Multi-agent-based information processing method and device and electronic equipment

Abstract

The invention provides an information processing method and device based on multiple agents and electronic equipment, and relates to the technical field of artificial intelligence. According to the specific implementation scheme, an information processing request is received, and the information processing request comprises input information; inputting input information to the first intelligent agent, determining one or more second intelligent agents from the intelligent agent set by the first intelligent agent according to the input information, and acquiring output information of the first intelligent agent; and obtaining response information corresponding to the input information according to the output information of the first intelligent agent and the second intelligent agent.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, and in particular to a multi-agent based information processing method, device and electronic equipment. Background Art

[0002] At present, when using intelligent agents as intelligent customer service to communicate with users, traditional intelligent agents lack in-depth understanding of multiple rounds of dialogue and context, which makes it impossible to accurately grasp the user's true intention. Moreover, traditional intelligent agents cannot provide personalized responses based on the user's historical dialogue, user profile and other information, resulting in poor user experience. Summary of the invention

[0003] The present disclosure provides a method, device and electronic device for multi-agent based information processing.

[0004] According to one aspect of the present disclosure, a multi-agent based information processing method is provided, comprising: receiving an information processing request, wherein the information processing request includes input information; inputting the input information to a first agent, and the first agent determines one or more second agents from an agent set based on the input information, and obtains output information of the first agent; and obtaining response information corresponding to the input information based on the output information of the first agent and the second agent.

[0005] According to another aspect of the present disclosure, a multi-agent based information processing device is provided, including: a receiving module for receiving an information processing request, wherein the information processing request includes input information; a first generating module for inputting the input information to a first agent, and the first agent determines one or more second agents from an agent set based on the input information, and obtains output information of the first agent; and a second generating module for obtaining response information corresponding to the input information based on the output information of the first agent and the second agent.

[0006] According to another aspect of the present disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the multi-agent-based information processing method described in the above-mentioned one aspect embodiment.

[0007] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, on which a computer program / instructions are stored, and the computer instructions are used to enable the computer to execute the multi-agent-based information processing method described in the above-mentioned embodiment.

[0008] According to another aspect of the present disclosure, a computer program product is provided, including a computer program / instruction, which, when executed by a processor, implements the multi-agent-based information processing method described in the above-mentioned embodiment.

[0009] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0011] Figure 1 A flowchart of a multi-agent-based information processing method provided in an embodiment of the present disclosure;

[0012] Figure 2 A flowchart of another multi-agent-based information processing method provided in an embodiment of the present disclosure;

[0013] Figure 3 A flowchart of another multi-agent-based information processing method provided in an embodiment of the present disclosure;

[0014] Figure 4 A flowchart of another multi-agent-based information processing method provided in an embodiment of the present disclosure;

[0015] Figure 5 A flowchart of another multi-agent-based information processing method provided in an embodiment of the present disclosure;

[0016] Figure 6 A schematic diagram of the structure of a multi-agent-based information processing method provided in an embodiment of the present disclosure;

[0017] Figure 7 A schematic diagram of the structure of a multi-agent-based information processing device provided in an embodiment of the present disclosure;

[0018] Figure 8 The present invention is a block diagram of an electronic device used to implement the multi-agent based information processing method according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0019] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0020] The following describes the multi-agent based information processing method, device and electronic device of the embodiments of the present disclosure with reference to the accompanying drawings.

[0021] Artificial Intelligence (AI) is a discipline that studies how computers can simulate certain thought processes and intelligent behaviors of human beings (such as learning, reasoning, thinking, planning, etc.). It includes both hardware-level technologies and software-level technologies. AI hardware technologies generally include computer vision technology, speech recognition technology, natural language processing technology, as well as learning / deep learning, big data processing technology, knowledge graph technology, and other aspects.

[0022] An intelligent agent is a computer system or entity that can act autonomously, perceive the environment, make decisions, and interact with the environment. Intelligent agents are systems or machines created by humans that can perform tasks that usually require human intelligence, such as visual recognition, language understanding, decision-making, and translation. Intelligent agents usually rely on large language models as their core decision-making and processing units, and have the ability to think independently and call tools to gradually complete given goals.

[0023] The multi-agent based information processing method provided by the present invention is applicable to multiple industry scenarios including e-commerce platforms, financial institutions, medical consultation, educational counseling, etc., and can provide users with an interactive experience with low configuration cost, controllable logic, high accuracy, and personalized customization.

[0024] The multi-agent-based information processing method provided by the present disclosure is applicable to online platforms that require large-scale user services, customer-oriented enterprise services, and other occasions that require intelligent and highly automated customer service systems. Through the collaboration of multiple agents, the present disclosure can better achieve business demand docking, improve user conversion rate, customer satisfaction and service efficiency, and reduce the transfer rate to manual work.

[0025] Figure 1 A flowchart of a multi-agent based information processing method provided in an embodiment of the present disclosure.

[0026] like Figure 1 As shown, the multi-agent based information processing method may include:

[0027] S101, receiving an information processing request, wherein the information processing request includes input information.

