live broadcast service startup strategy, the live broadcast service startup strategy is used to determine the startup timing of the target live broadcast service, and the target live broadcast service is a digital human live broadcast service for target merchants provided to the user end; a startup timing determination unit, used to determine the startup timing of the target live broadcast service based on the live broadcast service startup strategy; a live broadcast service startup instruction generation unit, when the startup timing of the target live broadcast service arrives, generates a live broadcast service startup instruction for instructing the cloud to start the target live broadcast service to the user end; a live broadcast service startup instruction sending unit, sending the live broadcast service startup instruction to the cloud.

[0018] In a fourth aspect, the present application provides a live broadcast data processing device, including: a live broadcast service start instruction obtaining unit, used to obtain a live broadcast service start instruction for instructing to start a target live broadcast service to a user end, the target live broadcast service is a digital human live broadcast service for a target merchant, and the live broadcast service start instruction is generated based on a live broadcast service start strategy, and the live broadcast service start strategy is a strategy for determining the timing of starting the target live broadcast service, and the live broadcast service start strategy is pre-configured by the target merchant or generated by a live broadcast service end that provides the digital human live broadcast service; a live broadcast data stream generating unit, used to generate a digital human live broadcast data stream for the target merchant as a target live broadcast data stream in response to the live broadcast service start instruction; a live broadcast data stream pushing unit, used to push the target live broadcast data stream to the user end.

[0019] In a fifth aspect, the present application provides a live broadcast data processing system, including a cloud and a live broadcast service end: the cloud is used to obtain a live broadcast service start instruction, the live broadcast service start instruction is used to instruct to start a target live broadcast service to a user end, the target live broadcast service is a digital human live broadcast service for a target merchant, the live broadcast service start instruction is generated based on a live broadcast service start strategy, the live broadcast service start strategy is a strategy for determining the start time of the target live broadcast service, the live broadcast service start strategy is pre-configured by the target merchant or generated by a live broadcast service end that provides a digital human live broadcast service; in response to the live broadcast service start instruction, a digital human live broadcast data stream for the target merchant is generated as a target live broadcast data stream; the target live broadcast data stream is pushed to the user end; the live broadcast service end is used to determine a live broadcast service start strategy, the live broadcast service start strategy is used to determine the start time of the target live broadcast service, the target live broadcast service is a digital human live broadcast service for a target merchant provided to the user end; based on the live broadcast service start strategy, the start time of the target live broadcast service is determined; when the start time of the target live broadcast service arrives, a live broadcast service start instruction is generated for instructing the cloud to start the target live broadcast service to the user end; and the live broadcast service start instruction is sent to the cloud.

[0020] In a sixth aspect, the present application provides an electronic device comprising: a processor, a memory, and computer program instructions stored in the memory and executable on the processor; when the processor executes the computer program instructions, it implements the method described in any one of the first and second aspects.

[0021] In a seventh aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by a processor, they are used to implement the method described in any one of the first and second aspects.

[0022] Compared with the prior art, this application has the following advantages: The live broadcast data processing method provided in the embodiment of the present application is applied to the live broadcast server. The server can determine the live broadcast service startup strategy to control the timing of starting the live broadcast. When the startup timing arrives, it automatically generates a startup instruction to instruct the live broadcast service to start on time, thereby realizing the automation of starting the live broadcast service; the live broadcast server can also generate a product listing strategy to control when and which product to put on the shelf during the live broadcast process. During the broadcast, the live broadcast server will generate corresponding product listing trigger data based on the live broadcast material data. The listing trigger data can trigger the listing effect of the corresponding product, thereby realizing the automatic listing of the corresponding product during the live broadcast service process.

[0023] The live broadcast data processing method provided in the embodiment of the present application generates a live broadcast service start instruction based on the live broadcast service start strategy, and in response to the live broadcast service start instruction, opens the digital human live broadcast service for the target merchant to the user end. Since the live broadcast service start instruction is generated based on the live broadcast service start strategy, and the live broadcast service start strategy can be pre-configured by the merchant or generated by the live broadcast service end, the target product listing strategy is also generated by the live broadcast service end. Therefore, by adopting the live broadcast data processing method provided in the embodiment of the present application, the merchant does not need to create a live broadcast room, send a start instruction, and other real-time operations when the broadcast starts, and the start instruction can be automatically generated and the broadcast task can be completed. There is no need for manual identification and manual listing of the products when they are listed, and the listing of the corresponding products can be achieved, thereby promoting the development of automated live broadcasting. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A schematic diagram of an application scenario provided for this embodiment.

[0025] Figure 2 A flowchart of the live broadcast data processing method provided in this embodiment.

[0026] Figure 3 A flowchart of the live broadcast data processing method provided in Example 1.

[0027] Figure 4 This is a schematic diagram of the merchant terminal configuration interface provided in Example 1.

[0028] Figure 5 This is a flow chart of the method for automatically putting goods on shelves provided in Example 1.

[0029] Figure 6 This is a flow chart of the interactive information response method provided in Example 1.

[0030] Figure 7 This is a flowchart of the live broadcast data processing method provided in Example 2.

[0031] Figure 8 A schematic diagram of a live broadcast data processing device provided in Example 3.

[0032] Fig. 9 A schematic diagram of a live broadcast data processing device provided in Example 4.

[0033] Fig.10 A schematic diagram of a live broadcast data processing system provided in Example 5.

[0034] Fig.11 This is a schematic diagram of an electronic device provided in Example 6. DETAILED DESCRIPTION

[0035] In order to enable those skilled in the art to better understand the technical solution of the present application, the present application is described clearly and completely below in conjunction with the drawings in the embodiments of the present application. However, the present application can be implemented in many other ways different from the following description. Therefore, based on the embodiments provided by the present application, all other embodiments obtained by those of ordinary skill in the art without creative work should fall within the scope of protection of the present application.

[0036] It should be noted that the terms "first", "source domain", "third", etc. in the claims, description and drawings of the present application are used to distinguish similar objects and are not used to describe a specific order or sequence. The data used in this way are interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including", "having" and their variants are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0037] Live broadcasting is an innovative way of information dissemination and is very popular among consumers. In real life, live broadcasting has become the main form of product promotion and introduction for many electronic product service platforms. Electronic product service platforms display products to the audience through live broadcasting. The anchor and the user can understand the merchant (store) and products through interactive methods such as likes, attention, comments, and bullet screens, which also increases the purchase conversion rate. Electronic product service platforms can provide good support for live broadcasting. Electronic product service platforms can support live broadcasting platforms. Live broadcasting platforms are also electronic product service platforms that can recommend products. Therefore, unless otherwise specified in the following text, live broadcast platforms and electronic product service platforms (referred to as e-commerce platforms) have the same meaning. At present, most real anchors broadcasting live broadcasts throughout the whole process is a more common form of live broadcasting. The anchors display and introduce products online independently and respond to users' questions by obtaining bullet screens. This traditional form of live broadcasting requires a real anchor to display and introduce the products. At the same time, a field control personnel is required to set up the live broadcast in real time and interact with the anchor to achieve the live broadcast effect. It can be seen that traditional live broadcasting usually requires at least two people to cooperate and complete it together. Since traditional live broadcasting requires a large amount of manpower, and when the interaction is hot, the interactive information is refreshed quickly, and the host cannot capture and answer it in time, which obviously reduces the user experience. In addition, due to the limitations of the working time and physical strength of the real anchor in the traditional live broadcasting format, the live broadcast time cannot be adjusted all day or arbitrarily. Therefore, the existing technology proposes digital human live broadcasting, which uses digital humans to replace real anchors.

[0038] Digital human live broadcast, also known as virtual anchor or digital human live broadcast, is a new form of live broadcast that combines virtual humans (i.e. digital humans, virtual characters or virtual anchors) with video live broadcast. Due to the relatively novel image, form and interactive experience of digital humans, digital human live broadcast has gradually been widely used in recent years. In addition, digital human live broadcast can overcome the problems of high labor costs and poor interactive experience in traditional live broadcast to a certain extent. At present, the digital human live broadcast in the prior art only replaces the anchor with a digital human, but the live broadcast preparation, control operation, personalization issues, etc. still require real-time online human participation. In other words, the digital human live broadcast process in the prior art cannot be completed by digital humans alone. In addition, whether it is a real anchor or a digital anchor, a special local live broadcast device is required for face capture or real-time operation, which increases the equipment investment cost. In order to solve the above problems existing in digital human live broadcast, the present application proposes a live broadcast data processing method to further reduce the labor cost of live broadcast. The live broadcast data processing method proposed in the present application generates a live broadcast service start instruction through a live broadcast service start strategy pre-configured by the target merchant or generated by the live broadcast service end providing the digital human live broadcast service, so as to instruct the user end to start the target live broadcast service. The live broadcast data processing method of the present application can realize automated live broadcast and reduce manpower and equipment costs.

[0039] In order to facilitate understanding of the method embodiments of the present application, the application scenarios of the embodiments of the present application are introduced. Figure 1 , Figure 1Schematic diagram of the application scenario provided for this embodiment. The application scenario mainly includes a live broadcast service end 101, a user end 102, a cloud end 103 and an anchor end 104. In simple terms, the live broadcast service end 101 provides live broadcast services, and the user end 102 can be used to watch live broadcasts. The live broadcast service end 101 is mainly responsible for the registration and scheduling of cloud machines, timed on / off broadcasts, and provides bullet screen interactive capabilities. The user end 102 refers to the user's live broadcast end, including mobile client, computer client and other portals for watching live broadcasts. The user end 102 supports users to send bullet screen interactions, and provides functions such as product pop-up cards in the live broadcast room. The product pop-up card pops up a product card in the live broadcast room interface, and users can jump to the product details page to purchase by clicking on the product card. The cloud end 103 is used to provide cloud machines, interactive services, cloud data support and cloud streaming services. The anchor end 104 supports receiving merchants to configure products in advance, pre-generate materials, configure interactions and configure on / off broadcast time. The anchor end 104 can refer to the live broadcast anchor console. The cloud 103 includes multiple cloud machines, each of which can run multiple service instances, and each service instance can provide different services to different merchants. The cloud can also push streams to the user end. A cloud push stream cluster is deployed on the cloud. The cloud push stream cluster is a technical architecture based on cloud computing architecture, which is specifically used for video streaming media push (pushing stream) and distribution. Therefore, the cloud can push the live broadcast room to the user end. Since the cloud resources required for live broadcast may vary according to factors such as material size, quantity, live broadcast cache, etc., the cloud can also expand or shrink its own cloud resources to support the live broadcast service of the target merchant to improve resource utilization and data processing efficiency. With the collaboration of the live broadcast service end 101, the user end 102, the cloud 103, and the anchor end 104, the live broadcast is completed. In this process, the merchant does not need to prepare a local device specifically for live broadcast, thereby effectively saving the merchant's machine cost. The application scenario can also include a merchant end, which can display the live broadcast configuration interface provided by the anchor end for the merchant, and the information set by the merchant through the live broadcast configuration interface can be synchronized to the anchor end.

[0040] Please see Figure 2 , Figure 2 A flowchart of the live broadcast data processing method provided in this embodiment. Figure 2 It can also reflect Figure 1 The data processing method provided in the embodiment of the present application mainly includes the following steps: S201: Cloud machine registration. The prepared cloud operating system image is deployed on the cloud machine, and the cloud machine starts multiple service instances. The cloud machine registers machine information with the live broadcast server for subsequent machine scheduling by the live broadcast server. The service instance may include different services. The registered machine information includes the IP address of the machine, the unique identifier of the machine, the resource status, the load status, etc. The live broadcast server schedules the operation of the service instance through the registered machine information. The service instance is allocated to the merchant so that the digital human can be put into use, and the merchant can obtain the rights and interests of configuring the service instance (for example, digital human live broadcast). In this step, a flexible and dynamic cloud service scheduling system can be realized through the production of cloud operating system images, multi-instance deployment of cloud machines, and registration and scheduling mechanism of machine information. The scheduling mechanism may include two scheduling methods, one is to directly schedule cloud machines through the cloud, and the other is to schedule cloud machines through cloud scheduling servers, and then the cloud servers schedule cloud machines to provide service instances. The above cloud scheduling method can not only improve the availability and scalability of services, but also optimize resource utilization and reduce operating costs.

[0041] S202: Merchant information is sent. The live broadcast server sends the merchant information and binds the machine registered in the cloud to the merchant for subsequent merchant material pulling and broadcast event recording.

