Media content production and distribution integrated system based on multi-agent collaboration

Abstract

The invention discloses a media content production and distribution integrated system based on multi-agent collaboration, and relates to the technical field of media content production and distribution. The system comprises a content creation multi-task intelligent system, an audience analysis multi-task intelligent system and a media content distribution module, and the content creation multi-task intelligent system comprises a topic selection planning agent, a text creation agent, an article illustration agent and a content evaluation agent. The audience analysis multi-task intelligent system comprises a data processing agent, a user portrait agent and an interest prediction agent, and through functions and cooperation of the agents, an AI large language model and a multi-agent cooperation technology can be introduced, so that a full-service process from media content production to media content distribution can be automatically opened, and the efficiency is improved. And automatic production and personalized distribution of the media content are realized, the production efficiency and the propagation effect of the media content are greatly improved, and practical application and popularization are facilitated.

Description

Technical Field

[0001] The present invention belongs to the technical field of media content production and distribution, and specifically relates to an integrated system for media content production and distribution based on multi-agent collaboration. Background Art

[0002] With the rapid development of the Internet and social media, the media industry is undergoing profound changes. The traditional media content production model relies on manual labor, which is not only inefficient and difficult to meet the growing demand for personalized content, but also appears to be stretched in the face of massive content and rapidly iterating market environments. In addition, existing media content distribution solutions also face the challenge of how to accurately reach the target audience and improve content interactivity and user engagement. In this context, the media industry is in urgent need of an efficient and intelligent solution for media content production and distribution, and breakthroughs in related technologies such as agents with AI (Artificial Intelligence) large language models as the core have brought opportunities for this. However, the application of technologies such as large language models in the industry is still limited to discrete small scenes and functional points, and has not been well integrated and utilized in the field of content production and distribution. There are still problems such as cumbersome content production processes and limited innovation, lack of accuracy and personalization in distribution strategies, low user satisfaction and participation, and greatly reduced content dissemination effects. Therefore, how to provide a new solution for media content production and distribution based on multi-agent collaboration, so as to automatically open up the entire business process from media content production to distribution by introducing AI large language models and multi-agent collaboration technologies, and realize the automated production and personalized distribution of media content, is a topic that technicians in this field urgently need to study. Summary of the invention

[0003] The purpose of the present invention is to provide an integrated media content production and distribution system based on multi-agent collaboration, so as to solve the problems existing in existing media content production and distribution solutions, such as cumbersome content production process and limited innovation, lack of accuracy and personalization of distribution strategy, and greatly reduced content dissemination effect.

[0004] In order to achieve the above object, the present invention adopts the following technical solutions: The present invention provides an integrated system for media content production and distribution based on multi-agent collaboration, including a content creation multi-task intelligent system, an audience analysis multi-task intelligent system and a media content distribution module, wherein the content creation multi-task intelligent system includes a topic planning intelligent agent, a text creation intelligent agent, an article illustration intelligent agent and a content evaluation intelligent agent, and the audience analysis multi-task intelligent system includes a data processing intelligent agent, a user portrait intelligent agent and an interest prediction intelligent agent; The topic planning agent is used to automatically generate or recommend topics that meet market trends based on dynamic information crawling results through big data analysis technology and user demand prediction technology; The text creation agent is used to generate text content under the topic according to the topic selected by the topic planning agent, in combination with the outline generation module, the knowledge reference module and the text generation module; The article picture matching agent is used to retrieve or generate pictures matching the text content according to the text content from the text creation agent; The content evaluation agent is used to perform quality inspection and compliance review on media content to ensure the legality and accuracy of the media content, wherein the media content includes text content from the text creation agent and pictures from the article illustration agent; The data processing agent is used to collect user data from multiple channels and pre-process the user data to obtain pre-processed user data, wherein the multiple channels include social media, search engines and / or user behavior logs, and the pre-processing includes data cleaning processing, data standardization processing and / or data desensitization processing; The user portrait agent is used to generate a multi-dimensional user portrait based on the pre-processed user data from the data processing agent, wherein the multi-dimensionality includes a physical attribute dimension, an interest preference dimension and / or a social relationship dimension; The interest prediction agent is used to group the user portraits from the user portrait agent based on a cluster analysis algorithm, identify user groups with similar interests, and predict the changing trends of user interests of different user groups based on a time series prediction model to obtain user interest prediction results for each user group; The media content distribution module is used to associate and match the user interest prediction results from the interest prediction agent and each user group with the media content from the content evaluation agent that has passed quality inspection and compliance review, determine personalized matching media content for each user group and distribute it.

[0005] Based on the above invention content, a new solution for media content production and distribution based on multi-agent collaboration is provided, which includes a content creation multi-task intelligent system, an audience analysis multi-task intelligent system and a media content distribution module, wherein the content creation multi-task intelligent system includes a topic planning intelligent agent, a text creation intelligent agent, an article illustration intelligent agent and a content evaluation intelligent agent, and the audience analysis multi-task intelligent system includes a data processing intelligent agent, a user portrait intelligent agent and an interest prediction intelligent agent, and through their functions and collaboration, the entire business process from media content production to distribution can be automatically opened up by introducing AI large language models and multi-agent collaboration technology, and the automated production and personalized distribution of media content can be realized, which greatly improves the production efficiency and dissemination effect of media content and facilitates practical application and promotion.