[0028] It should be noted that the execution subject of the multi-agent-based information processing method in the disclosed embodiment may be a hardware device with data processing capabilities and / or the necessary software required to drive the hardware device to work. Optionally, the execution subject may include a server, a user terminal and other intelligent devices. Optionally, the user terminal includes but is not limited to a mobile phone, a computer, an intelligent voice interaction device, etc. Optionally, the server includes but is not limited to a network server, an application server, and may also be a server of a distributed system, or a server combined with a blockchain, etc. The disclosed embodiment is not specifically limited.

[0029] In some implementations, the information processing request may be generated based on the user's input information. That is, the information processing request is generated based on the input information by obtaining the user's input information. Optionally, the query statement input by the user may be used as the input information.

[0030] Optionally, the user's input information can be obtained from the agent's data center.

[0031] S102, input information is input to the first agent, and the first agent determines one or more second agents from the agent set according to the input information, and obtains output information of the first agent.

[0032] In some implementations, input information is input into a first agent, which processes the input information to obtain the user's intention from the input information, and plans the agents in the agent set according to the user's intention to determine one or more second agents. The first agent may be a global planning agent.

[0033] Optionally, the input information may be classified to determine the type of intent corresponding to the input information as the user intent. Optionally, the input information may be matched using a pre-set matching rule to determine the user intent. A pre-trained large model may also be used to identify the user intent based on the input information.

[0034] In some implementations, after receiving input information, the first agent can also generate agent planning status and standard operating procedure (SOP) process information, and use the input information, agent planning status, and SOP process information as output information so that the second agent can respond according to the output information.

[0035] Optionally, after receiving the input information, the first agent may determine whether the input information needs to be adjusted, and if so, the adjusted input information is used as one of the output information; if not, the input information is used as one of the output information. Optionally, it may be determined whether the input information needs to be adjusted based on the format and content of the input information.

[0036] Optionally, the agent set includes but is not limited to: speech generation agent, retrieval enhancement generation agent, active dialogue agent, dialogue image agent, dialogue video agent, image recognition agent and information search agent, etc.

[0037] It can be understood that the speech generation agent is used to generate speech, and a large language model (LLM) with parameters ranging from tens to hundreds of billions can be used as the speech generation agent.

[0038] The retrieval-augmented generation agent is responsible for vector retrieval recall of the data center (storing semi-structured and structured data) based on the input information and generating replies, clarifications or rejections. The vector retrieval part uses an arbitrary retrieval-augmented generation (RAG) system. The definitions of replies, clarifications, and rejections can all be configured and defined through natural language, generally using LLMs with parameters of 10 billion or less.

[0039] The active dialogue agent is responsible for generating active dialogue scripts based on the user's historical dialogue, user profile and other information when the user is silent. Generally, LLM with parameters ranging from tens to hundreds of billions is used.

[0040] Dialogue-to-image agent refers to the process of converting text descriptions into images using artificial intelligence technology. The agent can understand the semantic information in the text and convert it into corresponding image content.

[0041] The dialogue-generating video agent is a machine that uses artificial intelligence technology to transform descriptions into dynamic video content. The agent not only needs to understand the semantic information in the text, but also needs to understand the movement and interaction of objects in three-dimensional space to generate video information.

[0042] Image recognition agent is an intelligent system that can autonomously perceive image information, analyze and recognize images, and make decisions or perform related tasks based on them. It uses advanced image recognition technologies, such as convolutional neural networks, to extract features, classify and recognize input images.

[0043] Information search agent is an intelligent system that can autonomously search, filter and integrate information on the Internet and provide corresponding information services according to user needs. It uses advanced search engine technology, natural language processing technology and data mining technology to achieve rapid search and accurate matching of network information.

[0044] For example, if the input information is "introduce item A", the first agent identifies the input information and determines the speech generation agent as the second agent to generate an introduction to item A. If the input information is "introduce item A in style X", the first agent identifies the input information and determines the speech generation agent and the search enhancement generation agent as the second agent to generate an introduction to item A in style X.

[0045] S103, obtaining response information corresponding to the input information according to the output information of the first agent and the second agent.

[0046] In some implementations, a prompt word of the second agent can be generated based on the output information of the first agent, so that the second agent can respond to the input information based on the prompt word to obtain response information corresponding to the input information. The response information can be obtained by inputting the prompt word into the second agent.

[0047] Optionally, the auxiliary information of the second agent can be obtained to combine the auxiliary information and the output information to determine the prompt word of the second agent. In other words, the auxiliary information and the output information can be used as the prompt word of the second agent.

[0048] Optionally, auxiliary information of the second agent may be obtained from the configuration center of the agent, wherein the auxiliary information includes but is not limited to: dialogue strategy, user portrait, historical dialogue, etc.

[0049] In some implementations, when there are multiple second agents, the upstream and downstream relationships between the second agents are determined, and the output information of the upstream agent is used as the prompt word of the downstream agent, so that the second agent can determine the response information according to the input information.