[0042] After ensuring that the cloud machine has registered with the server and stored its own machine information when it starts, the merchant information is sent to the cloud machine, and the merchant is bound to the cloud machine, and the merchant's allocation information is recorded. The merchant information includes the merchant's basic information, such as merchant ID, merchant name, etc., as well as the machine information it uses. After the merchant is bound to the cloud machine, the merchant can pull the required materials from the cloud. For example, the merchant may need to pull some live broadcast material templates or default materials. When the merchant is ready to start live broadcast, the server needs to detect the start event and respond.

[0043] S203: Pulling material information. The cloud initiates a request to pull material information to the live broadcast server, which is used for the subsequent cloud to pre-download the digital human materials configured by the merchant, the product explanation materials and interactive materials of the bound merchant, etc.

[0044] S204: Pre-download of materials. To improve the response time of the broadcast and streaming, the cloud needs to prepare the materials required for the live broadcast room in advance. In the previous step, the cloud initiates a request to pull materials to the live broadcast server, and the live broadcast server will respond to it. Then the cloud can pre-download the material information in advance. The merchant materials include the merchant's digital human materials.

[0045] S205: Generate scheduled on / off broadcast information. The on / off broadcast information can be pre-set by the merchant, such as the on / off broadcast time and broadcast session number and other information related to on / off broadcast, or the on / off broadcast time and session number information can be generated by the live broadcast server. After the on / off broadcast information is pre-configured or generated, the live broadcast server will record it for subsequent on / off broadcast event delivery. It should be noted that this step has no necessary sequential relationship with the previous steps. Taking a live broadcast that is about to begin as an example, this step can be performed simultaneously with S203, or before, after, or simultaneously with any step after S203 and before S206.

[0046] S206: Sending a start event. According to the start time set by the merchant, the live broadcast server sends a start event to the cloud at a fixed time.

[0047] S207: Create a scene. After receiving the start event sent by the live broadcast server, the cloud creates a cloud rendering scene and renders the corresponding digital human screen and materials in real time. The live broadcast screen is composed of the scene screen and the digital human screen. By rendering the scene and the digital human separately, support is provided for the subsequent user end to display the live broadcast screen.

[0048] S208: Start streaming. When the live broadcast image is rendered, the cloud pushes high-definition, low-latency streams to GRTN (Global Real-Time Network) through real-time communication services to meet the live broadcast needs.

[0049] S209: Receive barrage messages. After the streaming is successful, the cloud can receive barrage messages from the user side. Barrage messages are a type of interactive messages. In addition, it may also receive interactive messages such as comments, likes, gifts, etc. sent by users watching the live broadcast. The strategy for receiving barrage messages can be determined according to the frequency of users sending barrages. For example, when the frequency of receiving barrage messages is low, the batch of barrages can be pulled in batches at intervals. When the frequency of sending barrage messages is high, pull them in real time to avoid missing certain barrage messages. The transmission path of barrage messages can be from the user side to the cloud, then to the live broadcast server, and then to the anchor side.

[0050] S210: Regular expression matching. The cloud matches the acquired bullet screen message with the pre-set bullet screen database to search for the reply content in the bullet screen database. The bullet screen database is set in advance, which stores the questions that users may ask and the corresponding answers, as well as some preset actions and texts after users like, interact, and place orders. Since the bullet screen database is pre-set with previously used questions and answers, in this step, it is only necessary to search for the corresponding answer in the bullet screen database without additional processing. Therefore, the regular expression matching method can improve the efficiency of digital humans automatically replying to bullet screen messages.

[0051] S211: Digital human replies. If the regular match is successful, the cloud will reply the reply content obtained by the match to the user end through the digital human, realizing the interaction between the digital human and the users in the live broadcast room. If the regular match is successful in the previous step, it means that the answer corresponding to this barrage message exists in the preset barrage database, so it is used as the reply content and feedback is given to the user through the digital human.

[0052] S212: Barrage input. If the regular expression matching fails in the above steps, the cloud needs to input the barrage content and the user information sending the barrage into the large language model LLM, and use the text output by the large language model as the reply content of the barrage message, so as to provide feedback to the user based on the reply content.

[0053] LLM (Large Language Model) refers to a pre-trained model based on deep learning technology, especially in the field of natural language processing, with a large number of parameters and the ability to handle complex language tasks. Through training on a large amount of text data, the large language model can generate, understand, translate, answer questions, summarize, write and other language-related tasks. This step uses the LLM model to generate the reply content to the target user's bullet message.

[0054] The large language model can be trained based on the copywriting of massive commodity data and reference commodity vertical information. Specifically, a data set containing multiple commodity vertical information and commodity data copywriting examples on the electronic commodity service platform is obtained as an initial sample; the initial sample is scored to generate a quality score for the initial sample; the initial sample with a quality score greater than a preset score is selected as a training sample; the copywriting generation model is pre-trained based on the training sample to obtain a pre-trained copywriting generation model; and a copywriting generation model is generated based on the pre-trained copywriting generation model. Among them, the vertical information can be divided according to commodity type, application scenario, and user characteristics; the scoring of the initial sample to generate a copywriting score as the quality score of the initial sample includes: using natural language processing and other technologies to score the initial sample, and the score may include at least one of the following methods: language fluency, creativity, emotional resonance, and target matching. The pre-training of the copywriting generation model based on the training sample includes: pre-training the training sample using a deep neural network architecture; and enabling the copywriting generation model to learn language structure and grammatical rules through self-supervised learning tasks.

[0055] The method of generating a trained copy generation model based on a pre-trained copy generation model includes: labeling the training samples to obtain labeling information of the training samples; obtaining training samples and preference labels generated by multiple models; fine-tuning the initial pre-trained copy generation model using a supervised fine-tuning technique SFT (Supervised Fine-Tuning), wherein the fine-tuning is based on the labeling information of the training samples; optimizing the fine-tuned model using a direct preference optimization DPO (Direct Preference Optimization) method, and improving the generation quality and accuracy of the model by adjusting the preference order of the model output; and obtaining a copy generation model.

[0056] Therefore, when a new bullet screen message is received, the bullet screen message is input into the copy generation model for inference to generate reply content. Model inference refers to the use of the model to predict new and unseen data after model training is completed. The core task of inference is to apply the trained model to make decisions or predictions without adjusting the parameters of the model. Specifically, the product rights and interests information of the electronic commodity service platform is integrated into the prompt information (the input prompt of the model), so that the product rights and interests information of the electronic commodity service platform is added to the copy generated by the large language model. By adding the product rights and interests information of the electronic commodity service platform (such as coupons, full-reduction activities, red envelopes, etc.) of the electronic commodity service platform to the prompt, the model can provide personalized product recommendations based on the user's needs or historical behavior. For example, if the user has certain specific rights and interests (such as points, membership discounts, etc.), the model can recommend products or preferential activities that meet the user's rights and interests based on this information. The method generates copy containing the electronic commodity service platform commodity rights and interests information by adding the electronic commodity service platform commodity rights and interests information to the prompt information of the copy generation model, thereby improving the diversity of the copy content, enriching the information related to the copy rights and interests, making the copy content seen by the user more personalized, and thus improving the user experience.

[0057] S213: Parse the reply content. The server sends the reply content output by LLM to the cloud. The cloud determines whether the LLM call is successful and parses the reply content output by LLM so that the anchor can reply according to the reply content.

[0058] S214: Barrage reply. Based on the reply content and target user information, the cloud replies to the target user through private barrage chat and digital human explanation. The content of the digital human explanation comes from the copywriting generation model.

[0059] S215: Sending a live broadcast closing message. According to the end time of the live broadcast room pre-configured by the merchant or generated by the live broadcast server, the server sends an end live broadcast event to the cloud to control the end of the live broadcast and close the live broadcast room.

[0060] S216: End streaming. The cloud receives the end live event, ends streaming, and closes the live scene screen.

[0061] S217: Review data. The logs generated during the live broadcast are temporarily stored in the cloud. The host can obtain the logs by sending a request message to the cloud through the live broadcast server. The cloud can also automatically synchronize the logs to the host during or after the live broadcast, and then the host sends them to the merchant. The logs may include bullet screen messages, live broadcast time, visits, topics of interest, etc. The live broadcast server will also analyze and process the data in the logs according to the cloud request content and send them to the merchant for the merchant to refer to the live broadcast data in the logs for optimizing live broadcast, goods and other related services.

[0062] The live broadcast data processing method provided in the embodiment of the present application realizes automated digital human live broadcast by utilizing capabilities such as a large language model based on commodity data, timed on / off broadcast, automatic commodity shelving, and low-latency cloud streaming. This effectively reduces the complexity of product use and helps merchants to automatically start broadcasting. At the same time, merchants do not need additional manpower to participate in the live broadcast process or special local equipment, which effectively saves merchants' live broadcast costs.

[0063] Embodiment 1 Embodiment 1 provides a live broadcast data processing method, and the execution subject of the method can be a cloud.

[0064] Please see Figure 3 , Figure 3 This is a flowchart of a live data processing method provided in Example 1, and the live data processing method includes the following steps: S301: Obtain a live broadcast service start instruction for instructing to start a target live broadcast service for a user terminal, wherein the target live broadcast service is a digital human live broadcast service for a target merchant, and the live broadcast service start instruction is generated based on a live broadcast service start strategy, which is a strategy for determining a start time for the target live broadcast service, and the live broadcast service start strategy is pre-configured by the target merchant or generated by a live broadcast service terminal that provides a digital human live broadcast service.

[0065] This step is used to instruct the user end to automatically start live broadcast at the specified time.

[0066] In this step, the live broadcast service start instruction is sent from the cloud to the live broadcast service end. When the live broadcast service end receives the live broadcast service start instruction, the live broadcast can be started. The live broadcast service start instruction mainly includes the time when the live broadcast service is started, and the time when the live broadcast service is started is determined by the live broadcast service start strategy. The live broadcast service start strategy is pre-configured by the target merchant or generated by the live broadcast service end that provides the digital human live broadcast service. Before this step, the cloud machine and the merchant have been bound through steps such as cloud machine registration and merchant information issuance. The merchant bound to the cloud machine is the target merchant that initiates the live broadcast. The target merchant can pre-configure the live broadcast service start strategy such as product information, interactive settings, and broadcast time on the merchant end. The anchor end receives the configuration information of the merchant end and sends the configuration information to the live broadcast service end. After receiving it, the live broadcast service end records it for subsequent scheduled broadcast. For example, the target merchant enters the time to start the live broadcast on the merchant end and selects the product to be introduced in this live broadcast. These configuration information will be synchronized from the anchor end to the live broadcast service end, and the live broadcast service end will record the configuration information and the corresponding target merchant information. Please refer to Figure 4 , Figure 4 Schematic diagram of the merchant end configuration interface provided in Example 1. In the merchant end configuration interface 400, the merchant can set the parameters related to the switch broadcast, goods, digital human, etc. For example, setting the broadcast start time and the broadcast end time can control the switch broadcast of the live broadcast, and setting the product list can provide the product information on the shelves in the live broadcast. The target merchant configures the live broadcast related data through the configuration interface on the merchant end, and the configured data will be synchronized to the anchor end through the configuration interface, and then sent by the anchor end to the live broadcast service end. Of course, the live broadcast service start strategy can also be generated by the live broadcast service end that provides the digital human live broadcast service. While providing the basic machine registration, scheduling and timed switch broadcast and barrage interaction basic capabilities, the live broadcast service end can also provide digital human live broadcast service and control the digital human to start live broadcast at a predetermined time. In the prior art, the configuration process is determined by the merchant's real anchor to determine the live broadcast content, time, etc. The configuration interface of Example 1 provides the conditions for the realization of automated live broadcast, which can be broadcast according to the merchant's requirements without the merchant's full participation. In addition, the configuration process can also be generated by the live broadcast service end, which further reflects the improvement of the degree of automation.

[0067] S302: In response to the live broadcast service start instruction, a digital human live broadcast data stream for the target merchant is generated as a target live broadcast data stream.

[0068] This step is used to automatically push the live broadcast to the user end.

[0069] When the cloud receives the live broadcast service start instruction sent by the live broadcast service end, it generates the digital human live broadcast content of the target merchant, and then generates the digital human live broadcast data stream for the target merchant as the target live broadcast data stream. The generation of the digital human live broadcast content of the target merchant can refer to step S207. The digital human live broadcast data stream for the target merchant includes: generating the digital human live broadcast data stream for the target merchant according to the target live broadcast material data.

[0070] S303: Push the target live data stream to the user end.

[0071] This step enables users to watch live broadcasts on the user side.