[0006] In a possible design, the content creation multi-task intelligent system also includes a video creation agent; The video creation agent is used to create and encapsulate video content under the topic according to the topic selected by the topic planning agent through script writing, material integration, sound synthesis and video synthesis. The media content also includes video content from the video creation agent.

[0007] In a possible design, there is also a multi-terminal distribution multi-task intelligent system, wherein the multi-terminal distribution multi-task intelligent system includes a channel content adaptation agent and a channel content distribution agent; The channel content adaptation agent is used to set corresponding rules and models according to the channel characteristics of each distribution channel, and automatically adjust the format, style and / or presentation method of the media content from the content evaluation agent and passing the quality inspection and compliance review based on the corresponding rules and models for each distribution channel to obtain the corresponding media content to be displayed; The channel content distribution agent is used to determine the suitable push distribution channel and suitable push timing of the media content based on the manuscript type and release time window of the media content that comes from the content evaluation agent and passes the quality check and compliance review, by analyzing historical data through a large language model and / or a pre-set distribution strategy, and push the to-be-displayed media content that comes from the channel content adaptation agent and corresponds to the suitable push distribution channel to the suitable push distribution channel according to the suitable push timing.

[0008] In a possible design, the multi-terminal distribution multi-task intelligent system further includes a user feedback interaction agent; The user feedback interaction agent is used to collect user feedback data after pushing the corresponding media content to be displayed to the suitable push distribution channel, and then identify high-value users based on the user feedback data and automatically trigger interaction with the high-value users.

[0009] In a possible design, a feedback optimization multi-task intelligent system is also included, wherein the feedback optimization multi-task intelligent system includes an effect evaluation agent and a feedback optimization agent; The effect evaluation agent is used to evaluate and analyze the content quality, user consumption index and / or key factors of the distribution strategy of the distributed media content to obtain multi-dimensional key indicators, wherein the multi-dimensional key indicators include content quality score, user click rate and / or user forwarding rate; The feedback optimization agent is used to automatically generate content optimization suggestions, distribution strategy adjustment plans and / or content lifecycle management plans based on the multi-dimensional key indicators from the effect evaluation agent.

[0010] In a possible design, the feedback optimization multi-task intelligent system also includes an emergency response agent; The emergency response agent is used to monitor negative feedback of the distributed media content, and if it is found during the monitoring process that the negative feedback has triggered a preset risk warning, it automatically triggers the preset emergency measures for the distributed media content.

[0011] In one possible design, each intelligent agent in all intelligent systems is built based on an intelligent agent capability base, wherein the intelligent agent capability base is used to implement a combination of the selection of corresponding large language models driven by the intelligent agent business, task decomposition decisions, knowledge data introduction, internal business interface connection and external capability integration, so as to build intelligent agents with different types of functions.

[0012] In one possible design, the agent capability base is a visual agent process orchestration platform that supports the orchestration of large language model nodes, plug-in nodes, knowledge base nodes, workflow nodes, conditional nodes and / or code nodes; The large language model node is used to support the selection of different AI large language models to process the generation task, wherein the AI ​​large language model includes a cloud service AI large language model and / or a private deployment AI large language model; The plug-in node is used to provide public plug-ins and / or custom plug-ins, wherein the public plug-ins include web search plug-ins, audio processing plug-ins, image understanding plug-ins and / or video processing plug-ins, and the custom plug-ins are used to support the integration of existing API capabilities into the intelligent agent through parameter configuration; The knowledge base node is used to support the RAG framework with a vector database as the engine and / or the GraphRAG framework with a graph database as the engine, so as to build a knowledge brain based on data and enhance the retrieval capability of a large language model; The workflow node is used to support decomposing a complex task into multiple manageable subtasks and facilitate manual definition of the running process of each subtask in the multiple subtasks; The conditional node is used to support logical judgment driven by the joint use of rules and large language models, and to implement decision-making and control of path selection and / or behavior selection for complex tasks; The code node is used to support writing code through an IDE to process input parameters and return output values, thereby realizing the integration of custom module functions.

[0013] In a possible design, the interest prediction agent is also used to dynamically adjust the corresponding user interest prediction results for each user group in combination with context-aware technology and according to environmental factors and corresponding user real-time behaviors.

[0014] In a possible design, an intermediate result review and adjustment module is also included; The intermediate result review and adjustment module is used to display the intermediate results generated by the topic planning agent, the text creation agent, the article illustration agent, the data processing agent, the user portrait agent and / or the interest prediction agent, and respond to manual review and adjustment operations on the intermediate results to optimize the final output of the corresponding agent.