[0050] According to the multi-agent information processing method provided by the embodiment of the present disclosure, an information processing request containing input information is obtained, and the first agent determines the required one or more second agents and the output information of the first agent based on the input information. Further, the second agent determines the response information corresponding to the input information based on the output information, so as to improve the efficiency and accuracy of the agent's response to the input information, and improve the problem solving efficiency and recall rate. Based on multiple agents to respond, the number of interactions and dialogue effects between users and agents are effectively improved.

[0051] Figure 2A flowchart of a multi-agent based information processing method provided in an embodiment of the present disclosure.

[0052] like Figure 2 As shown, the multi-agent based information processing method may include:

[0053] S201, receiving an information processing request, wherein the information processing request includes input information.

[0054] The relevant contents of step S201 can be found in the above embodiment and will not be described again here.

[0055] S202, input information is input to the first agent, and the first agent determines one or more second agents from the agent set according to the input information, and obtains output information of the first agent.

[0056] In some implementations, user intent can be determined based on input information, and based on the user intent, one or more agents can be determined to ensure personalized and differentiated agent responses, thereby providing response information with emotional value.

[0057] In some implementations, the input information can be used to identify the user's intention based on the auxiliary information of the first agent. Optionally, user portraits and historical behavior data can be obtained from the data center of the agent as the auxiliary information of the first agent.

[0058] Furthermore, the first agent identifies the intent of the input information based on the auxiliary information, obtains an intent information set, the intent information set includes at least one user intent, and determines one or more second agents from the agent set based on the user intent.

[0059] In some implementations, before determining the second agent, you can also first determine the corresponding agent set based on the current business scenario to determine one or more second agents from the agent set, so as to screen out the agent set that best matches the business needs, which helps to ensure that the selected second agent can accurately meet the business needs.

[0060] Optionally, the current target business scenario can be determined based on the input information, and the agent associated with the target business scenario can be obtained. Then, the agent set corresponding to the target business scenario can be determined based on the agent associated with the target business scenario. Optionally, an association relationship between the business scenario and the agent can be established in advance, and after the target business scenario is determined, the agent associated with the target business scenario can be determined by querying the association relationship.

[0061] In some implementations, it is also possible to determine sets of agents corresponding to different industries based on the industry to which the agent belongs and the industry of the user, so as to provide automated and intelligent interaction and improve the service efficiency of different industries. Optionally, by obtaining the industry to which the intelligent service belongs and the industry of the user corresponding to the input information, and according to the industry to which the intelligent service belongs and the industry of the user corresponding to the input information, the set of agents is obtained.

[0062] Optionally, the agent set includes at least one of the following agents: a speech generation agent, a retrieval enhancement generation agent, an active dialogue agent, a dialogue image agent, a dialogue video agent, an image recognition agent and an information search agent.

[0063] S203, determining the auxiliary information required by the second agent, where the auxiliary information includes at least one of a pre-configured dialogue strategy, a user portrait, and historical dialogues.

[0064] In the disclosed embodiment, the second agent is one, that is, the second agent can be a speech generation agent, a retrieval enhancement generation agent, or an active dialogue agent.

[0065] In some implementations, in order to improve the execution efficiency and accuracy of the second agent and optimize the user's interaction experience with the second agent, the auxiliary information required by the second agent can be determined, and the prompt word can be determined based on the auxiliary information.

[0066] Optionally, the target business corresponding to the second agent can be obtained, and according to the target business, the configuration information of the target business corresponding to the second agent can be obtained, and the auxiliary information required by the second agent can be obtained from the configuration information. For example, the configuration information can be determined from the configuration center, and the auxiliary information can be determined from the configuration information. The auxiliary information includes at least one of the pre-configured dialogue strategy, user profile, and historical dialogue.

[0067] Optionally, the configuration center includes a configuration information list, and the configuration information in the configuration information list corresponds to the target service. The matching information list can be queried according to the target service to determine the configuration information of the target service, and the auxiliary information can be determined from the configuration information.

[0068] S204, generating prompt words for the second agent according to the auxiliary information and the output information of the first agent.

[0069] Optionally, a prompt word template may be pre-set, and the auxiliary information and the output information of the first agent may be added to the prompt word template to obtain the prompt word of the second agent.

[0070] S205, obtaining response information corresponding to the input information according to the prompt word of the second agent.

[0071] In some implementations, the prompt word is input into the second agent, and the second agent generates response information corresponding to the input information according to the prompt word. Optionally, the response information corresponding to the input information can be generated according to a set output format.

[0072] According to the multi-agent information processing method provided by the embodiment of the present disclosure, an information processing request containing input information is obtained, and the first agent identifies the user intent in the input information to determine one or more required second agents and the output information of the first agent according to the user intent. Further, the auxiliary information of the second agent is determined, and the prompt word of the second agent is determined based on the auxiliary information and the output information, and the second agent determines the response information corresponding to the input information based on the prompt word, so as to improve the efficiency and accuracy of the agent's response to the input information, and improve the problem solving efficiency and recall rate. Responding based on multiple agents effectively improves the number of interactions and dialogue effects between users and agents.