[0072] During the push process, you can set the push parameters according to the merchant's needs, including the push protocol and live stream quality. For example, the target merchant needs to push the live broadcast to GRTN, use the RTMP (Real-Time Messaging Protocol) protocol, and set the video resolution, frame rate and bit rate. The push protocol can be adapted according to the push requirements of the target live platform. In addition to RTMP, there are also HLS / HTTP-FLV (HTTP Live Streaming, HTTP real-time streaming protocol / HTTP Flash Video, HTTP Flash video streaming protocol), WebRTC (Web Real-Time Communication, web real-time communication) and other protocols. The specific selection can be based on the merchant's needs.

[0073] The obtaining of the live broadcast service start instruction for instructing to start the target live broadcast service to the user terminal includes: obtaining the live broadcast service start instruction issued by the live broadcast service end. The live broadcast service end generates the live broadcast service start instruction by generating a live broadcast service start strategy. The live broadcast service end issues a live broadcast service start instruction to instruct the user terminal to start the target live broadcast service. The live broadcast service start strategy includes a live broadcast service start time for starting the target live broadcast service to the user terminal.

[0074] In the prior art, if the target product needs to be put on the shelf in the live broadcast room, the anchor or assistant needs to perform the putting operation at the right time and display the target product card or link. In the first embodiment, the live broadcast data processing method can also provide a function of automatically putting products on the shelf in the live broadcast room. Therefore, the live broadcast data processing method provided in this embodiment can also provide a method for automatically putting products on the shelf. Please refer to Figure 5 , Figure 5 This is a flow chart of the method for automatically putting goods on shelves provided in Example 1.

[0075] Optionally, the live broadcast data processing method provided in the first embodiment further includes the following steps, which can be used to automatically put products on the shelves in the live broadcast room: S501: Obtain a target commodity listing instruction for indicating listing of a target commodity in a target live broadcast room corresponding to the target live broadcast data stream, wherein the target commodity is a target commodity that the target merchant can provide to users, and the target commodity listing instruction is generated based on a target commodity merchant policy, and the target commodity listing policy is a policy for indicating listing of the target commodity in the target live broadcast room, and the target commodity listing policy is pre-configured by the target merchant or generated by the live broadcast server.

[0076] This step is used to provide a target product listing instruction for automatically listing the product.

[0077] Specifically, the target merchant pre-configures or the live broadcast server generates a target product listing strategy, the live broadcast server generates a target product listing instruction and sends it to the cloud, and the cloud obtains the target product listing instruction to instruct the live broadcast room corresponding to the target live broadcast data stream to automatically list the target product. The live broadcast server determines the listing time and conditions of the target product according to the target product listing strategy, and the listing time and conditions may include listing the product when the name of the target product is mentioned in the digital human commentary, the name of the target product is included in the barrage message, the time node, and the information related to the target product mentioned in the digital human commentary, such as functions, matching information with other products, etc.

[0078] S502: Obtain the live broadcast data of the target product in the target live broadcast room as the target product live broadcast data. This step is used to generate the live broadcast data of the target product.

[0079] After determining the listing strategy and listing instructions for the target product, relevant preparations need to be made for the listing of the target product in the live broadcast room. Live broadcast data may include all data in various forms during the live broadcast, such as the information that needs to be displayed for the target product in the live broadcast room and the address and content to jump to after clicking the target product. If the target merchant requires the listing of the target product to be displayed in the live broadcast room in the form of a pop-up card, and the name, service introduction and preview image of the target product are displayed in the card, when the user in the live broadcast room clicks on the card, it will jump to the details page of the target product, which provides the user with detailed product information, and the user can also place an order or add to the purchase on the details page.

[0080] S503: Add the target product live broadcast data to the target live broadcast data stream. This step enables the target product live broadcast data to be pushed to the user end together with other data in the live broadcast room.

[0081] During the live broadcast process, the above steps do not require the participation of the target merchants. The digital human can complete them independently and can also automatically put the target products on the shelves, further providing a basis for the realization of digital human live broadcast automation.

[0082] In order to increase the streaming speed, after the live broadcast data stream is configured, the material can be preloaded: the cloud loads the live broadcast material data from the live broadcast server in advance. Optionally, the live broadcast data processing method also includes: obtaining the live broadcast material data for the target live broadcast service provided by the live broadcast server, as the target live broadcast material data, the target live broadcast material data is pre-set by the target merchant or generated by the live broadcast server. The target live broadcast material data includes digital human materials, explanation materials and interactive materials, etc., and can refer to steps S203-S204 for details. The digital human live broadcast data stream for the target merchant includes: generating a digital human live broadcast data stream for the target merchant based on the target live broadcast material data.

[0083] Before the above-mentioned preloading of the material, it is necessary to initiate a request to the live broadcast server to obtain the target live broadcast material data provided for the material data request message. Optionally, the live broadcast data processing method also includes: sending a material data request message to the live broadcast server to request the target live broadcast material data; the obtaining of the live broadcast material data of the digital human live broadcast service of the target merchant provided by the live broadcast server includes: obtaining the target live broadcast material data provided by the live broadcast server for the material data request message. After the server receives the material data request message sent by the cloud, it provides the live broadcast material data of the target live broadcast service to the cloud. Obtaining the live broadcast material data from the live broadcast server through the cloud and preloading it effectively improves the response time for starting broadcasting and streaming.

[0084] One of the reasons why live streaming is popular is that many people can interact with the anchor in real time. Therefore, the interaction between the anchor and the users who watch the live streaming is an important part of the live streaming process. The users who watch the live streaming communicate with the anchor by sending interactive messages in the form of barrage, and the anchor will respond after receiving the interactive messages sent by the users who watch the live streaming, thus completing the live streaming interaction. Figure 6 , Figure 6 This is a flow chart of the interactive information response method provided in Example 1. Optionally, the live broadcast data processing method further includes the following steps: S601: Obtaining interactive information sent by the user end for the target live broadcast service as target user interactive information. In this step, the cloud obtains interactive information sent by the user end for the target live broadcast service, and the interactive information includes real-time bullet screens, gifts sent, likes, etc. sent by users watching the live broadcast, and the interactive information sent by the target user is used as the target user interactive information. Any user watching the live broadcast in the live broadcast room can be a target user.

[0085] S602: Generate response information for the target user's interactive information as the target anchor's interactive information according to the interactive strategy for the target live broadcast service. The interactive strategy is pre-configured by the target merchant or generated by the live broadcast service. This step is used to provide feedback on the target user's interactive information. After receiving the target user's interactive information in the cloud, the cloud generates response information according to the interactive strategy of the target live broadcast service, and responds to the target user's interactive information to achieve real-time interaction between the target anchor and the target user. Among them, the interactive strategy includes a pre-set barrage database, a text generation model, etc., and the form of feedback and response is not limited. It can be public barrage, private chat barrage, private chat messages, digital human explanations, digital human expressions, body movements, live broadcast room background subtitles and other forms that users like.

[0086] S603: Send the target anchor interaction information to the user terminal. The user terminal receives the target anchor interaction information and obtains feedback of the target user interaction information. Usually, the target anchor interaction information is used to answer questions in the target user interaction information.

[0087] The corresponding information for the target user interaction information in step S602 can be obtained by the first method: The method of generating response information for the target user interaction information according to the interaction strategy for the target live broadcast service includes: determining whether there is interaction information matching the target user interaction information in the pre-set user interaction information, and using the interaction information matching the target user interaction information in the pre-set user interaction information as the matching interaction information; if so, obtaining the anchor interaction information corresponding to the matching interaction information as the matching anchor interaction information; and generating response information for the target user interaction information according to the matching anchor interaction information.

[0088] This method matches the target user interaction information with the pre-set user interaction information. For example, the target user sends a bullet screen and matches the pre-set user interaction information. If there is interaction information matching the bullet screen sent by the target user in the pre-set user interaction information, the interaction information is used as the matching interaction information, and the corresponding host interaction information is obtained as the matching host interaction information. Then, based on the matching host interaction information, response information for the target user interaction information is generated to provide feedback on the target user interaction information. The response information can be a reply and feedback using the host interaction information, or it can be content generated after changing the expression form, expression style, sentence structure, etc. of the host interaction information.

[0089] After receiving the target user interaction information, the cloud can process and respond according to the above steps, or send the received target user interaction information to the live broadcast server and request the live broadcast server to respond to the target user interaction information. Therefore, the response information to the target user interaction information in the above step S502 can also be obtained in the second way: The generating of response information for the target user interaction information according to the interaction strategy for the target live broadcast service includes: sending an anchor interaction information acquisition request message for requesting anchor interaction information for the target user interaction information to the live broadcast service end; obtaining anchor interaction information for the target user interaction information provided by the live broadcast service end in response to the anchor interaction information acquisition request message; and generating response information for the target user interaction information according to the anchor interaction information for the target user interaction information.

[0090] In the second method, after receiving the target user interaction information, the cloud sends a host interaction information acquisition request message to the live broadcast server to request the host interaction information provided by the live broadcast server for the target user interaction information. When the cloud receives the host interaction information provided by the live broadcast server, it generates response information for the target user interaction information based on the host interaction information to complete the interaction between the host and the target user. In practical applications, the first or second method for obtaining response information for the target user interaction information can be used, or the first method for obtaining response information for the target user interaction information can be used as the preferred method, and the second method can be used as the method that can be selected after the match fails in the first case. Combining the judgment conditions and successively using the first and second methods can reduce the data processing pressure of the live broadcast server.

[0091] The above steps provide a method for processing interactive information sent by the target user during live broadcasting. This method can achieve the communication and interaction between the anchor and the viewing user without human participation in the live broadcasting process. The live broadcasting server and the cloud can work together to complete the communication and interaction between the anchor and the viewing user. The process of receiving and replying to the barrage in the above steps can also be understood in conjunction with the contents of steps S209-S214.

[0092] Corresponding to the start of the live broadcast service, the end of the live broadcast is triggered by the live broadcast service closing instruction, and the start of the live broadcast to the end of the live broadcast is a complete live broadcast process. Optionally, the live broadcast data processing method also includes: obtaining a live broadcast service closing instruction for instructing to close the target live broadcast service to the user end, the live broadcast service closing instruction is generated based on the live broadcast service closing strategy, the live broadcast service closing strategy is a strategy for determining the timing of closing the target live broadcast service, the live broadcast service closing strategy is pre-configured by the target merchant or generated by the live broadcast service end; in response to the live broadcast service closing instruction, stop pushing the target live broadcast data stream to the user end. Among them, the obtaining of the live broadcast service closing instruction for instructing to close the target live broadcast service to the user end includes: obtaining the live broadcast service closing instruction issued by the live broadcast service end. The live broadcast service closing strategy includes the live broadcast service closing time for closing the target live broadcast service to the user end.

[0093] Optionally, the live broadcast data processing method also includes: sending a registration request message to the live broadcast server to request that the target device information in the cloud be registered to the live broadcast server; obtaining the target merchant information of the target merchant returned by the live broadcast server in response to the registration request message; binding the target merchant information with the target device information to establish a binding relationship between the target merchant information and the target device information; after obtaining the live broadcast service start instruction, obtaining the target merchant information from the live broadcast service start instruction; obtaining the target device information based on the binding relationship between the target merchant information and the target device information; pushing the target live broadcast data stream to the user end includes: sending a target device scheduling instruction to the target device corresponding to the target device information to instruct the target live broadcast service to be started to the user end, so that the target device pushes the target live broadcast data stream to the user end.

[0094] The live broadcast data processing method provided in Example 1 is based on the live broadcast service opening and closing strategy, and opens and closes the digital human live broadcast service for the target merchant to the user end; based on the target commodity merchant instruction, the target commodity is automatically put on the shelf in the target live broadcast room; based on the interactive strategy of the target live broadcast service, the live broadcast room host can automatically interact with the target user. Using the live broadcast data processing method provided in Example 1, the merchant does not need to create a live broadcast room, send a broadcast instruction, and other real-time operations when starting the broadcast, and can automatically generate a broadcast instruction, complete the broadcast task, and end the live broadcast. At the same time, the target commodity can be automatically put on the shelf and the interactive process between the host and the viewing user can be automatically completed, achieving the technical effect of improving the automation of live broadcasting.

[0095] Embodiment 2 Embodiment 2 provides a live data processing method, and the execution subject is a live server. Since this embodiment corresponds to the steps of the live data processing method in Embodiment 1, but the execution subject is different, the description of some contents is relatively simple. For the relevant technical details and the effects achieved, please refer to the contents of the live data processing method provided in Embodiment 1.

[0096] Please see Figure 7 , Figure 7 This is a flow chart of a live data processing method provided in Example 2. The live data processing method comprises the following steps: S701: Determine a live broadcast service start-up strategy for determining the start-up time of a target live broadcast service, wherein the target live broadcast service is a digital human live broadcast service for a target merchant provided to a user terminal. The live broadcast service start-up strategy can determine the start-up time of the target live broadcast service. The live broadcast service start-up strategy is used for the generation of a subsequent live broadcast service start-up instruction.