[0015] Beneficial effects of the above scheme: (1) The present invention provides a new solution for media content production and distribution based on multi-agent collaboration, namely, it includes a content creation multi-task intelligent system, an audience analysis multi-task intelligent system and a media content distribution module, wherein the content creation multi-task intelligent system includes a topic planning intelligent agent, a text creation intelligent agent, an article illustration intelligent agent and a content evaluation intelligent agent, and the audience analysis multi-task intelligent system includes a data processing intelligent agent, a user portrait intelligent agent and an interest prediction intelligent agent, and through their functions and collaboration, the whole business process from media content production to distribution can be automatically opened up by introducing AI large language model and multi-agent collaboration technology, and the automated production and personalized distribution of media content can be realized, which greatly improves the production efficiency and dissemination effect of media content and facilitates practical application and promotion; (2) Through the construction of the intelligent agent capability base, it can provide solid technical support for media content production and distribution solutions: Based on the large language model, the intelligent agent system that performs media content production and distribution has certain execution, reasoning, decision-making, memory and reflection capabilities; by building a visual intelligent agent process orchestration platform, it not only simplifies the construction process of multiple types of intelligent agents, but also significantly improves the flexibility and scalability of the intelligent agent system; it can support a variety of node types, not only can it flexibly select multiple types of large language models with differentiated advantages, but also access small models, tool sets, etc. to provide intelligent agents with rich functions and data processing capabilities. At the same time, it can well connect the private data and existing businesses of the media, and support autonomous planning of tasks and different manually defined configurations, thereby improving the performance and business applicability of intelligent agents; it can greatly enhance the automation and personalization level of media content production and distribution, and provide users with a richer and more accurate content experience; (3) In terms of collaborative creation and content innovation, it can significantly improve the automation and efficiency of media content production, while enhancing the diversity and innovation of content: through the collaborative operation of multiple intelligent agents, the content creation process can be optimized, reducing human intervention and shortening the production cycle; the intermediate result output and human intervention mechanism between intelligent agents ensure the controllability of content quality, so that the content not only meets market demand but also has high user appeal; the introduction of personalized creation processes further meets users' demand for novel and personalized content, improves user satisfaction and participation, and brings significant competitive advantages to the media content production industry; (4) It can significantly improve the accuracy of media content distribution and user satisfaction in terms of automated and precise audience analysis: Through automated and precise audience analysis, it can accurately capture user interest preferences and demand changes, and achieve accurate matching of content and users. This not only improves the exposure and click-through rate of content, but also enhances user stickiness and participation, bringing more traffic and revenue to the media platform; it can also reduce the cost and complexity of manual intervention, improve the efficiency and automation level of content distribution, and provide an efficient and intelligent solution for the production and distribution of media content; (5) It can achieve multi-dimensional, flexible, personalized and multi-terminal distribution: Through automated channel adaptation and content optimization, the display of content on different platforms is more in line with user preferences, which improves the attractiveness and conversion rate of content; the multi-channel content distribution strategy ensures that the content can cover a wider audience group, and at the same time, the publishing time is intelligently arranged according to the user's active time to maximize the exposure opportunities of the content; the introduction of user feedback and interactive mechanisms not only enhances user participation, but also provides valuable feedback for content creation and distribution through data analysis, forming a virtuous circle of continuous improvement; overall, it can greatly improve the accuracy, coverage and user satisfaction of media content distribution; (6) Adaptive media content feedback optimization: The production and distribution process of media content has achieved a qualitative leap from static to dynamic, and from one-way to two-way. Through automated effect evaluation and data analysis, enterprises can accurately grasp the market performance and user feedback of content, providing a scientific basis for the formulation of optimization strategies. The application of intelligent feedback loops and optimization strategies enables content to be continuously iterated and upgraded, better in line with user needs, and improves the competitiveness of content and user stickiness. The establishment of emergency response and risk management mechanisms effectively reduces the impact of negative events on content distribution and ensures the healthy dissemination of content. Overall, it can significantly improve the efficiency and quality of media content production and distribution, and enhance the market adaptability and user satisfaction of enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 A schematic diagram of the structure of an integrated system for media content production and distribution based on multi-agent collaboration provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the present invention will be briefly introduced below in combination with the drawings and the description of the embodiments or the prior art. Obviously, the following description of the structures of the drawings is only some embodiments of the present invention. For ordinary technicians in this field, other embodiments can be obtained based on these embodiments without creative work. It should be noted that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation of the present invention.

[0019] It should be understood that although the terms first and second, etc. may be used herein to describe various objects, these objects should not be limited by these terms. These terms are only used to distinguish one object from another object. For example, a first object can be referred to as a second object, and similarly, a second object can be referred to as a first object without departing from the scope of the exemplary embodiments of the present invention.

[0020] It should be understood that the term "and / or" that may appear in this document is merely a description of the association relationship between associated objects, indicating that there may be three relationships. For example, A and / or B can indicate three situations: A exists alone, B exists alone, or A and B exist at the same time. For another example, A, B and / or C can indicate the existence of any one of A, B and C or any combination of them. The term " / and" that may appear in this document describes another type of association object relationship, indicating that there may be two relationships. For example, A / and B can indicate two situations: A exists alone or A and B exist at the same time. In addition, the character " / " that may appear in this document generally indicates that the previous and next associated objects are in an "or" relationship.