[0073] Figure 3 A flowchart of a multi-agent based information processing method provided in an embodiment of the present disclosure.

[0074] like Figure 3 As shown, the multi-agent based information processing method may include:

[0075] S301, receiving an information processing request, wherein the information processing request includes input information.

[0076] S302, input information is input to the first agent, and the first agent determines one or more second agents from the agent set according to the input information, and obtains output information of the first agent.

[0077] The relevant contents of steps S301 - S302 can be found in the above embodiment and will not be described again here.

[0078] S303: Determine the execution order of the plurality of second agents and the dependency relationship between the second agents according to the output information of the first agent.

[0079] In some implementations, when it is determined that there are multiple second agents, the output information of the first agent can be parsed to determine the execution order of the multiple second agents contained in the output information and the dependency relationship between the second agents.

[0080] For example, the execution order of multiple second agents and the dependencies between the second agents can be determined based on the agent planning status and SOP process information in the output information.

[0081] S304, calling multiple second agents in the execution order, and determining the input information of the currently called second agent according to the dependency relationship.

[0082] In some implementations, the upstream and downstream relationships between the second agents can be determined based on the dependency relationship, and the output of the upstream agent can be used as the input of the downstream agent to optimize the task process and improve the collaborative efficiency of the system. In other words, the upstream agent that the currently called second agent depends on can be determined based on the dependency relationship, and the upstream agent that is relied on is at least one of the multiple second agents.

[0083] Furthermore, the output information of the upstream agent on which it depends is obtained and determined as the input information of the second agent currently called. For example, the second agent includes: a speech generation agent and a search enhancement generation agent. The second agent currently called is a speech generation agent, and the upstream agent it depends on is a search enhancement generation agent. That is to say, search enhancement is performed first, and then the output information of the search enhancement generation agent is used as the input information of the speech generation agent.

[0084] S305: Generate a prompt word for the currently called second agent according to the auxiliary information of the second agent and the corresponding input information.

[0085] Optionally, a prompt word template may be pre-set, and the auxiliary information of the second agent and the corresponding output information may be added to the prompt word template, so as to obtain the prompt word of the currently called second agent.

[0086] S306, obtaining the second output information corresponding to the currently called second agent according to the prompt word of the second agent, until the sequential execution of multiple second agents is completed, and the response information corresponding to the input information is obtained.

[0087] In some implementations, a prompt word is input into the second agent currently called, and the second agent currently called generates second output information based on the prompt word, and uses the second output information as input information of a downstream agent corresponding to the second agent currently called, and generates a corresponding prompt word based on the input information of the downstream agent. Further, the downstream agent can generate output information based on the prompt word until multiple second agents are called according to the execution order, and the output information of the last called second agent is used as response information corresponding to the input information.

[0088] According to the multi-agent information processing method provided by the embodiment of the present disclosure, an information processing request containing input information is obtained, and the first agent determines the required one or more second agents and the output information of the first agent based on the input information. Further, by determining the execution order and dependency relationship of the second agent, and determining the input information of the second agent currently called based on the dependency relationship, and determining the prompt word of the second agent based on the input information, so as to determine the second output information corresponding to the second agent currently called based on the prompt word, and further completing the call of the second agent according to the execution order, the response information corresponding to the input information is obtained, so as to improve the efficiency and accuracy of the agent's response to the input information, and improve the problem solving efficiency and recall rate. Responding based on multiple agents effectively improves the number of interactions and dialogue effects between users and agents.

[0089] Figure 4 A flowchart of a multi-agent based information processing method provided in an embodiment of the present disclosure.

[0090] like Figure 4 As shown, the multi-agent based information processing method may include:

[0091] S401, receiving an information processing request, wherein the information processing request includes input information.

[0092] S402, input information to the first agent, the first agent determines one or more second agents from the agent set according to the input information, and obtains output information of the first agent.

[0093] The relevant contents of steps S401 - S402 can be found in the above embodiment and will not be described again here.

[0094] S403: For each second intelligent agent, determine the auxiliary information required by the second intelligent agent.

[0095] In some implementations, the target business of each second agent is determined, and the configuration information corresponding to the target business is determined, so as to obtain the auxiliary information required by the second agent from the configuration information. For example, the configuration information can be determined from the configuration center according to the target business, and the auxiliary information can be determined from the configuration information.

[0096] S404: Determine associated output information matching the second agent from the output information.

[0097] In some implementations, the output information may be screened to determine the associated output information that matches the second agent. Alternatively, the associated output information that matches the second agent may be screened from the output information based on characteristic information of the second agent, such as characteristics, requirements, functions, etc.

[0098] S405, generating a prompt word for each second agent according to the auxiliary information of the second agent and the matched associated output information.

[0099] Optionally, for each second agent, a prompt word for each second agent may be determined based on a preset prompt word template. The prompt word for each second agent may be obtained by adding the auxiliary information of the second agent and the matched associated output information to the prompt word template.