[0097] S702: Determine the start time of the target live service based on the live service start policy. The live service end determines the start time of the target live service according to the determined live service start policy. For example, the live service end determines when to start streaming according to the live service start policy.

[0098] S703: When the activation timing of the target live broadcast service arrives, generate a live broadcast service activation instruction for instructing the cloud to activate the target live broadcast service for the user end.

[0099] S704: Send the live broadcast service start instruction to the cloud.

[0100] For example, when the broadcast start time arrives, the live broadcast service end generates a live broadcast service start instruction and sends the live broadcast service start instruction to the cloud end to instruct the cloud end to start the target live broadcast service for the user end. The determination of the live broadcast service start strategy for determining the start time of the target live broadcast service includes: obtaining the live broadcast service start strategy configured by the target merchant; or generating the live broadcast service start strategy. The live broadcast service start strategy includes the live broadcast service start time for starting the target live broadcast service for the user end.

[0101] In order to achieve a better automated live broadcast effect, the live broadcast data processing method of the second embodiment also provides a method for listing a target product, including: determining a target product listing strategy, the target product listing strategy is a strategy for indicating listing a target product in a target live broadcast room, the target live broadcast room is a live broadcast room corresponding to the target live broadcast service, and the target product is a target product that the target merchant can provide to users; based on the target product listing strategy, a target product listing instruction for indicating listing the target product in the target live broadcast room is generated; and the target product listing instruction is sent to the cloud. Among them, determining the target product listing strategy includes: obtaining the target product listing strategy configured by the target merchant; or generating the target product listing strategy. For example, the merchant pre-configures the target product listing strategy on the merchant side: when the anchor mentions the target product in the explanation, the product is put on the shelf and removed from the shelf within 10 minutes after the listing. Before the target product is mentioned, the target product is not on the shelf; or, the live broadcast service generates a target product listing strategy: the live broadcast service obtains historical live broadcast data, and obtains the time, copy, number of users, target product click rate and target product transaction rate according to the historical live broadcast data to estimate the estimated target product listing heat index. When the listing heat index reaches a certain threshold, the time point corresponding to the threshold is used as the actual listing time point of the target product. How to determine the target product listing strategy by generating the target product listing strategy will be described in detail in other parts of Example 2.

[0102] Optionally, the method further includes: determining live broadcast material data for the target live broadcast service as target live broadcast material data; and sending the target live broadcast material data to the cloud. Obtaining a material data request message sent by the cloud for requesting to obtain the target live broadcast material data; the sending of the target live broadcast material data to the cloud includes: sending the target live broadcast material data to the cloud in response to the material data request message.

[0103] When the live broadcast service receives a material data request message sent by the cloud to request the target live broadcast material data, the live broadcast material data of the target live broadcast service is determined as the target live broadcast material data, and the target live broadcast material data is returned to the cloud so that the cloud can preload the target live broadcast material data. The determination of the live broadcast material data for the target live broadcast service includes: obtaining the target live broadcast material data configured by the target merchant; or, generating the target live broadcast material data. The target live broadcast material data configured by the target merchant includes pictures, texts, display effects, etc. that are customized by the merchant according to its own store and products. Of course, the target live broadcast material data can also be generated by the target live broadcast service itself, which can be obtained through the default material library, or directly generated according to the product and store characteristics and categories uploaded by the merchant.

[0104] The user watching the live broadcast sends interactive information in the form of bullet screens, etc. The user who sends the interactive information is the target user. After receiving the interactive information sent by the target user, the anchor will respond to complete the live interaction. The live broadcast data processing method also includes: obtaining an anchor interactive information acquisition request message issued by the cloud for requesting to obtain anchor interactive information for the target user interactive information, the target user interactive information is the interactive information issued by the user end for the target live broadcast service; obtaining anchor interactive information for the target user interactive information; returning anchor interactive information for the target user interactive information to the cloud in response to the anchor interactive information acquisition request message. The interactive information includes real-time bullet screens, gifts, likes, etc. sent by the target user, and the bullet screen message is used as an example for explanation here. The user end sends a bullet screen message, and after the cloud receives the bullet screen message, it sends an anchor interactive information acquisition request message to the live broadcast service end, and the live broadcast service end obtains the anchor interactive information for the target user interactive information, and returns the anchor interactive information for the target user interactive information to the cloud end.

[0105] Optionally, the live broadcast data processing method further includes: determining a live broadcast service shutdown strategy for determining the target live broadcast service shutdown timing; determining the target live broadcast service shutdown timing based on the live broadcast service shutdown strategy; generating a live broadcast service shutdown instruction for instructing the cloud to shut down the target live broadcast service to the user end when the target live broadcast service shutdown timing arrives; and sending the live broadcast service shutdown instruction to the cloud. The determination of the live broadcast service shutdown strategy for determining the target live broadcast service shutdown timing includes: obtaining the live broadcast service shutdown strategy configured by the target merchant; or generating the live broadcast service shutdown strategy. The live broadcast service shutdown strategy includes the live broadcast service shutdown time for shutting down the target live broadcast service to the user end.

[0106] Optionally, the live broadcast data processing method also includes: obtaining a registration request message issued by the cloud for requesting to register the target device information of the cloud to the live broadcast server; determining the target merchant information of the target merchant to be bound to the target device information; and returning the target merchant information to the cloud in response to the registration request message to establish a binding relationship between the target merchant information and the target device information.

[0107] The above method determines the live broadcast service start-up strategy for determining the start-up time of the target live broadcast service, determines the start-up and end time of the target live broadcast service based on the live broadcast service start-up and close strategy, generates a live broadcast service start-up instruction or a close-down instruction when the start-up and close time of the target live broadcast service arrives, and sends the live broadcast service start-up instruction or the close-down instruction to the cloud to instruct the cloud to start or close the target live broadcast service to the user end, thereby achieving the technical effects of automated interaction and automatic listing of goods, while reducing labor costs and improving the automation of live broadcast. In order to further improve the automation of live broadcast, on the basis of the above, the live broadcast data processing method is optimized and improved.

[0108] In the above method, the determination of the live broadcast service activation strategy for determining the timing of activation of the target live broadcast service includes: obtaining the live broadcast service activation strategy configured by the target merchant; or, generating the live broadcast service activation strategy. Since the amount of historical access data can reflect the popularity of the target merchant to a certain extent, it can also predict the possible amount of access data in a certain time period in the future. Therefore, the live broadcast service end can generate a live broadcast service activation strategy based on the historical access data. The generation of the live broadcast service activation strategy includes the following steps: obtaining the historical access data of the target merchant. The historical visit data can reflect the scenarios and times in which users are more interested in the target merchant; based on the historical visit data of the target merchant, determine whether the historical visit data of the target merchant in each same historical time period in a preset number of historical visit data of the target merchant exceeds a preset visit data threshold, and whether the visit data difference between the historical visit data of the target merchant in every two same historical time periods is less than a preset visit data difference threshold; if so, obtain future time period data corresponding to the same historical time period; based on the future time period data, generate live broadcast service start time data for starting the target live broadcast service to the user end; and generate the live broadcast service start strategy based on the live broadcast service start time data. For example, the same historical time period is 20:00-21:00 Beijing time, if the target merchant's historical visit data from 20:00-21:00 every day for ten consecutive days exceeds the preset visit data threshold, and the visit data difference between the historical visit data from 20:00-21:00 on A month A day and 20:00-21:00 on X month X day is less than the preset visit data difference threshold, then it is judged as yes, and the future time period data corresponding to the same historical time period is obtained, that is, 20:00-21:00 on a future day is obtained as the future time period data.

[0109] By adopting the above-mentioned live broadcast service activation strategy generation method, it is possible to predict when the number of visits will be high in the future, and the historical number of visits data obtained has a certain stability, so it can be reasonably inferred that the live broadcast will be able to obtain a high number of visits in the future time period.

[0110] Optionally, generating the live service start time data for starting the target live service to the user terminal according to the future time period data includes: obtaining the start time point data of the future time period corresponding to the same historical time period according to the future time period data, and determining the start time point data as the live service start time data. For example, if 20:00-21:00 of a certain future day obtained above is used as the future time period data, wherein 20:00 is the start time point data of the future time period corresponding to the same historical time period, then 20:00 of a certain future day is used as the live service start time data.

[0111] The live broadcast service activation strategy generated by the live broadcast server can be determined based on the historical visit data of the target merchant, or according to steps S801-S804, or determined based on the historical visit data of the target merchant in steps S701-S705 and then determined according to steps S801-S804.

[0112] Optionally, generating the live broadcast service start-up strategy includes the following steps: after determining the live broadcast service start time for the target live broadcast service, generating a start time detection task for the live broadcast service start time of the target live broadcast service, and obtaining an identifier of the start time detection task as a start time detection task identifier; generating a start instruction generation task, and obtaining an identifier of the start instruction generation task as a start instruction generation task identifier, the start instruction generation task being used to indicate that when the live broadcast service start time is detected to have arrived according to the start time detection task, the live broadcast service start instruction is generated; generating a start instruction sending task, and obtaining an identifier of the start instruction sending task as a start instruction sending task identifier, the start instruction sending task being used to indicate that after the live broadcast service start instruction is generated according to the start instruction generation task, the live broadcast service start instruction is sent to the cloud; recording the start time detection task identifier, the start instruction generation task identifier and the start instruction sending task identifier in the task data item of the live broadcast service start policy template to generate the live broadcast service start strategy. That is to say, the generated live broadcast service startup strategy includes the startup time detection task, the startup instruction generation task, and the startup instruction sending task. Subsequent use requires the execution of the above tasks to complete the execution of the live broadcast service startup strategy.

[0113] In the above steps, when the live broadcast service end receives the live broadcast service start time, the live broadcast service end will record the live broadcast service start time, specifically, it can be achieved by generating a start time detection task for the live broadcast service start time. That is to say, after determining the live broadcast service start time for the target live broadcast service, the live broadcast service end needs to generate a start time detection task to detect and respond to the live broadcast service start time, and also needs to generate a start instruction generation task and a start instruction sending task to send the live broadcast service start instruction to the cloud. The live broadcast service start strategy generated according to the above method includes a start time detection task identifier, a start instruction generation task identifier and a start instruction sending task identifier, and these identifiers are recorded in the live broadcast service start strategy in the form of task data items. These identifiers can store the task content corresponding to the identifier in a relatively lightweight data form to reduce the occupation of resources. In fact, these identifiers represent the corresponding task content, such as the essence of the start time detection task identifier is the start time detection task. The task data item can also include other data related to the live broadcast service start strategy, such as merchant identification, name, data relationship, etc., which can be set according to merchant needs.

[0114] It is worth noting that the live broadcast service startup strategy can be used immediately after being generated, or it can be generated in advance and used when needed after a certain interval. When applying the live broadcast service startup strategy, optionally, determining the startup timing of the target live broadcast service based on the live broadcast service startup strategy includes: obtaining the startup time detection task identifier from the live broadcast service startup strategy; obtaining the startup time detection task according to the startup time detection task identifier; and determining the live broadcast service startup time according to the startup time detection task. Optionally, when the startup timing of the target live broadcast service arrives, generating a live broadcast service startup instruction for instructing the cloud to open the target live broadcast service to the user end includes: detecting whether the live broadcast service startup time has arrived according to the startup time detection task; if so, obtaining the startup instruction generation task identifier from the live broadcast service startup strategy; obtaining the startup instruction generation task according to the startup instruction generation task identifier; and generating the live broadcast service startup instruction according to the startup instruction generation task.