[0021] Example like Figure 1 As shown, the media content production and distribution integrated system provided by this embodiment and based on multi-agent collaboration includes but is not limited to a content creation multi-task intelligent system, an audience analysis multi-task intelligent system, and a media content distribution module, etc., wherein the content creation multi-task intelligent system includes but is not limited to a topic planning intelligent agent, a text creation intelligent agent, an article illustration intelligent agent, and a content evaluation intelligent agent, etc., and the audience analysis multi-task intelligent system includes but is not limited to a data processing intelligent agent, a user portrait intelligent agent, and an interest prediction intelligent agent, etc. The aforementioned intelligent agents are instantiated from different types of large language models, and each intelligent agent plays a specific role and assumes corresponding functions. In addition, the aforementioned intelligent agents not only need to have autonomy and decision-making capabilities, but also should be able to understand and predict the behavior of other intelligent agents, as well as transmit information to other intelligent agents. Therefore, when designing and implementing multi-task intelligent systems based on the subsequent intelligent agent capability base, it is also necessary to pay special attention to the interaction mechanism and collaboration strategy between intelligent agents. That is, unlike the independent working mode of a single-agent system, the aforementioned multi-task intelligent system emphasizes the collaborative cooperation between intelligent agents in order to give play to the advantages of collective decision-making. Through the interaction and collaboration between intelligent agents, the flexibility and adaptability of the multi-agent collaborative system are significantly enhanced, and it can complete more complex and challenging tasks than a single agent.

[0022] The topic planning intelligent agent is used to automatically generate or recommend topics that meet market trends based on dynamic information crawling results through big data analysis technology and user demand prediction technology.

[0023] The text creation agent is used to generate text content under the topic according to the topic selected by the topic planning agent in combination with text generation function modules such as outline generation module, knowledge reference module and text generation module. In the specific process of generating the aforementioned text content, text generation function modules such as title generation module and / or summary generation module can also be combined to efficiently generate high-quality text content.

[0024] The article picture matching agent is used to retrieve or generate pictures matching the text content according to the text content from the text creation agent. The visual appeal of the media content can be enhanced by the above picture retrieval or generation.

[0025] The content evaluation agent is used to perform quality inspection and compliance review on media content to ensure the legality and accuracy of the media content, wherein the media content includes but is not limited to text content from the text creation agent and pictures from the article illustration agent.

[0026] By designing the above-mentioned key agents such as the topic planning agent, the text creation agent, the article illustration agent, and the content evaluation agent, a set of efficient and collaborative content creation multi-task intelligent system can be formed to achieve full-chain intelligence and process visualization from creative conception to finished product output. In addition, the above-mentioned agents can be seamlessly connected through standardized interfaces and data exchange protocols to ensure smooth and efficient information flow; and also allow flexible selection of overall processes or specific module processes according to content types or field requirements.

[0027] The data processing agent is used to collect user data from multiple channels and pre-process the user data to obtain pre-processed user data, wherein the multiple channels include but are not limited to social media, search engines and / or user behavior logs, and the pre-processing includes but is not limited to data cleaning processing, data standardization processing and / or data desensitization processing.

[0028] The user portrait agent is used to generate a multi-dimensional user portrait based on the pre-processed user data from the data processing agent, wherein the multi-dimensionality includes but is not limited to physical attribute dimension, interest preference dimension and / or social relationship dimension, etc.

[0029] The interest prediction agent is used to group user portraits from the user portrait agent based on a cluster analysis algorithm, identify user groups with similar interests, and predict the changing trends of user interests of different user groups based on a time series prediction model to obtain user interest prediction results for each user group.

[0030] Based on the above design and key agents such as the data processing agent, the user portrait agent, and the interest prediction agent, a set of efficient and collaborative audience analysis multi-task intelligent system can also be formed, which not only improves the accuracy of media content distribution and user satisfaction, but also helps to bring more traffic and revenue to the media platform. In addition, after the flexible combination of key agents such as the data processing agent, the user portrait agent, and the interest prediction agent, the purpose of supporting the working mode combining large model autonomous planning with manual preset can also be achieved through the subsequent intermediate result review and adjustment module.

[0031] The media content distribution module is used to associate and match the user interest prediction results from the interest prediction agent and each user group with the media content from the content evaluation agent that has passed quality inspection and compliance review, determine personalized matching media content for each user group and distribute it.

[0032] Based on the detailed description of the above-mentioned integrated media content production and distribution system, it can be seen that this embodiment provides a new media content production and distribution solution based on multi-agent collaboration, namely, it includes a content creation multi-task intelligent system, an audience analysis multi-task intelligent system and a media content distribution module, wherein the content creation multi-task intelligent system includes a topic planning intelligent agent, a text creation intelligent agent, an article illustration intelligent agent and a content evaluation intelligent agent, and the audience analysis multi-task intelligent system includes a data processing intelligent agent, a user portrait intelligent agent and an interest prediction intelligent agent, and through their functions and collaboration, the entire business process from media content production to distribution can be automatically opened up by introducing AI large language models and multi-agent collaboration technology, and the automated production and personalized distribution of media content can be realized, which greatly improves the production efficiency and dissemination effect of media content, and facilitates practical application and promotion.

[0033] Preferably, the content creation multi-task intelligent system also includes a video creation intelligent agent; the video creation intelligent agent is used to create and package video content under the topic selected by the topic planning intelligent agent through key links such as script writing, material integration, sound synthesis and video synthesis; the media content also includes video content from the video creation intelligent agent. Through the creation and packaging of the aforementioned video content, the media content can be further enriched and the visual appeal of the media content can be enhanced.

[0034] Preferably, the integrated media content production and distribution system also includes a multi-terminal distribution multi-task intelligent system, wherein the multi-terminal distribution multi-task intelligent system includes but is not limited to a channel content adaptation intelligent agent and a channel content distribution intelligent agent.