[0100] S406, obtaining first output information corresponding to the associated output information according to the prompt word of the second agent.

[0101] S407: Generate response information corresponding to the input information according to the first output information of each second agent.

[0102] In some implementations, the prompt word is input into the second agent, and the second agent generates first output information corresponding to the input information according to the prompt word. Further, the first output information corresponding to each second agent is spliced ​​to obtain the response information corresponding to the input information.

[0103] Optionally, the first output information may be respectively used as response information, that is, one input information may correspond to multiple response information.

[0104] According to the multi-agent information processing method provided by the embodiment of the present disclosure, an information processing request containing input information is obtained, and the first agent determines the required one or more second agents and the output information of the first agent based on the input information. Further, the auxiliary information required by each second agent is determined, and the associated output information matched by the second agent is determined from the output information to determine the prompt word of the second agent, so that the second agent determines the response information corresponding to the input information based on the prompt word, so as to improve the efficiency and accuracy of the agent's response to the input information, and improve the problem solving efficiency and recall rate. Responding based on multiple agents effectively improves the number of interactions and dialogue effects between users and agents.

[0105] Figure 5 A flowchart of a multi-agent based information processing method provided in an embodiment of the present disclosure.

[0106] like Figure 5 As shown, the multi-agent based information processing method may include:

[0107] S501: Receive an information processing request, where the information processing request includes input information.

[0108] S502: Input information is input to the first agent, and the first agent determines one or more second agents from the agent set according to the input information, and obtains output information of the first agent.

[0109] S503: Acquire response information corresponding to the input information according to the output information of the first agent and the second agent.

[0110] The relevant contents of steps S501-S503 can be found in the above embodiment and will not be described again here.

[0111] S504, start at least one third agent for dialogue monitoring, and obtain dialogue content during the dialogue process.

[0112] Optionally, the third agent includes: a conversation analysis agent and a conversation quality inspection agent. By starting the third agent and obtaining the conversation content between the user and the second agent, the third agent can analyze the conversation content to promptly discover the problems encountered by the user in the conversation, and quickly respond and solve them, thereby improving user satisfaction, and optimizing the agent based on the analysis results, thereby improving the learning ability of the agent.

[0113] S505, performing intelligent analysis on the conversation content through a third intelligent agent.

[0114] In some implementations, if the third agent is a dialogue analysis agent, the third agent can be used to extract information from the dialogue content to obtain key information, which is used to optimize subsequent dialogues. The third agent can also be used to clarify and reflect on the dialogue content to obtain dialogue reflection results, and optimize the agent based on the reflection results, thereby improving the learning ability of the agent and enhancing the adaptability of the agent.

[0115] Optionally, by determining the key information and the dialogue reflection results as the analysis results of the dialogue analysis agent, and performing structured processing on the analysis results of the dialogue analysis agent, first structured data is obtained. Further, the first structured data is determined to be shared data and stored in a data center, and the first structured data can be called by at least one of the first agent and the second agent.

[0116] That is to say, by calling the structured data in the data center, the first agent and the second agent can optimize themselves according to the structured data to improve the learning ability of the agent and enhance the adaptability of the agent.

[0117] In some implementations, if the third agent is a conversation quality inspection agent, the third agent can be used to perform hallucination monitoring and compliance monitoring on the conversation content, obtain quality inspection results, and fine-tune the agent based on the quality inspection results, so as to correct the agent's errors and hallucinations and improve the accuracy of the agent's responses.

[0118] Optionally, based on the quality detection results, abnormal conversation content can be determined and collected as sample data for fine-tuning the agent. The sample data is used to fine-tune the relevant agent to obtain a fine-tuned agent, and the fine-tuned agent is used for the next conversation.

[0119] According to the multi-agent-based information processing method provided by the embodiment of the present disclosure, an information processing request containing input information is obtained, and the first agent determines the required one or more second agents and the output information of the first agent based on the input information. Further, the second agent determines the response information corresponding to the input information based on the output information to improve the efficiency and accuracy of the agent's response to the input information. Responding based on multiple agents effectively improves the number of interactions and dialogue effects between users and agents. By obtaining the content of the dialogue and using a third agent to analyze the content of the dialogue, problems encountered by users in the dialogue can be discovered in a timely manner, and quickly responded to and resolved, thereby improving user satisfaction, and optimizing the agent based on the analysis results to improve the learning ability of the agent.

[0120] By way of example, the information processing method based on multi-agent provided by the embodiment of the present disclosure is explained by taking the pre-sales scenario in the marketing field to acquire customers and collect customer leads as an example:

[0121] Among them, the first agent is the global planning agent, the second agent is the speech generation agent, the retrieval enhancement generation agent, and the active dialogue agent, and the third agent is the dialogue analysis agent and the dialogue quality inspection agent.