[0115] Correspondingly, when it is determined that the historical visit volume data of the target merchant does not satisfy the following two conditions: for each historical time period, among the historical visit volume data of the target merchant in a preset number of the same historical time period, the historical visit volume data of the target merchant in each same historical time period exceeds the preset visit volume data threshold, and the visit volume data difference between the historical visit volume data of each two same historical time periods of the target merchant is less than the preset visit volume data difference threshold, that is, when a stable and reliable regularity cannot be obtained from the historical visit volume data, a live broadcast service activation strategy can be generated through a visit volume detection task generated at a future detection time point, and the generation of the live broadcast service activation strategy includes the following steps: obtaining the historical visit volume data of the target merchant ; Determine whether the historical visit volume data of the target merchant meets the following conditions: for each historical time period, among the historical visit volume data of the target merchant in a preset number of the same historical time period, whether the historical visit volume data of the target merchant in each same historical time period exceeds the preset visit volume data threshold, and whether the difference in visit volume data between the historical visit volume data of the target merchant in every two same historical time periods is less than the preset visit volume data difference threshold; if not, generate a visit volume detection task, and obtain the identifier of the visit volume detection task as the visit volume detection task identifier, the visit volume detection task is used to indicate whether the visit volume data of the target merchant exceeds the preset visit volume data threshold at a future detection time point. The future detection time point is not necessarily related to the historical visit volume data in the previous text, and can be adjusted according to the needs of the merchant; generate an enable instruction generation task, and obtain the identifier of the enable instruction generation task as the enable instruction generation task identifier, the enable instruction generation task is used to indicate that when the visit volume data of the target merchant is detected to exceed the preset visit volume data threshold at the future detection time point according to the visit volume detection task, the live broadcast service enable instruction is generated; generate an enable instruction sending task, and obtain the identifier of the enable instruction sending task as the enable instruction sending task identifier, the enable instruction sending task is used to indicate that after the live broadcast service enable instruction is generated according to the enable instruction generation task, the live broadcast service enable instruction is sent to the cloud; record the visit volume detection task identifier, the enable instruction generation task identifier and the enable instruction sending task identifier in the task data item of the live broadcast service enable policy template to generate the live broadcast service enable policy.

[0116] The above steps provide a method for generating a live broadcast service activation strategy for situations where the regularity of historical visit volume data cannot be obtained. After the live broadcast service activation strategy is generated according to the above steps, the activation timing of the target live broadcast service can be determined based on the live broadcast service activation strategy. Optionally, determining the activation timing of the target live broadcast service based on the live broadcast service activation strategy includes: obtaining the visit volume detection task identifier from the live broadcast service activation strategy; obtaining the visit volume detection task according to the visit volume detection task identifier; detecting whether the visit volume data of the target merchant exceeds a preset visit volume data threshold at the future detection time point according to the visit volume detection task; if so, determining that the activation timing of the target live broadcast service has arrived. The future detection time points may be multiple or one.

[0117] When it is determined that no regularity can be obtained from the historical visit volume data of the target merchant, a live broadcast service activation strategy can also be generated by generating a visit volume detection task. Optionally, the generating of the live broadcast service activation strategy includes: Obtaining historical visit data of the target merchant; Determine whether the historical visit volume data of the target merchant meets the following conditions: for each historical time period, among a preset number of historical visit volume data of the target merchant in the same historical time period, the historical visit volume data of the target merchant in each same historical time period exceeds a preset visit volume data threshold, and the visit volume data difference between every two historical visit volume data of the target merchant in the same historical time period is less than a preset visit volume data difference threshold; If not, a visit volume detection task is generated, and an identifier of the visit volume detection task is obtained as the visit volume detection task identifier, wherein the visit volume detection task is used to indicate whether the visit volume data of the target merchant exceeds a preset visit volume data threshold at a future detection time point; Generate a start timing detection task, and obtain an identifier of the start timing detection task as the start timing detection task identifier, the start timing detection task is used to indicate whether the start timing of the target live broadcast service has arrived when the visit volume data of the target merchant detected at a future detection time point exceeds a preset visit volume data threshold, the start timing detection task includes an identifier of a start timing detection model, the start timing detection model is used to obtain a detection result of whether the start timing of the target live broadcast service has arrived based on the visit volume data of the target merchant detected at the future detection time point, the number of merchants of the same type that provide live broadcast services to the user terminal in the time period of the future detection time point, and the number of user terminals that visit merchants of the same type in the time period of the future detection time point in the visit volume data of the target merchant detected at the future detection time point, the merchants of the same type are merchants that provide the same type of services to the user terminal as the target merchant; Generate a start instruction generation task, and obtain an identifier of the start instruction generation task as the start instruction generation task identifier, wherein the start instruction generation task is used to indicate that when the start timing of the target live broadcast service is detected to have arrived according to the start timing detection task, the live broadcast service start instruction is generated; Generate a start instruction sending task, and obtain an identifier of the start instruction sending task as the start instruction sending task identifier, wherein the start instruction sending task is used to indicate that after the live broadcast service start instruction is generated according to the start instruction generating task, the live broadcast service start instruction is sent to the cloud; The visit volume detection task identifier, the start timing detection task identifier, the start instruction generation task identifier and the start instruction sending task identifier are recorded in the task data item of the live broadcast service start policy template to generate the live broadcast service start policy.

[0118] Among them, the historical time period can be customized. For example, with an interval of one hour, a day can be divided into 24 time periods. Any of 8:00-9:00, 13:00-14:00, and 20:00-21:00 can be used as a time period. The amount of historical access data in each time period is judged. If the judgment results of the historical access data for the 24 historical time periods cannot obtain the law of the historical access data, then the dependence on the historical access data is abandoned, and the detection results of the future detection time point are used as the basis for generating the live service start strategy. When the access data at the future detection time point exceeds the preset access data threshold, it can be further judged whether the user will stay in the live broadcast room of the target merchant. Therefore, the start timing detection task is generated, and the start timing detection model is used to detect whether the start timing has arrived, so as to further provide a basis for the generation of the start strategy. When there are merchants or products of the same channel and the same category, when different merchants start live broadcasts at the same time, it will affect the viewing volume of the target merchant. By inputting the visit volume data of the target merchant detected at the future detection time point, the number of merchants of the same type that provide live broadcast services to the user terminal in the time period of the future detection time point, and the number of user terminals that visit the same type of merchants in the time period of the future detection time point in the visit volume data of the target merchant detected at the future detection time point into the start timing detection model, the detection result of whether the start timing of the target live broadcast service has been arrived can be obtained, thereby reducing the data impact of live broadcasts of merchants of the same category and similar categories on the target merchant's live broadcast.

[0119] Optionally, the number of user terminals that visit the same type of merchant in the time period where the future detection time point is located is obtained in the following manner: input the time period data, user terminal data and merchant data of the same type in the time period where the future detection time point is located into the merchant access detection model, and obtain a merchant access detection result of whether the user terminal will visit the same type of merchant in the time period where the future detection time point is located; according to the merchant access detection result, count the user terminals that will visit the same type of merchant in the time period where the future detection time point is located, and obtain the number of user terminals that visit the same type of merchant in the time period where the future detection time point is located. The number of user terminals that visit the same type of merchant in the time period where the future detection time point is located can represent the number of users who visit merchants of the same type as the target merchant in the time period where the future detection time point is located. For example, if the user is interested in the live broadcast of a merchant of sports equipment, and the target merchant type is sports equipment, then the number of user terminals that visit other merchants of the sports equipment type in the time period where the future detection time is located is obtained. This data can reflect the probability, duration, and time point of users visiting live broadcast rooms of other merchants of the same type while they are interested in sports equipment merchants. Obtaining these parameters is helpful for the target merchant's live broadcast service activation strategy. For example, you can avoid the time when there are more live broadcasts of merchants of the same type, so that there are fewer live broadcasts of merchants of the same type at the start of the broadcast, which can reduce the direct impact on the data of the target merchant's live broadcast. Among them, there can also be multiple detection time points in the future. Selecting more regular and stable data from multiple different time points as a reference can improve the accuracy of the detection results.

[0120] The step S702 of determining the start timing of the target live broadcast service based on the live broadcast service start strategy includes: obtaining the visit volume detection task identifier from the live broadcast service start strategy; obtaining the visit volume detection task according to the visit volume detection task identifier; detecting whether the visit volume data of the target merchant exceeds a preset visit volume data threshold at the future detection time point according to the visit volume detection task; if it is detected that the visit volume data of the target merchant exceeds the preset visit volume data threshold at the future detection time point, obtaining the start timing detection task identifier from the live broadcast service start strategy; obtaining the start timing detection task according to the start timing detection task identifier; detecting whether the start timing of the target live broadcast service has arrived according to the start timing detection task; if so, determining that the start timing of the target live broadcast service has arrived. The visit volume detection task identifier, the start timing detection task identifier, the start instruction generation task identifier and the start instruction sending task identifier are recorded in the task data item of the live broadcast service start policy template, and the live broadcast service start policy of the target merchant is generated, wherein the visit volume detection task identifier, the start timing detection task identifier, the start instruction generation task identifier and the start instruction sending task identifier can respectively represent the visit volume detection task, the start timing detection task, the start instruction generation task and the start instruction sending task. According to the live broadcast service start policy, when determining the start timing, the tasks represented by the above-mentioned each identifier are executed accordingly to determine whether the start timing of the target live broadcast service has arrived. If it is confirmed that the start timing has arrived, the live broadcast service start instruction is sent to the cloud to control the user end to start the live broadcast service.

[0121] Considering that network quality may cause live broadcast network delays, resulting in asynchronous communication between the anchor and the target users, untimely response to interactive information, etc., and the live broadcast room will occupy cloud resources when streaming. Therefore, it is necessary to determine the relevant configuration of the live broadcast according to the network quality during the live broadcast.

[0122] In step S702, the generation of the live broadcast service activation strategy includes the following steps: obtaining the historical visit data of the target merchant; judging whether the historical visit data of the target merchant meets the following conditions: for each historical time period, among the historical visit data of the target merchant in a preset number of the same historical time periods, the historical visit data of the target merchant in each same historical time period exceeds the preset visit data threshold, and whether the visit data difference between the historical visit data of the target merchant in each two same historical time periods is less than the preset visit data difference threshold; If not, a link quality detection task is generated, and an identifier of the link quality detection task is obtained as the link quality detection task identifier, wherein the link quality detection task is used to indicate whether the quality of the data transmission link between the visiting user terminal of the target merchant and the cloud reaches a preset link quality threshold at a future detection time point; Generate an enable instruction generation task, and obtain an identifier of the enable instruction generation task as the enable instruction generation task identifier, wherein the enable instruction generation task is used to indicate that when the quality of the data transmission link between the visiting user terminal of the target merchant and the cloud reaches a preset link quality threshold at a future detection time point, the live broadcast service enable instruction is generated; Generate a start instruction sending task, and obtain an identifier of the start instruction sending task as the start instruction sending task identifier, wherein the start instruction sending task is used to indicate that after the live broadcast service start instruction is generated according to the start instruction generating task, the live broadcast service start instruction is sent to the cloud; The link quality detection task identifier, the start instruction generation task identifier and the start instruction sending task identifier are recorded in the task data item of the live broadcast service start policy template to generate the live broadcast service start policy.

[0123] Among them, determining the start time of the target live broadcast service based on the live broadcast service start-up strategy includes: obtaining the link quality detection task identifier from the live broadcast service start-up strategy; obtaining the link quality detection task according to the link quality detection task identifier; detecting whether the data transmission link quality between the visiting user terminal of the target merchant and the cloud reaches a preset link quality threshold at a future detection time point according to the link quality detection task; if so, determining that the start time of the target live broadcast service has arrived.

[0124] Link quality can reflect network quality and latency. As an important factor affecting the live broadcast experience, it determines the generation of live broadcast service activation strategy. For example, when a merchant is broadcasting live, if the merchant's network quality is poor, the normal live broadcast screen cannot be pushed to the user end, which is obviously not suitable for live broadcasting; when the link quality of a certain user end is poor, the live broadcast screen seen by the user end may be loaded abnormally and may also affect the use of other network services on the user end. If the link quality of other user ends is normal at this time, the live broadcast room screen can be pushed normally and displayed. At the same time, each user end will occupy cloud resources when watching live broadcasts, and the user end with poor link quality can neither watch live broadcasts nor occupy bandwidth resources, affecting the resource utilization of the cloud. Therefore, link quality detection for each user end is necessary. According to the link quality detection results of each user end, it is decided whether to push the stream to the user end. Displaying the live broadcast room is a way to improve the user end experience and reduce the waste of cloud resources.

[0125] Optionally, when the activation timing of the target live broadcast service arrives, generating a live broadcast service activation instruction for instructing the cloud to activate the target live broadcast service to the user end includes: after determining that the activation timing of the target live broadcast service arrives, obtaining the activation instruction generation task identifier from the live broadcast service activation strategy; obtaining the activation instruction generation task according to the activation instruction generation task identifier; and generating the live broadcast service activation instruction according to the activation instruction generation task. Sending the live broadcast service activation instruction to the cloud includes: obtaining the activation instruction sending task identifier from the live broadcast service activation strategy; obtaining the activation instruction sending task according to the activation instruction sending task identifier; and sending the live broadcast service activation instruction to the cloud according to the activation instruction sending task.