[0035] The channel content adaptation agent is used to set corresponding rules and models according to the channel characteristics of each distribution channel, and for each distribution channel, automatically adjust the format, style and / or presentation method of the media content from the content evaluation agent that has passed the quality inspection and compliance review based on the corresponding rules and models to obtain the corresponding media content to be displayed. Specific examples of the aforementioned channel characteristics include but are not limited to: whether it is a social media, whether it is a news website, whether it is a video platform, user group characteristics, content preferences and / or interface design, etc. Through the aforementioned channel content adaptation, it can be ensured that the same media content can obtain the best display effect on different distribution channels.

[0036] The channel content distribution agent is used to determine the suitable push distribution channel and suitable push timing of the media content based on the manuscript type and release time window of the media content from the content evaluation agent and that has passed the quality inspection and compliance review, by analyzing historical data with a large language model / and a pre-set distribution strategy (i.e., analyzing historical data with a large language model, or analyzing historical data with a large language model and a pre-set distribution strategy), and push the to-be-displayed media content from the channel content adaptation agent and corresponding to the suitable push distribution channel to the suitable push distribution channel according to the suitable push timing. Specific examples of the aforementioned manuscript types include, but are not limited to, news articles, entertainment articles, and / or educational articles. Specific examples of the aforementioned distribution strategy include, but are not limited to, giving priority to pushing to platforms or time periods with high user activity.

[0037] Based on the above design and key intelligent entities such as the channel content adaptation intelligent entity and the channel content distribution intelligent entity, a set of efficient and collaborative multi-terminal distribution and multi-task intelligent system can be formed, thereby improving the accuracy and coverage of media content distribution, and bringing significant competitive advantages to the media content production and distribution industry.

[0038] Further preferably, the multi-terminal distribution multi-task intelligent system also includes a user feedback interaction intelligent agent; the user feedback interaction intelligent agent is used to collect user feedback data after pushing the corresponding media content to be displayed to the suitable push distribution channel, and then identify high-value users based on the user feedback data, and automatically trigger interaction with the high-value users. Specific examples of the aforementioned user feedback data include but are not limited to: user comments, likes and / or shares, etc.; specific examples of the aforementioned interactions include but are not limited to: targeted push of related content and / or invitations to participate in discussions, etc. Through the aforementioned user feedback interaction, a closed-loop optimization of content distribution can also be formed, and user participation and satisfaction can be enhanced, further bringing significant competitive advantages to the media content production and distribution industry.

[0039] Preferably, the integrated media content production and distribution system also includes a feedback optimization multi-task intelligent system, wherein the feedback optimization multi-task intelligent system includes but is not limited to an effect evaluation agent and a feedback optimization agent.

[0040] The effect evaluation agent is used to evaluate and analyze the content quality of the distributed media content, user consumption indicators and / or key factors of the distribution strategy to obtain multi-dimensional key indicators, wherein the multi-dimensional key indicators include but are not limited to content quality scores, user click-through rates and / or user forwarding rates, etc.

[0041] The feedback optimization agent is used to automatically generate content optimization suggestions, distribution strategy adjustment plans and / or content lifecycle management plans based on the multi-dimensional key indicators from the effect evaluation agent. Specific examples of the aforementioned content optimization suggestions include, but are not limited to: adjusting content style and / or adding interactive elements, etc.; specific examples of the aforementioned distribution strategy adjustment plans include, but are not limited to: dynamically adjusting distribution strategies based on user preferences and consumption trends, such as optimizing push time and / or adding distribution channels, etc.; the aforementioned content lifecycle management plan is used to ensure continuous optimization and updating of media content during its lifecycle.

[0042] Based on the above design and key agents such as the effect evaluation agent and the feedback optimization agent, a set of efficient and collaborative feedback optimization multi-task intelligent system can be formed, which not only improves the efficiency and quality of media content production and distribution, but also enhances the company's market adaptability and user satisfaction.

[0043] Further preferably, the feedback optimization multi-task intelligent system also includes an emergency response agent; the emergency response agent is used to monitor the negative feedback of the distributed media content, and if it is found that the negative feedback has triggered a preset risk warning during the monitoring process, it automatically triggers the preset emergency measures for the distributed media content. The aforementioned preset emergency measures include, but are not limited to: suspending distribution, deleting content, etc., so as to ensure the safety and stability of content distribution.

[0044] Preferably, each agent in all intelligent systems is built based on the agent capability base, wherein the agent capability base is used to realize the combined application of the corresponding large language model selection, task decomposition decision, knowledge data introduction, internal business interface connection and external capability integration driven by the agent business, so as to build agents with different types of functions. In order to better support the rapid construction of multiple types of agents, it is further preferred that the agent capability base is a visual agent process orchestration platform that supports node orchestration such as large language model nodes, plug-in nodes, knowledge base nodes, workflow nodes, conditional nodes and / or code nodes. Specifically, advanced technologies such as Vue.js (i.e., a JavaScript framework for building user interfaces) can be used at the front end to build the user interaction interface of the agent process orchestration platform to ensure that the editing operation is intuitive and convenient; at the same time, a core business logic processing unit can be built at the back end based on the Python Django framework (i.e., an advanced Python Web framework that can quickly develop secure and maintainable websites) to manage data storage and communication with each node.