[0122] By inputting the input information into the global planning agent, the second agent can be determined as the speech generation agent, the search enhancement generation agent, and the active dialogue agent, and the planning strategy, the final goal, and the secondary goal are generated as the output information of the first agent. Among them, the planning strategy is a specific description of the strategic planner of the marketing scenario, the final goal is to acquire customers, and the secondary goal is to obtain customer lead information.

[0123] By determining the execution order of the second agent as the speech generation agent, the retrieval enhancement generation agent, and the active dialogue agent, further, according to the auxiliary information of the speech generation agent, the prompt words of the speech generation agent are determined: the pre-sales service role in the marketing scenario, obtaining a specific description of customer lead information, and providing a high-quality customer service dialogue experience. By inputting the prompt words into the speech generation agent, the corresponding second output information is obtained, and then the prompt words of the retrieval enhancement generation agent are generated based on the second output information: for answering questions in the sales scenario. By inputting the prompt words into the retrieval enhancement generation agent, the corresponding second output information is obtained, and then the prompt words of the active dialogue agent are generated based on the second output information: a personal assistant in the private domain marketing scenario, combining historical conversations with users to regain contact.

[0124] Furthermore, by acquiring the conversation content during the conversation process and using a conversation quality inspection agent and a conversation analysis agent, the conversation content is analyzed to optimize the agent based on the analysis results.

[0125] like Figure 6 Shown is a schematic diagram of the structure of the multi-agent based information processing method. Figure 6 It includes a data center, a configuration center, and multiple agents: a global planning agent, a speech generation agent, a retrieval enhancement generation agent, an active dialogue agent, a dialogue analysis agent, and a dialogue quality inspection agent.

[0126] Among them, the data center stores user data in structured and semi-structured forms, such as historical conversations, conversation content analysis results, etc.; the configuration center stores auxiliary information of the intelligent agent, such as conversation strategies, trigger strategies, user portraits, business descriptions, business SOPs and other information.

[0127] Based on the user's input information, the global planning agent generates the planning logic and execution list of the second agent as output information, and calls one or more second agents. Taking the second agent as a speech generation agent as an example, it obtains dialogue strategies and user portraits from the configuration center as auxiliary information, and generates prompt words based on the auxiliary information and the output information of the global planning agent to determine the response information corresponding to the input information according to the prompt words.

[0128] Taking the second agent as the retrieval-augmented generation agent as an example, the dialogue strategy and the retrieval-augmented generation (RAG) system are obtained from the configuration center as auxiliary information, and prompt words are generated based on the auxiliary information and the output information of the global planning agent to determine the response information corresponding to the input information according to the prompt words.

[0129] Taking the second agent as an active dialogue agent as an example, by obtaining dialogue strategies and trigger strategies from the configuration center as auxiliary information, and generating prompt words based on the auxiliary information and the output information of the global planning agent, the response information corresponding to the input information is determined according to the prompt words.

[0130] In the process of agent-based dialogue, at least one of the dialogue analysis agent and the dialogue quality inspection agent can be used to monitor the dialogue content, and the agent can be used to analyze the dialogue content. Taking the dialogue analysis agent as an example, key information can be obtained by extracting information from the dialogue content, and the key information is used to optimize subsequent dialogues. The dialogue content can also be clarified and reflected to obtain dialogue reflection results.

[0131] Taking the dialogue quality inspection agent as an example, by performing hallucination monitoring and compliance monitoring on the dialogue content, the quality inspection results are obtained, and the relevant agents are fine-tuned and trained based on the quality inspection results.

[0132] Corresponding to the multi-agent based information processing methods provided in the above-mentioned embodiments, an embodiment of the present disclosure also provides a multi-agent based information processing device. Since the multi-agent based information processing device provided in the embodiment of the present disclosure corresponds to the multi-agent based information processing methods provided in the above-mentioned embodiments, the implementation methods of the above-mentioned multi-agent based information processing methods are also applicable to the multi-agent based information processing device provided in the embodiment of the present disclosure, and will not be described in detail in the following embodiments.

[0133] Figure 7 A schematic diagram of the structure of a multi-agent based information processing device provided in an embodiment of the present disclosure.

[0134] like Figure 7 As shown, the multi-agent based information processing device 700 of the embodiment of the present disclosure includes a receiving module 701, a first generating module 702 and a second generating module 703.

[0135] The receiving module 701 is used to receive an information processing request, wherein the information processing request includes input information;

[0136] A first generating module 702 is used to input the input information to a first agent, and the first agent determines one or more second agents from an agent set according to the input information, and obtains output information of the first agent;

[0137] The second generating module 703 is used to obtain response information corresponding to the input information according to the output information of the first agent and the second agent.

[0138] In one embodiment of the present disclosure, the second generation module 703 is further used to: determine the auxiliary information required by the second agent, the auxiliary information including at least one of a pre-configured dialogue strategy, a user portrait and a historical dialogue; generate a prompt word for the second agent based on the auxiliary information and the output information of the first agent; and obtain response information corresponding to the input information based on the prompt word of the second agent.