[0126] The live broadcast data processing method provided in the above embodiment 2 combines the historical visit data of the target merchant and predicts the visit data of the target merchant in the future, so as to determine the time of higher visit data as the start time of the live broadcast service and generate a live broadcast service start strategy; wherein according to the judgment conditions, a visit detection task, a start time detection task, and a link quality detection task are generated, and a method for generating a live broadcast service start strategy under different circumstances is provided. The target live broadcast service start timing determined by the live broadcast service start strategy generated by the above method is more accurate, more reasonable, and more intelligent. The live broadcast service start timing determined by this method can obtain better user visits, user viewing volume, and interactive effects, thereby improving the user's interactive experience, thereby promoting user orders, and improving the transaction conversion rate.

[0127] In the above method, in order to facilitate the generation of the target live broadcast service activation strategy for the target merchant, a live broadcast service activation strategy template is adopted, and the live broadcast service activation strategy template includes various task data items. Similarly, in order to facilitate the generation of the target product listing strategy on the live broadcast service side, a product listing strategy template is pre-set, and the template product listing strategy can be generated by recording the data that needs to be considered and may affect the target product listing strategy into the product listing strategy template. The pre-set product listing strategy template normalizes the data, which is helpful for standardized management and modification. The data items in the product listing strategy template can be added, deleted, and modified, and can be used to generate the product listing strategy for the target product.

[0128] According to different merchant needs and personalized settings, the target product listing strategy is also different. Optionally, the generating the target product listing strategy includes: Generate listing trigger data for the target product according to the target live broadcast material data, wherein the target live broadcast material data is live broadcast material data for the target live broadcast service, and the listing trigger data for the target product is data output in the target live broadcast room for triggering listing of the target product in the target live broadcast room; Generate a shelf trigger data detection task, and obtain an identifier of the shelf trigger data detection task as the shelf trigger data detection task identifier, wherein the shelf trigger data detection task is used to indicate whether the data output in the target live broadcast room contains shelf trigger data for the target product; Generate a target product listing instruction generation task, and obtain an identifier of the target product listing instruction generation task as the target product listing instruction generation task identifier, wherein the target product listing instruction generation task is used to indicate that when it is detected that the data output in the target live broadcast room contains listing trigger data for the target product, the target product listing instruction is generated; Generate a target product listing instruction sending task, and obtain an identifier of the target product listing instruction sending task as the target product listing instruction sending task identifier, wherein the target product listing instruction sending task is used to indicate that after the target product listing instruction is generated, the target product listing instruction is sent to the cloud; The shelf trigger data detection task identifier, the target product shelf instruction generation task identifier and the target product shelf instruction sending task identifier are recorded in the task data item of the product shelf strategy template to generate the target product shelf strategy.

[0129] Different live broadcast rooms, merchants, and products all have different characteristics, which can determine the click-through rate and transaction rate of users in the live broadcast room. Therefore, the unique feature data of the target product is obtained from the template live broadcast material data and recorded in the product listing strategy template to generate a more targeted template product listing strategy. Among them, the unique feature data can be the value of a unique feature. The generation of listing trigger data for the target product based on the target live broadcast material data includes: obtaining the unique feature data of the target product from the target live broadcast material data as the unique feature data of the target product, and the unique feature data of the target product is used to indicate the difference between the target product and other products provided by the target merchant; obtaining the association data between the target product and the other products from the target live broadcast material data as product association data; generating listing trigger data for the target product based on the unique feature data of the target product and the product association data.

[0130] The obtaining of the association data between the target product and the other products from the target live broadcast material data includes: obtaining the product usage data of the target product from the target live broadcast material data as the target product usage data, the target product usage data is the data marked as the usage information of the target product in the target live broadcast material data; obtaining the product usage data of the other products from the target live broadcast material data as the other product usage data, the other product usage data is the data marked as the usage information of the other products in the target live broadcast material data; inputting the target product usage data and the other product usage data into the usage matching detection model to obtain the usage matching detection result of whether the target product usage data matches the other product usage data; if the usage matching detection result indicates that the target product usage data matches the other product usage data, then recording the other product usage data as the product association data. When the data associated with it appears in the product data being displayed, the associated product can be put on the shelf by using the product association data as a trigger condition. For example, during the live broadcast in the live broadcast room, earphones and earphone chargers are used together. Therefore, when the earphones are introduced in the live broadcast, the listing link or product card of the earphone charger can be displayed in the live broadcast room. In the context of this scenario, users can understand the purpose and usage of the product more clearly and realistically, and obtain more comprehensive product information, thereby enhancing users' viewing experience and promoting orders.

[0131] The method of generating a target commodity listing instruction for instructing to list the target commodity in the target live broadcast room based on the target commodity listing strategy includes: obtaining the listing trigger data detection task identifier from the target commodity listing strategy; obtaining the listing trigger data detection task according to the listing trigger data detection task identifier; detecting whether the data output in the target live broadcast room contains the listing trigger data for the target commodity according to the listing trigger data detection task; if it is detected that the data output in the target live broadcast room contains the listing trigger data for the target commodity, obtaining the target commodity listing instruction generation task identifier from the target commodity listing strategy; obtaining the target commodity listing instruction generation task according to the target commodity listing instruction generation task identifier; and generating the target commodity listing instruction according to the target commodity listing instruction generation task. The sending of the target product listing instruction to the cloud comprises: after generating the target product listing instruction, obtaining the target product listing instruction sending task identifier from the target product listing strategy; obtaining the target product listing instruction sending task according to the target product listing instruction sending task identifier; and sending the target product listing instruction to the cloud according to the target product listing instruction sending task.

[0132] There may be multiple merchants of the same type on the electronic goods service platform, and it is possible that multiple merchants of the same type will provide live broadcast services, and may even broadcast live at the same time. The type may also be the type of service provided, such as physical goods delivered by express delivery, virtual goods that provide virtual services, goods that provide venues, etc. When users choose merchants, if it is an offline service, they often give priority to distance. When the distance factor is less different, in order to make the merchant more attractive, it is necessary to make the live broadcast room material more distinctive and different from other similar merchants, so that more users stay in the live broadcast room of the target merchant and increase the viewing volume and duration of the target live broadcast service. The generating of the target live broadcast material data comprises: obtaining the service mode data of the target merchant; if the service mode data of the target merchant indicates that the service mode provided by the target merchant is an in-store service mode, obtaining the geographic location data of the target merchant and obtaining the service item type data of the target merchant; judging whether there are other merchants in the merchant list whose geographic location matches that of the target merchant and whose product types match those of the target merchant as matching merchants according to the geographic location data of the target merchant and whose product types match those of the target merchant in the merchant list; if the matching merchant exists in the merchant list, obtaining the live broadcast material data of the matching merchant as matching merchant live broadcast material data; obtaining the merchant feature data of the target merchant and the product feature data of the target merchant; inputting the merchant feature data of the target merchant, the product feature data of the target merchant and the matching merchant live broadcast material data into a live broadcast material data generation model to generate the target live broadcast material data, wherein, when training the live broadcast material data generation model, the live broadcast material data sample of the matching merchant sample is used as a negative sample of the live broadcast material data. Detect similar merchants near the merchant, and screen the live data samples when training the live material data generation model. Abandon the live materials used by other merchants of the same type that are close to the merchant, and use other materials that have not been used and are very different in style from other merchants as the target live material data to create a live broadcast room, thereby generating a live broadcast room of the target merchant with a distinctive style, which is different from other nearby merchants of the same type, increases its attractiveness to users, and can also highlight the characteristics of the merchant and the product, so that users can clearly understand the product positioning at a glance.

[0133] In addition, when users interact in a live broadcast room, they often ask the same questions, but the way, style, and tone of each user's questions are different. The host usually uses the same answer to the same question asked in different ways, but this method usually makes users feel that it is less interesting or not taken seriously enough. Therefore, a method for generating host interaction information for the target user interaction information is provided to improve the interactive interest and user experience. The method of obtaining the host interaction information for the target user interaction information includes: obtaining the characteristic data of the target merchant as the target merchant characteristic data; obtaining the characteristic data of the target user who sends the target user interaction information as the target user characteristic data; obtaining the interactive scene characteristic data when the target user sends the target user interaction information; inputting the target user interaction information and the replied user interaction information into the user interaction information matching model to obtain a user interaction information matching result indicating whether the target user interaction information matches the replied user interaction information, wherein the replied user interaction information is the user interaction information that has replied to the user terminal; if the user interaction information matching result indicates that the target user interaction information matches the replied user interaction information, then the reply information is the user interaction information that has replied to the user terminal; if the user interaction information matching result indicates that the target user interaction information matches the replied user interaction information, then the reply information is the user interaction information that has replied to the user terminal; if the user interaction information matching result indicates that the target user interaction information matches the replied user interaction information, then the reply information is the user interaction information that has replied to the user terminal; if the user interaction information matching result indicates that the target user interaction information matches the replied user interaction information, then the reply information is the user interaction information that has replied to the user terminal; if the user interaction information matching result indicates that the target user interaction information matches the replied user interaction information, then the reply information is the user interaction information that has replied to the user terminal. If the target user characteristic data matches the interactive user characteristic data, the interactive user characteristic data is the characteristic data of the user who sent the replied user interactive information; if the target user characteristic data does not match the interactive user characteristic data, the anchor interactive information for the replied user interactive information is obtained as the comparative anchor interactive information; the target merchant characteristic data, the target user characteristic data, the interactive scene characteristic data, the interactive user characteristic data and the comparative anchor interactive information are input into the anchor interactive information generation model to generate the anchor interactive information for the target user interactive information, wherein when training the anchor interactive information generation model, the comparative anchor interactive information sample is used as the negative sample of the anchor interactive information. This step can be automatically replied by digital human or bullet screen. When the live broadcast service end receives the bullet screen message sent by the user, the interactive information of the target user will be analyzed, for example, the user's language habits, favorite language style, topics of interest, etc. are obtained, and the analysis results are used as the target user characteristic data. The interactive scene characteristic data can be determined in combination with the current live broadcast atmosphere, current time, etc. The target merchant feature data, target user feature data, interactive scene feature data, interactive user feature data, and comparative anchor interaction information involving merchant features, target user features, interactive scene feature data, user features that have had the same interaction, and anchor interaction information that has replied to user interaction information are input into the anchor interaction information generation model to generate anchor interaction information for the target user interaction information. The anchor interaction information generation model outputs anchor interaction information that is different from the comparative anchor interaction information sample to improve the personalization of the anchor interaction information.

[0134] The above is an explanation of the relevant process of starting live broadcast and live broadcast interaction, and the following is an explanation of the relevant process of closing live broadcast. The generation of the live broadcast service closing strategy includes: obtaining the historical visit volume data of the target merchant; judging, based on the historical visit volume data of the target merchant, whether the historical visit volume data of the target merchant in each historical time period in a preset number of historical time periods are lower than the preset visit volume data threshold, and whether the difference in visit volume data between the historical visit volume data of the target merchant in each two historical time periods is less than the preset visit volume data difference threshold; if so, obtaining the future time period data corresponding to the same historical time period; generating the live broadcast service closing time data for closing the target live broadcast service to the user end according to the future time period data; generating the live broadcast service closing strategy according to the live broadcast service closing time data. The generation of the live broadcast service closing time data for closing the target live broadcast service to the user end according to the future time period data includes: determining any time point data in the future time period data as the live broadcast service closing time data. Corresponding to the live broadcast service opening strategy, the live broadcast service closing strategy can also find a pattern based on the historical visit volume data. Usually, when the number of visits is low, the effect of live broadcast is poor. Therefore, the historical time period with stable and low visit volume is determined based on the historical visit volume data, and the corresponding future time period data is used as the live broadcast service shutdown time data to generate a live broadcast service shutdown strategy. The live broadcast service shutdown strategy generated in this way can ensure that the live broadcast is shut down within a certain range when the number of visits is low, reducing unnecessary live broadcast investment and resource occupation.