[0045] The large language model node is used to support the selection of different AI large language models to process the generation task, wherein the AI ​​large language model includes but is not limited to a cloud service AI large language model and / or a privately deployed AI large language model, etc. Specifically, the large language model node can support the selection of different AI large language models by establishing a standardized interface connection with major cloud service providers and enterprise private models and setting a clear model selection and parameter configuration area in the user interaction interface.

[0046] The plug-in node is used to provide public plug-ins and / or custom plug-ins, wherein the public plug-ins include but are not limited to web search plug-ins, audio processing plug-ins, image understanding plug-ins and / or video processing plug-ins, etc., and the custom plug-ins are used to support the integration of existing API capabilities into the intelligent body through parameter configuration. Specifically, the plug-in node can be developed with an independent plug-in library, the public plug-in is used to encapsulate common tool functions (that is, the web search plug-in, the audio processing plug-in, the image understanding plug-in and the video processing plug-in are all existing conventional tools), and the custom plug-in will provide standardized development templates and parameter configuration wizards to support the integration of existing API (Application Programming Interface) capabilities into the intelligent body through parameter configuration.

[0047] The knowledge base node is used to support the RAG (Retrieval-Augmented Generation) framework based on the vector database as the engine, specifically by combining the retrieval model and the generation model to improve the relevance and quality of the generated content, and / or the GraphRAG (Graph-based Retrieval Augmented Generation) framework based on the graph database as the engine, so as to build a knowledge brain based on data and enhance the retrieval capability of the large language model. Specifically, the knowledge base node can first select the Milvus (which is an open source vector database designed for processing vector search tasks, especially good at processing similarity retrieval of large-scale vector data) vector database and the Neo4j (which is a high-performance NOSQL graph database that stores structured data on the network instead of in a table) graph database, and then develop targeted data import and conversion tools to build the knowledge brain, and implement the natural language and keyword-driven knowledge retrieval interface on the agent process orchestration platform.

[0048] The workflow node is used to support decomposing a complex task into multiple manageable subtasks and facilitate manual definition of the running process of each subtask in the multiple subtasks. Specifically, the workflow node can realize the above functions by developing a visual task editing interface and an execution engine.

[0049] The conditional node is used to support logical judgment driven by the joint use of rules and large language models to achieve decision-making and control of path selection and / or behavior selection for complex tasks. Specifically, the conditional node can achieve the above functions by establishing a rule and model library and integrating intelligent judgment logic.

[0050] The code node is used to support writing code through IDE (Integrated Development Environment, which is an application used to provide a program development environment, generally including tools such as code editors, compilers, debuggers, and graphical user interfaces) to process input parameters and return output values, and realize the integration of custom module functions. Specifically, the code node can build a secure code running container environment (such as Docker) by embedding a full-featured IDE (such as Monaco Editor) and develop a code library management function module.

[0051] In the agent capability base, information sharing, task allocation and collaborative control between agents can also be achieved through communication and coordination mechanisms, thereby improving the performance and efficiency of the entire system, and ultimately achieving the complete construction and effective operation of the agent capability base. In addition, in the process of creating agent instances, the type, number and characteristics of agents need to be designed according to the complexity of the problem and the required functions.

[0052] Preferably, the interest prediction agent is also used to dynamically adjust the corresponding user interest prediction results for each user group in combination with context awareness technology and according to environmental factors and corresponding user real-time behaviors. This can further accurately predict user interest changes and potential needs, providing strong support for personalized content distribution.

[0053] Preferably, the media content production and distribution integrated system also includes an intermediate result review and adjustment module; the intermediate result review and adjustment module is used to display the intermediate results generated by the topic planning agent, the text creation agent, the article illustration agent, the data processing agent, the user portrait agent and / or the interest prediction agent, and respond to the manual review and adjustment operation of the intermediate result to optimize the final output of the corresponding agent. In this way, it can also support the working mode of combining the autonomous planning of the large language model with the manual preset, ensuring that the content creation is both efficient and meets the personalized needs, that is, it not only improves the efficiency and quality of media content production, but also meets the user's urgent needs for personalized, high-quality media content. In addition, for the key agents such as the video creation agent, the channel content adaptation agent, the channel content distribution agent, the user feedback interaction agent, the effect evaluation agent, the feedback optimization agent and the emergency response agent, the intermediate result review and adjustment module can also be used to achieve the purpose of supporting the working mode of combining the autonomous planning of the large language model with the manual preset, ensuring the accuracy and efficiency of content distribution, and ensuring that the media content can continue to adapt to changes in market demand and user preferences.