[0139] In one embodiment of the present disclosure, the second generation module 703 is also used to: determine the execution order of multiple second agents and the dependency relationship between the second agents based on the output information of the first agent; call the multiple second agents in accordance with the execution order, and determine the input information of the currently called second agent based on the dependency relationship; generate a prompt word for the currently called second agent based on the auxiliary information of the second agent and the corresponding input information; obtain the second output information corresponding to the currently called second agent based on the prompt word of the second agent, until the sequential execution of the multiple second agents is completed and the response information corresponding to the input information is obtained.

[0140] In one embodiment of the present disclosure, the second generation module 703 is also used to: determine the upstream agent on which the second agent currently called depends based on the dependency relationship, and the dependent upstream agent is at least one of the multiple second agents; obtain the output information of the dependent upstream agent and determine it as the input information of the second agent currently called.

[0141] In one embodiment of the present disclosure, the second generation module 703 is also used to: determine the auxiliary information required by each second intelligent agent; determine the associated output information matching the second intelligent agent from the output information; generate a prompt word for each second intelligent agent based on the auxiliary information of the second intelligent agent and the matched associated output information; obtain the first output information corresponding to the associated output information based on the prompt word of the second intelligent agent; and generate response information corresponding to the input information based on the first output information of each second intelligent agent.

[0142] In one embodiment of the present disclosure, the device also includes: an acquisition module for starting at least one third agent for conversation monitoring and acquiring conversation content during the conversation; and an analysis module for performing intelligent analysis of the conversation content through the third agent.

[0143] In one embodiment of the present disclosure, the analysis module is further used to: extract information from the conversation content through the third agent to obtain key information, and the key information is used to optimize subsequent conversations; or, clarify and reflect on the conversation content through the third agent to obtain a conversation reflection result.

[0144] In one embodiment of the present disclosure, the analysis module is further used to: determine that the key information and the dialogue reflection results are the analysis results of the dialogue analysis agent; perform structured processing on the analysis results of the dialogue analysis agent to obtain first structured data; determine that the first structured data is shared data and store it in a data center, and the first structured data can be called by at least one of the first agent and the second agent.

[0145] In one embodiment of the present disclosure, the analysis module is further used to: perform hallucination monitoring and compliance monitoring on the conversation content through the third agent to obtain quality detection results.

[0146] In one embodiment of the present disclosure, the analysis module is also used to: determine abnormal conversation content based on the quality detection result, and collect the abnormal conversation content as sample data for fine-tuning the agent, and the sample data is used to fine-tune the relevant agent.

[0147] In one embodiment of the present disclosure, the second generation module 703 is further used to: obtain the target business corresponding to the second agent; obtain configuration information of the target business corresponding to the second agent, and obtain auxiliary information required by the second agent from the configuration information.

[0148] In one embodiment of the present disclosure, the first generation module 702 is also used to: obtain user portraits and historical behavior data as auxiliary information of the first agent; the first agent identifies the intent of the input information based on the auxiliary information to obtain an intent information set, wherein the intent information set includes at least one user intent; and determine the one or more second agents from the agent set based on the user intent.

[0149] In one embodiment of the present disclosure, the first generation module 702 is further used to: determine the current target business scenario based on the input information; obtain the intelligent agent associated with the target business scenario; and determine the intelligent agent set corresponding to the target business scenario based on the intelligent agent associated with the target business scenario.

[0150] In one embodiment of the present disclosure, the first generation module 702 is further used to: obtain the industry to which the intelligent service belongs and the industry of the user corresponding to the input information; and obtain a set of intelligent agents based on the industry to which the intelligent service belongs and the industry of the user corresponding to the input information.

[0151] In one embodiment of the present disclosure, the agent set includes at least one of the following agents: a speech generation agent, a retrieval enhancement generation agent, an active dialogue agent, a dialogue image agent, a dialogue video agent, an image recognition agent and an information search agent.

[0152] According to the information processing device based on multiple agents provided by the embodiment of the present disclosure, an information processing request containing input information is obtained, and the first agent determines the required one or more second agents and the output information of the first agent based on the input information. Further, the second agent determines the response information corresponding to the input information based on the output information, so as to improve the efficiency and accuracy of the agent's response to the input information, and improve the problem solving efficiency and recall rate. Based on multiple agents to respond, the number of interactions and dialogue effects between users and agents are effectively improved.