[0135] The generation of the live service shutdown strategy includes: after determining the live service shutdown time for the target live service, generating a shutdown time detection task for the live service shutdown time of the target live service, and obtaining the identifier of the shutdown time detection task as the shutdown time detection task identifier; generating a shutdown instruction generation task, and obtaining the identifier of the shutdown instruction generation task as the shutdown instruction generation task identifier, the shutdown instruction generation task is used to indicate that when the live service shutdown time is detected to have arrived according to the shutdown time detection task, the live service shutdown instruction is generated; generating a shutdown instruction sending task, and obtaining the identifier of the shutdown instruction sending task as the shutdown instruction sending task identifier, the shutdown instruction sending task is used to indicate that after the live service shutdown instruction is generated according to the shutdown instruction generation task, the live service shutdown instruction is sent to the cloud; recording the shutdown time detection task identifier, the shutdown instruction generation task identifier and the shutdown instruction sending task identifier in the task data item of the live service shutdown strategy template, and generating the live service shutdown strategy. In the above steps, when the live service end receives the live service shutdown time, the live service end can realize the recording of the live service shutdown time by generating a shutdown time detection task for the value service shutdown time. That is to say, after determining the live service shutdown time for the live service, the live service end needs to detect and respond to the live service shutdown time, and also generate a shutdown instruction generation task and a shutdown instruction sending task to send the live service shutdown instruction to the cloud. The live service shutdown strategy generated according to the above method includes a shutdown time detection task representation, a shutdown instruction generation task identifier, and a shutdown instruction sending task identifier, and these identifiers are recorded in the live service shutdown strategy in the form of task data items. The task data item may also include other data related to shutting down the live service.

[0136] Determining the shutdown timing of the target live service based on the live service shutdown strategy includes: obtaining the shutdown time detection task identifier from the live service shutdown strategy; obtaining the shutdown time detection task according to the shutdown time detection task identifier; and determining the live service shutdown time according to the shutdown time detection task.

[0137] When the shutdown timing of the target live broadcast service is reached, a live broadcast service shutdown instruction is generated to instruct the cloud to shut down the target live broadcast service to the user end, including: detecting whether the live broadcast service shutdown time is reached according to the shutdown time detection task; if so, obtaining the shutdown instruction generation task identifier from the live broadcast service shutdown strategy; obtaining the shutdown instruction generation task according to the shutdown instruction generation task identifier; and generating the live broadcast service shutdown instruction according to the shutdown instruction generation task.

[0138] The generation of the live broadcast service closing strategy includes: obtaining the historical visit volume data of the target merchant; judging whether the historical visit volume data of the target merchant meets the following conditions: for each historical time period, among the historical visit volume data of the target merchant in a preset number of the same historical time periods, the historical visit volume data of the target merchant in each same historical time period is lower than the preset visit volume data threshold, and the visit volume data difference between the historical visit volume data of the target merchant in every two same historical time periods is lower than the preset visit volume data difference threshold; if not, generating a visit volume detection task, and obtaining the identifier of the visit volume detection task as the visit volume detection task identifier, the visit volume detection task is used to indicate whether the visit volume data of the target merchant is lower than the preset visit volume at a future detection time point after the target live broadcast service is opened to the user end. volume data threshold; generate a closing instruction generation task, and obtain the identifier of the closing instruction generation task as the closing instruction generation task identifier, the closing instruction generation task is used to indicate that when the target merchant's visit volume data is detected to be lower than the preset visit volume data threshold at the future detection time point according to the visit volume detection task, the live broadcast service closing instruction is generated; generate a closing instruction sending task, and obtain the identifier of the closing instruction sending task as the closing instruction sending task identifier, the closing instruction sending task is used to indicate that after the live broadcast service closing instruction is generated according to the closing instruction generation task, the live broadcast service closing instruction is sent to the cloud; record the visit volume detection task identifier, the closing instruction generation task identifier and the closing instruction sending task identifier in the task data item of the live broadcast service closing policy template to generate the live broadcast service closing policy. The above steps provide a method for generating a live broadcast service closing policy for the situation where the regularity of historical visit volume data cannot be obtained. After the live broadcast service closing policy is generated according to the above steps, the closing timing of the target live broadcast service can be determined based on the live broadcast service closing policy. Optionally, determining the closing time of the target live broadcast service based on the live broadcast service closing strategy includes: obtaining the visit volume detection task identifier from the live broadcast service closing strategy; obtaining the visit volume detection task according to the visit volume detection task identifier; detecting whether the visit volume data of the target merchant is lower than a preset visit volume data threshold at the future detection time point according to the visit volume detection task; if so, determining that the closing time of the target live broadcast service has arrived. The future detection time points may be multiple or one.

[0139] Considering that network quality may cause live broadcast network delays, resulting in asynchronous communication between the anchor and the target user, untimely response to interactive information, etc., and the live broadcast service will occupy cloud resources, it is necessary to determine the live broadcast shutdown strategy based on the link quality. Optionally, the generation of the live broadcast service shutdown strategy includes: obtaining the historical visit data of the target merchant; judging whether the historical visit data of the target merchant meets the following conditions: for each historical time period, the historical visit data of the target merchant in a preset number of the same historical time periods, the historical visit data of the target merchant in each of the same historical time periods is lower than the preset visit data threshold, and the difference in visit data between the historical visit data of the target merchant in every two of the same historical time periods is less than the preset visit data difference threshold; if not, a link quality detection task is generated, and the identifier of the link quality detection task is obtained as the link quality detection task identifier. The link quality detection task is used to indicate the detection of the target merchant's data transmission link between the visiting user terminal and the cloud at a future detection time point after the target live broadcast service is opened to the user terminal. Whether the quality reaches a preset link quality threshold; generate a close instruction generation task, and obtain the identifier of the close instruction generation task as the close instruction generation task identifier, the close instruction generation task is used to indicate that when it is detected at a future detection time point that the data transmission link quality between the visiting user terminal of the target merchant and the cloud is lower than the preset link quality threshold, the live broadcast service close instruction is generated; generate a close instruction sending task, and obtain the identifier of the close instruction sending task as the close instruction sending task identifier, the close instruction sending task is used to indicate that after the live broadcast service close instruction is generated according to the close instruction generation task, the live broadcast service close instruction is sent to the cloud; the link quality detection task identifier, the close instruction generation task identifier and the close instruction sending task identifier are recorded in the task data item of the live broadcast service close policy template to generate the live broadcast service close policy. Among them, determining the shutdown timing of the target live broadcast service based on the live broadcast service shutdown strategy includes: obtaining the link quality detection task identifier from the live broadcast service shutdown strategy; obtaining the link quality detection task according to the link quality detection task identifier; detecting whether the data transmission link quality between the visiting user terminal of the target merchant and the cloud reaches a preset link quality threshold at a future detection time point according to the link quality detection task; if so, determining that the shutdown timing of the target live broadcast service has arrived.

[0140] Link quality can reflect network quality and latency. As an important factor affecting the live broadcast experience, it determines the generation of live broadcast service activation strategy. For example, when the link quality of a certain user end is poor, the live broadcast screen seen by the user end may load abnormally, and it may also affect the use of other network services on the user end. At the same time, each user end watching the live broadcast will occupy bandwidth resources. Therefore, link quality detection for each user end is necessary. Deciding whether to end the push stream of each user end based on the link quality detection results of each user end is a way to improve the user end experience and reduce the waste of cloud resources.

[0141] When the closing time of the target live broadcast service arrives, generating a live broadcast service closing instruction for instructing the cloud to close the target live broadcast service to the user end includes: after determining that the closing time of the target live broadcast service has arrived, obtaining the closing instruction generation task identifier from the live broadcast service closing strategy; obtaining the closing instruction generation task according to the closing instruction generation task identifier; and generating the live broadcast service closing instruction according to the closing instruction generation task. Sending the live broadcast service closing instruction to the cloud includes: obtaining the closing instruction sending task identifier from the live broadcast service closing strategy; obtaining the closing instruction sending task according to the closing instruction sending task identifier; and sending the live broadcast service closing instruction to the cloud according to the closing instruction sending task.

[0142] The live broadcast data processing method provided in the above embodiment combines the historical visit volume data of the target merchant and the visit volume data of the target merchant in the predicted future time, so as to determine the time of lower visit volume data as the timing to shut down the live broadcast service, and generate a live broadcast service shutdown strategy; wherein the visit volume detection task, the shutdown time detection task, and the link quality detection task are generated according to the judgment conditions, and a method for generating a live broadcast service shutdown strategy in different scenarios is provided. The target live broadcast service shutdown time determined by the live broadcast service shutdown strategy generated by the above method is more accurate and flexible. The live broadcast service startup timing determined by this method can obtain a better user experience and promote the rational use of cloud resources.

[0143] Embodiment 2 starts from the aspects of target live broadcast service activation strategy, target product listing strategy, anchor interaction information of target user interaction information, target live broadcast service shutdown strategy, etc., and explains the live broadcast data processing in detail. The live broadcast data processing method provided in Embodiment 2 can accurately and intelligently determine the time to start and stop the broadcast, the target product listing strategy is more merchant-specific, and the anchor interaction information is more personalized, achieving the technical effect of determining the start and stop time, target product listing, and anchor interaction information without manual operation, greatly improving the degree of live broadcast automation; at the same time, it improves the user experience and reduces the waste of cloud resources.

[0144] Embodiment 3 provides a live broadcast data processing device, which corresponds to the method in the aforementioned embodiment 1. Therefore, the following description of the image processing system embodiment is merely illustrative, and the relevant implementation details can be understood by referring to embodiment 1.

[0145] Please refer to Figure 8 , Figure 8 A schematic diagram of a live data processing device provided in Example 3. The live data processing device includes: a live service start instruction obtaining unit 801, used to obtain a live service start instruction for instructing to start a target live service to a user end, the target live service is a digital human live service for a target merchant, the live service start instruction is generated based on a live service start strategy, the live service start strategy is a strategy for determining the start time of the target live service, the live service start strategy is pre-configured by the target merchant or generated by a live service end that provides a digital human live service; a live data stream generating unit 802, used to generate a digital human live data stream for the target merchant as a target live data stream in response to the live service start instruction; a live data stream pushing unit 803, used to push the target live data stream to the user end.

[0146] By using the live broadcast data processing device provided in Example 3, it is possible to realize real-time operations such as creating a live broadcast room and sending a broadcast instruction when the live broadcast starts, and the broadcast instruction can be automatically generated, the broadcast task can be completed, and the live broadcast can be ended. At the same time, the target products in the live broadcast room can be automatically put on the shelves and the interaction process between the anchor and the viewing users can be automatically completed, achieving the technical effect of improving the automation of live broadcast.

[0147] Embodiment 4 provides a live broadcast data processing device, which corresponds to the method in the aforementioned embodiment 2. Therefore, the following description of the image processing system embodiment is merely illustrative, and the relevant implementation details can be understood by referring to embodiment 2.

[0148] Please refer to Fig. 9 , Fig. 9Schematic diagram of a live data processing device provided in Example 4. The live data processing device includes: a live service start-up strategy determination unit 901, used to determine the live service start-up strategy, the live service start-up strategy is used to determine the start-up timing of the target live service, the target live service is a digital human live service for target merchants provided to the user end; a start-up timing determination unit 902, used to determine the start-up timing of the target live service based on the live service start-up strategy; a live service start-up instruction generation unit 903, when the start-up timing of the target live service arrives, generates a live service start-up instruction for instructing the cloud to start the target live service to the user end; a live service start-up instruction sending unit 904, sending the live service start-up instruction to the cloud.

[0149] The live broadcast data processing device provided by the fourth embodiment can achieve the technical effect of determining the start and end broadcast time, the listing of target products, and the interactive information of the anchor without manual operation, which greatly improves the degree of automation of the live broadcast; at the same time, it improves the user experience and reduces the waste of cloud resources.

[0150] Embodiment 5 provides a live data processing system, which includes the live data processing method in Embodiment 1 and the execution body of the live data processing method in Embodiment 2. The following description of the embodiment of the live data processing system is only illustrative, and the relevant implementation details can be understood by referring to Embodiment 1 and Embodiment 2. Fig.10 , Fig.10 The live broadcast data processing system provided in Example 5 includes a cloud 103, a live broadcast server 101, a host terminal 104 and a user terminal 102: The cloud 103 is used to obtain a live broadcast service start instruction for instructing to start a target live broadcast service for a user terminal, wherein the target live broadcast service is a digital human live broadcast service for a target merchant, and the live broadcast service start instruction is generated based on a live broadcast service start strategy, wherein the live broadcast service start strategy is a strategy for determining the timing of starting the target live broadcast service, and the live broadcast service start strategy is pre-configured by the target merchant or generated by a live broadcast service terminal that provides the digital human live broadcast service; in response to the live broadcast service start instruction, a digital human live broadcast data stream for the target merchant is generated as a target live broadcast data stream; and the target live broadcast data stream is pushed to the user terminal.

[0151] The live broadcast service end 101 is used to determine the live broadcast service startup strategy, and the live broadcast service startup strategy is used to determine the startup timing of the target live broadcast service, and the target live broadcast service is a digital human live broadcast service for target merchants provided to the user end; based on the live broadcast service startup strategy, the startup timing of the target live broadcast service is determined; when the startup timing of the target live broadcast service arrives, a live broadcast service startup instruction is generated to instruct the cloud to start the target live broadcast service to the user end; and the live broadcast service startup instruction is sent to the cloud.