[0054] In summary, the integrated system for media content production and distribution provided by this embodiment has the following technical effects: (1) This embodiment provides a new solution for media content production and distribution based on multi-agent collaboration, namely, it includes a content creation multi-task intelligent system, an audience analysis multi-task intelligent system and a media content distribution module. The content creation multi-task intelligent system includes a topic planning agent, a text creation agent, an article illustration agent and a content evaluation agent, and the audience analysis multi-task intelligent system includes a data processing agent, a user portrait agent and an interest prediction agent. Through their functions and collaboration, the entire business process from media content production to distribution can be automatically connected by introducing an AI large language model and multi-agent collaboration technology, and the automated production and personalized distribution of media content can be realized, which greatly improves the production efficiency and dissemination effect of media content and facilitates practical application and promotion. (2) Through the construction of the intelligent agent capability base, it can provide solid technical support for media content production and distribution solutions: Based on the large language model, the intelligent agent system that performs media content production and distribution has certain execution, reasoning, decision-making, memory and reflection capabilities; by building a visual intelligent agent process orchestration platform, it not only simplifies the construction process of multiple types of intelligent agents, but also significantly improves the flexibility and scalability of the intelligent agent system; it can support a variety of node types, not only can it flexibly select multiple types of large language models with differentiated advantages, but also access small models, tool sets, etc. to provide intelligent agents with rich functions and data processing capabilities. At the same time, it can well connect the private data and existing businesses of the media, and support autonomous planning of tasks and different manually defined configurations, thereby improving the performance and business applicability of intelligent agents; it can greatly enhance the automation and personalization level of media content production and distribution, and provide users with a richer and more accurate content experience; (3) In terms of collaborative creation and content innovation, it can significantly improve the automation and efficiency of media content production, while enhancing the diversity and innovation of content: through the collaborative operation of multiple intelligent agents, the content creation process can be optimized, reducing human intervention and shortening the production cycle; the intermediate result output and human intervention mechanism between intelligent agents ensure the controllability of content quality, so that the content not only meets market demand but also has high user appeal; the introduction of personalized creation processes further meets users' demand for novel and personalized content, improves user satisfaction and participation, and brings significant competitive advantages to the media content production industry; (4) It can significantly improve the accuracy of media content distribution and user satisfaction in terms of automated and precise audience analysis: Through automated and precise audience analysis, it can accurately capture user interest preferences and demand changes, and achieve accurate matching of content and users. This not only improves the exposure and click-through rate of content, but also enhances user stickiness and participation, bringing more traffic and revenue to the media platform; it can also reduce the cost and complexity of manual intervention, improve the efficiency and automation level of content distribution, and provide an efficient and intelligent solution for the production and distribution of media content; (5) It can achieve multi-dimensional, flexible, personalized and multi-terminal distribution: Through automated channel adaptation and content optimization, the display of content on different platforms is more in line with user preferences, which improves the attractiveness and conversion rate of content; the multi-channel content distribution strategy ensures that the content can cover a wider audience group, and at the same time, the publishing time is intelligently arranged according to the user's active time to maximize the exposure opportunities of the content; the introduction of user feedback and interactive mechanisms not only enhances user participation, but also provides valuable feedback for content creation and distribution through data analysis, forming a virtuous circle of continuous improvement; overall, it can greatly improve the accuracy, coverage and user satisfaction of media content distribution; (6) Adaptive media content feedback optimization: The production and distribution process of media content has achieved a qualitative leap from static to dynamic, and from one-way to two-way. Through automated effect evaluation and data analysis, enterprises can accurately grasp the market performance and user feedback of content, providing a scientific basis for the formulation of optimization strategies. The application of intelligent feedback loops and optimization strategies enables content to be continuously iterated and upgraded, better in line with user needs, and improves the competitiveness of content and user stickiness. The establishment of emergency response and risk management mechanisms effectively reduces the impact of negative events on content distribution and ensures the healthy dissemination of content. Overall, it can significantly improve the efficiency and quality of media content production and distribution, and enhance the market adaptability and user satisfaction of enterprises.

[0055] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A media content production and distribution integrated system based on multi-agent collaboration, characterized in that: It includes a content creation multi-task intelligent system, an audience analysis multi-task intelligent system and a media content distribution module, wherein the content creation multi-task intelligent system includes a topic planning intelligent agent, a text creation intelligent agent, an article illustration intelligent agent and a content evaluation intelligent agent, and the audience analysis multi-task intelligent system includes a data processing intelligent agent, a user portrait intelligent agent and an interest prediction intelligent agent; The topic planning agent is used to automatically generate or recommend topics that meet market trends based on dynamic information crawling results through big data analysis technology and user demand prediction technology; The text creation agent is used to generate text content under the topic according to the topic selected by the topic planning agent, in combination with the outline generation module, the knowledge reference module and the text generation module; The article picture matching agent is used to retrieve or generate pictures matching the text content according to the text content from the text creation agent; The content evaluation agent is used to perform quality inspection and compliance review on media content to ensure the legality and accuracy of the media content, wherein the media content includes text content from the text creation agent and pictures from the article illustration agent; The data processing agent is used to collect user data from multiple channels and pre-process the user data to obtain pre-processed user data, wherein the multiple channels include social media, search engines and / or user behavior logs, and the pre-processing includes data cleaning processing, data standardization processing and / or data desensitization processing; The user portrait agent is used to generate a multi-dimensional user portrait based on the pre-processed user data from the data processing agent, wherein the multi-dimensionality includes a physical attribute dimension, an interest preference dimension and / or a social relationship dimension; The interest prediction agent is used to group the user portraits from the user portrait agent based on a cluster analysis algorithm, identify user groups with similar interests, and predict the changing trends of user interests of different user groups based on a time series prediction model to obtain user interest prediction results for each user group; The media content distribution module is used to associate and match the user interest prediction results from the interest prediction agent and each user group with the media content from the content evaluation agent that has passed quality inspection and compliance review, determine personalized matching media content for each user group and distribute it.