[0152] The host terminal 104 is used to pre-generate digital human materials, and to configure products, interactions, digital human products, and the merchant's start time, start time, and end time in advance. The user terminal 102 is used for viewing users to watch live broadcasts and interact, and receive bullet screen messages from the viewing users. The user terminal supports bullet screen interaction, product card pop-up, digital human identification and other capabilities, and can also support real-time bullet screen interaction between users and digital humans. The merchant terminal 1005 is used for merchants to configure live broadcast information and synchronize the configured information to the host terminal. The merchant terminal displays the interface for merchants to configure live broadcast information in advance. The configuration interface can be provided by the host terminal.

[0153] The cloud supports scheduling services, including machine registration, user material pulling, and barrage interaction. Multiple machines can be deployed on the cloud, and the machines can include operation management systems, training environments, cloud computing services, virtual machines, etc. The cloud also includes a cloud streaming cluster, which supports multi-opening capabilities, as well as image creation and cloning. It also pre-deploys basic dependencies such as computing tools and network security in the image, and supports scaling cloud resources based on usage. This cloud deployment eliminates the need for merchants to prepare local equipment dedicated to live broadcasts, effectively saving merchants' machine costs. In addition, based on the support of intelligent operation and maintenance and the technology center, high-definition, low-latency cloud direct push can be achieved, effectively reducing live broadcast and interactive delays and improving viewing user experience. Through the cloud direct push capability, merchants do not need to prepare local equipment, effectively saving merchants' machine preparation and operation costs.

[0154] The live broadcast data processing system provided in Example 5 can automatically generate broadcast instructions and complete the broadcast task without the merchant having to perform real-time operations such as creating a live broadcast room and sending broadcast instructions when starting the broadcast.

[0155] Embodiment 6 provides an electronic device embodiment for live data processing provided in the first embodiment. The electronic device is a user terminal. The live data processing methods of embodiments 1 and 2 can be executed on the user terminal. The user terminal can also include the live data processing system in embodiment 5. The following description of the electronic device embodiment is merely illustrative. For related content, please refer to the aforementioned embodiment.

[0156] Please see Fig.11 Understand the above electronic devices, Fig.11 Schematic diagram of an electronic device provided in Example 6. The electronic device provided in Example 6 includes: a processor 1101, a memory 1102, a communication bus 1103, and a communication interface 1104; The memory 1102 is used to store computer instructions for data processing. When the computer instructions are read and executed by the processor 1101, the steps of the first embodiment and the second embodiment are executed.

[0157] Embodiment 7 provides a computer-readable storage medium for implementing the methods described in Embodiment 1 and Embodiment 2. The computer-readable storage medium embodiment is described in a relatively simple manner. For the relevant parts, please refer to the corresponding description of the above method embodiment. The embodiments described below are only exemplary. The computer-readable storage medium stores computer-executable instructions, which are used to implement any of the methods described in Embodiment 1 and Embodiment 2 when executed by a processor.

[0158] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0159] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0160] 1. Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include non-transitory media such as modulated data signals and carrier waves.

[0161] 2. Those skilled in the art should understand that the embodiments of the present application can be provided as methods, systems or computer program products. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0162] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0163] Although the present application is disclosed as above in the form of a preferred embodiment, it is not intended to limit the present application. Any technical personnel in this field may make possible changes and modifications without departing from the spirit and scope of the present application. Therefore, the scope of protection of the present application shall be based on the scope defined by the claims of the present application.

Claims

1. A live broadcast data processing method, characterized in that: include: Determining a live broadcast service activation strategy for determining a target live broadcast service activation timing, wherein the target live broadcast service is a digital human live broadcast service for a target merchant provided to a user terminal; Based on the live broadcast service activation strategy, determining the activation timing of the target live broadcast service; When the activation timing of the target live broadcast service arrives, generating a live broadcast service activation instruction for instructing the cloud to activate the target live broadcast service for the user end; Sending the live broadcast service start instruction to the cloud; Determine a target product listing strategy, where the target product listing strategy is a strategy for listing the target product in the target live broadcast room; Generate a target product listing instruction for instructing to list the target product in the target live broadcast room based on the target product listing strategy; Sending the target product listing instruction to the cloud; Wherein, the determining of the target commodity listing strategy includes: obtaining the target commodity listing strategy configured by the target merchant, or generating the target commodity listing strategy; The generating of the target product listing strategy includes: generating listing trigger data for the target product based on target live broadcast material data, and generating the target product listing strategy based on the listing trigger data for the target product, wherein the listing trigger data for the target product is data output in the target live broadcast room for triggering the listing of the target product in the target live broadcast room.

2. The live broadcast data processing method according to claim 1, characterized in that: Also includes: Determining live broadcast material data for the target live broadcast service as target live broadcast material data; The target live broadcast material data is sent to the cloud.

3. The live broadcast data processing method according to claim 2, characterized in that: The determining of the live broadcast material data for the target live broadcast service includes: Obtain the target live broadcast material data configured by the target merchant; Alternatively, the target live broadcast material data is generated.

4. The live broadcast data processing method according to claim 2, characterized in that: Also includes: Obtaining a material data request message sent by the cloud for requesting to obtain the target live broadcast material data; The sending the target live broadcast material data to the cloud comprises: sending the target live broadcast material data to the cloud in response to the material data request message.

5. The live broadcast data processing method according to claim 1, characterized in that: Also includes: Obtaining a host interaction information acquisition request message sent by the cloud for requesting to obtain host interaction information for target user interaction information, wherein the target user interaction information is interaction information sent by a user end for the target live broadcast service; Obtaining host interaction information for the target user interaction information; In response to the anchor interaction information acquisition request message, anchor interaction information for the target user interaction information is returned to the cloud.

6. The live broadcast data processing method according to claim 1, characterized in that: The generating the target commodity listing strategy according to the listing trigger data for the target commodity comprises: Generate a shelf trigger data detection task and obtain the shelf trigger data detection task identifier, wherein the shelf trigger data detection task is used to indicate whether the data output in the target live broadcast room contains shelf trigger data for the target product; Generate a target product listing instruction generation task and obtain a target product listing instruction generation task identifier; Generate a task for sending a target product listing instruction and obtain a task identifier for sending a target product listing instruction; The shelf trigger data detection task identifier, the target product shelf instruction generation task identifier and the target product shelf instruction sending task identifier are recorded in the task data item of the product shelf strategy template to generate the target product shelf strategy.

7. A live broadcast data processing method, characterized in that: include: Obtaining a live broadcast service start instruction for instructing to start a target live broadcast service for a user terminal, wherein the target live broadcast service is a digital human live broadcast service for a target merchant, and the live broadcast service start instruction is generated based on a live broadcast service start strategy, wherein the live broadcast service start strategy is a strategy for determining a start time of the target live broadcast service, and the live broadcast service start strategy is pre-configured by the target merchant or generated by a live broadcast service terminal providing the digital human live broadcast service; In response to the live broadcast service start instruction, generating a digital human live broadcast data stream for the target merchant as a target live broadcast data stream; Pushing the target live data stream to the user end; Obtaining a target commodity listing instruction for instructing to list a target commodity in a target live broadcast room corresponding to the target live broadcast data stream, wherein the target commodity is a target commodity that the target merchant can provide to a user, and the target commodity listing instruction is generated based on a target commodity merchant policy, and the target commodity listing policy is a policy for indicating listing the target commodity in the target live broadcast room, and the target commodity listing policy is pre-configured by the target merchant or generated by the live broadcast server; Obtaining live broadcast data of the target product in the target live broadcast room as the target product live broadcast data; The target commodity live broadcast data is added to the target live broadcast data stream.

8. The live broadcast data processing method according to claim 7, characterized in that: Also includes: Obtaining live broadcast material data for the target live broadcast service provided by the live broadcast service end as target live broadcast material data, wherein the target live broadcast material data is preset by the target merchant or generated by the live broadcast service end; The digital human live broadcast data stream for the target merchant includes: generating a digital human live broadcast data stream for the target merchant based on the target live broadcast material data.

9. The live broadcast data processing method according to claim 8, characterized in that: Also includes: Sending a material data request message for requesting to obtain the target live broadcast material data to the live broadcast server; The obtaining of the live broadcast material data of the digital human live broadcast service of the target merchant provided by the live broadcast service end includes: obtaining the target live broadcast material data provided by the live broadcast service end in response to the material data request message.

10. The live broadcast data processing method according to claim 7, characterized in that: Also includes: Obtaining interactive information sent by the user terminal for the target live broadcast service as target user interactive information; Generate response information for the target user interaction information as target anchor interaction information according to the interaction strategy for the target live broadcast service, wherein the interaction strategy is pre-configured by the target merchant or generated by the live broadcast service end; The target anchor interaction information is sent to the user terminal.

11. The live broadcast data processing method according to claim 8, characterized in that: The step of generating response information for the target user interaction information according to the interaction strategy for the target live broadcast service includes: Determine whether there is interaction information matching the target user interaction information in the preset user interaction information, and use the interaction information matching the target user interaction information in the preset user interaction information as the matching interaction information; If yes, then obtaining the host interaction information corresponding to the matching interaction information as the matching host interaction information; According to the matching anchor interaction information, response information for the target user interaction information is generated.

12. The live broadcast data processing method according to claim 8, characterized in that: The step of generating response information for the target user interaction information according to the interaction strategy for the target live broadcast service includes: Sending an anchor interaction information acquisition request message to the live broadcast server for requesting to obtain anchor interaction information for the target user interaction information; Obtaining the anchor interaction information for the target user interaction information provided by the live broadcast server in response to the anchor interaction information acquisition request message; Generate response information for the target user interaction information based on the host interaction information for the target user interaction information.

13. A live broadcast data processing device, characterized in that: include: A live broadcast service activation strategy determination unit, used to determine a live broadcast service activation strategy, wherein the live broadcast service activation strategy is used to determine a target live broadcast service activation timing, wherein the target live broadcast service is a digital human live broadcast service provided to a user terminal for a target merchant; A start timing determination unit, configured to determine a start timing of the target live broadcast service based on the live broadcast service start strategy; A live broadcast service start instruction generating unit, when the start time of the target live broadcast service arrives, generates a live broadcast service start instruction for instructing the cloud to start the target live broadcast service to the user end; The live broadcast service start instruction sending unit sends the live broadcast service start instruction to the cloud.

14. A live broadcast data processing device, characterized in that: include: A live broadcast service start instruction obtaining unit, used for obtaining a live broadcast service start instruction for instructing to start a target live broadcast service to a user terminal, wherein the target live broadcast service is a digital human live broadcast service for a target merchant, and the live broadcast service start instruction is generated based on a live broadcast service start strategy, wherein the live broadcast service start strategy is a strategy for determining a start time of the target live broadcast service, and the live broadcast service start strategy is pre-configured by the target merchant or generated by a live broadcast service terminal providing a digital human live broadcast service; A live data stream generating unit, configured to generate a digital human live data stream for the target merchant as a target live data stream in response to the live service start instruction; The live data stream pushing unit is used to push the target live data stream to the user end.

15. A live broadcast data processing system, comprising a cloud end and a live broadcast server, characterized in that: include: The cloud end is used to obtain a live broadcast service start instruction, the live broadcast service start instruction is used to instruct the user end to start a target live broadcast service, the target live broadcast service is a digital human live broadcast service for a target merchant, the live broadcast service start instruction is generated based on a live broadcast service start strategy, the live broadcast service start strategy is a strategy for determining the start time of the target live broadcast service, the live broadcast service start strategy is pre-configured by the target merchant or generated by a live broadcast service end that provides a digital human live broadcast service; In response to the live broadcast service start instruction, generating a digital human live broadcast data stream for the target merchant as a target live broadcast data stream; Pushing the target live data stream to the user end; The live broadcast service end is used to determine a live broadcast service start-up strategy, and the live broadcast service start-up strategy is used to determine the start-up timing of a target live broadcast service, and the target live broadcast service is a digital human live broadcast service for a target merchant provided to the user end; Based on the live broadcast service activation strategy, determining the activation timing of the target live broadcast service; When the activation timing of the target live broadcast service arrives, a live broadcast service activation instruction is generated to instruct the cloud to activate the target live broadcast service to the user end; and the live broadcast service activation instruction is sent to the cloud.

16. An electronic device, characterized in that: include: A processor, a memory, and computer program instructions stored on the memory and executable on the processor; When the processor executes the computer program instructions, the method according to any one of claims 1 to 12 is implemented.

17. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by a processor, they are used to implement the method according to any one of claims 1 to 12.

Images