2. The integrated system for producing and distributing media content according to claim 1, characterized in that: The content creation multi-task intelligent system also includes a video creation agent; The video creation agent is used to create and encapsulate video content under the topic according to the topic selected by the topic planning agent through script writing, material integration, sound synthesis and video synthesis. The media content also includes video content from the video creation agent.

3. The integrated system for producing and distributing media content according to claim 1, characterized in that: It also includes a multi-terminal distribution multi-task intelligent system, wherein the multi-terminal distribution multi-task intelligent system includes a channel content adaptation agent and a channel content distribution agent; The channel content adaptation agent is used to set corresponding rules and models according to the channel characteristics of each distribution channel, and automatically adjust the format, style and / or presentation method of the media content from the content evaluation agent and passing the quality inspection and compliance review based on the corresponding rules and models for each distribution channel to obtain the corresponding media content to be displayed; The channel content distribution agent is used to determine the suitable push distribution channel and suitable push timing of the media content based on the manuscript type and release time window of the media content that comes from the content evaluation agent and passes the quality check and compliance review, by analyzing historical data through a large language model and / or a pre-set distribution strategy, and push the to-be-displayed media content that comes from the channel content adaptation agent and corresponds to the suitable push distribution channel to the suitable push distribution channel according to the suitable push timing.

4. The integrated system for producing and distributing media content according to claim 3, characterized in that: The multi-terminal distribution multi-task intelligent system also includes a user feedback interactive intelligent agent; The user feedback interaction agent is used to collect user feedback data after pushing the corresponding media content to be displayed to the suitable push distribution channel, and then identify high-value users based on the user feedback data and automatically trigger interaction with the high-value users.

5. The integrated system for producing and distributing media content according to claim 1, characterized in that: Also included is a feedback optimization multi-task intelligent system, wherein the feedback optimization multi-task intelligent system includes an effect evaluation agent and a feedback optimization agent; The effect evaluation agent is used to evaluate and analyze the content quality, user consumption index and / or key factors of the distribution strategy of the distributed media content to obtain multi-dimensional key indicators, wherein the multi-dimensional key indicators include content quality score, user click rate and / or user forwarding rate; The feedback optimization agent is used to automatically generate content optimization suggestions, distribution strategy adjustment plans and / or content lifecycle management plans based on the multi-dimensional key indicators from the effect evaluation agent.

6. The integrated system for producing and distributing media content according to claim 5, characterized in that: The feedback optimization multi-task intelligent system also includes an emergency response agent; The emergency response agent is used to monitor negative feedback of the distributed media content, and if it is found during the monitoring process that the negative feedback has triggered a preset risk warning, it automatically triggers the preset emergency measures for the distributed media content.

7. The integrated system for producing and distributing media content according to any one of claims 1 to 6, characterized in that: Each intelligent agent in all intelligent systems is built based on the intelligent agent capability base, wherein the intelligent agent capability base is used to realize the combined application of corresponding large language model selection, task decomposition decision, knowledge data introduction, internal business interface connection and external capability integration driven by the intelligent agent business, so as to build intelligent agents with different types of functions.

8. The integrated system for producing and distributing media content according to claim 7, characterized in that: The agent capability base is a visual agent process orchestration platform that supports the orchestration of large language model nodes, plug-in nodes, knowledge base nodes, workflow nodes, conditional nodes and / or code nodes; The large language model node is used to support the selection of different AI large language models to process the generation task, wherein the AI ​​large language model includes a cloud service AI large language model and / or a private deployment AI large language model; The plug-in node is used to provide public plug-ins and / or custom plug-ins, wherein the public plug-ins include web search plug-ins, audio processing plug-ins, image understanding plug-ins and / or video processing plug-ins, and the custom plug-ins are used to support the integration of existing API capabilities into the intelligent agent through parameter configuration; The knowledge base node is used to support the RAG framework with a vector database as the engine and / or the GraphRAG framework with a graph database as the engine, so as to build a knowledge brain based on data and enhance the retrieval capability of a large language model; The workflow node is used to support decomposing a complex task into multiple manageable subtasks and facilitate manual definition of the running process of each subtask in the multiple subtasks; The conditional node is used to support logical judgment driven by the joint use of rules and large language models, and to implement decision-making and control of path selection and / or behavior selection for complex tasks; The code node is used to support writing code through an IDE to process input parameters and return output values, thereby realizing the integration of custom module functions.

9. The integrated system for producing and distributing media content according to claim 1, characterized in that: The interest prediction agent is also used to dynamically adjust the corresponding user interest prediction results for each user group in combination with context perception technology and according to environmental factors and corresponding user real-time behaviors.

10. The integrated system for media content production and distribution as claimed in claim 1, characterized in that: It also includes an intermediate result review and adjustment module; The intermediate result review and adjustment module is used to display the intermediate results generated by the topic planning agent, the text creation agent, the article illustration agent, the data processing agent, the user portrait agent and / or the interest prediction agent, and respond to manual review and adjustment operations on the intermediate results to optimize the final output of the corresponding agent.

Images