[0153] In the technical solution disclosed herein, the acquisition, storage and application of user personal information involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0154] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0155] Figure 8 A schematic block diagram of an example electronic device 800 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0156] like Figure 8As shown, the device 800 includes a computing unit 801, which can perform various appropriate actions and processes according to a computer program / instruction stored in a read-only memory (ROM) 802 or a computer program / instruction loaded from a storage unit 806 to a random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the device 800 can also be stored. The computing unit 801, the ROM 802, and the RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0157] A number of components in the device 800 are connected to the I / O interface 805, including: an input unit 806 such as a keyboard, a mouse, etc.; an output unit 807 such as various types of displays, speakers, etc.; a storage unit 808 such as a disk, an optical disk, etc.; and a communication unit 809 such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 809 allows the device 800 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0158] The computing unit 801 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 801 performs the various methods and processes described above, such as a multi-agent-based information processing method. For example, in some embodiments, the multi-agent-based information processing method may be implemented as a computer software program, which is tangibly included in a machine-readable medium, such as a storage unit 806. In some embodiments, part or all of the computer program / instructions may be loaded and / or installed on the device 800 via the ROM 802 and / or the communication unit 809. When the computer program / instructions are loaded into the RAM 803 and executed by the computing unit 801, one or more steps of the multi-agent-based information processing method described above may be performed. Alternatively, in other embodiments, the computing unit 801 may be configured to execute a multi-agent-based information processing method in any other appropriate manner (eg, by means of firmware).

[0159] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs / instructions that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0160] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0161] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0162] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0163] The systems and techniques described herein can be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), the Internet, and a blockchain network.

[0164] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs / instructions running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0165] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in the disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in the disclosure can be achieved, and this document does not limit them here.

[0166] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A multi-agent based information processing method, wherein: The method comprises: receiving an information processing request, wherein the information processing request includes input information; Inputting the input information to the first agent, the first agent determining one or more second agents from the agent set according to the input information, and obtaining output information of the first agent; According to the output information of the first agent and the second agent, response information corresponding to the input information is obtained.

2. The method according to claim 1, wherein: The step of obtaining response information corresponding to the input information according to the output information of the first agent and the second agent includes: Determining auxiliary information required by the second agent, the auxiliary information comprising at least one of a pre-configured dialogue strategy, a user profile, and a historical dialogue; generating a prompt word for the second agent according to the auxiliary information and the output information of the first agent; Acquire response information corresponding to the input information according to the prompt word of the second agent.

3. The method according to claim 1, wherein: When there are multiple second agents, obtaining response information corresponding to the input information according to the output information of the first agent and the second agent includes: Determining, according to the output information of the first agent, the execution order of the plurality of second agents and the dependency relationship between the second agents; Invoke the plurality of second agents in the execution order, and determine input information of the second agent currently being invoked according to the dependency relationship; Generate a prompt word for the currently called second agent according to the auxiliary information of the second agent and the corresponding input information; The second output information corresponding to the second agent currently called is obtained according to the prompt word of the second agent, until the sequential execution of multiple second agents is completed, and the response information corresponding to the input information is obtained.

4. The method according to claim 3, wherein: The step of determining the input information of the second agent currently being called according to the dependency relationship includes: According to the dependency relationship, determining an upstream agent that the second agent currently called depends on, wherein the upstream agent that is dependent on is at least one of the plurality of second agents; The output information of the dependent upstream agent is obtained and determined as the input information of the second agent currently called.

5. The method according to claim 1, wherein: When there are multiple second agents, obtaining response information corresponding to the input information according to the output information of the first agent and the second agent includes: For each second agent, determining the auxiliary information required by the second agent; Determining associated output information matching the second agent from the output information; generating a prompt word for each second agent according to the auxiliary information of the second agent and the matched associated output information; Acquire first output information corresponding to the associated output information according to the prompt word of the second agent; According to the first output information of each of the second agents, response information corresponding to the input information is generated.

6. The method according to any one of claims 2 to 5, wherein: The method further comprises: Initiate at least one third agent for dialogue monitoring and obtain dialogue content during the dialogue process; The conversation content is intelligently analyzed by the third agent.

7. The method according to claim 6, wherein: The third agent is a conversation analysis agent, and the intelligent analysis of the conversation content by the third agent includes: Extracting information from the conversation content by the third agent to obtain key information, wherein the key information is used to optimize subsequent conversations; or The third agent clarifies and reflects on the content of the conversation to obtain a conversation reflection result.

8. The method according to claim 7, wherein: The method further comprises: Determine the key information and the dialogue reflection result as the analysis result of the dialogue analysis agent; Performing structured processing on the analysis result of the dialogue analysis agent to obtain first structured data; The first structured data is determined to be shared data and stored in a data center, and the first structured data can be called by at least one of the first agent and the second agent.

9. The method according to any one of claims 2 to 5, wherein: The third agent is a conversation quality inspection agent, and the intelligent analysis of the conversation content by the third agent includes: The third agent performs hallucination monitoring and compliance monitoring on the conversation content to obtain quality detection results.

10. The method according to claim 9, wherein: After obtaining the quality test result, the method further includes: Based on the quality detection results, abnormal conversation content is determined, and the abnormal conversation content is collected as sample data for fine-tuning the intelligent agent, and the sample data is used to perform fine-tuning training on the relevant intelligent agent.

11. The method according to any one of claims 2 to 5, wherein: The determining of the auxiliary information required for the second intelligent agent includes: Obtaining a target business corresponding to the second agent; The configuration information of the target service corresponding to the second agent is obtained, and the auxiliary information required by the second agent is obtained from the configuration information.

12. The method according to claim 1, wherein: The first agent determines one or more second agents from the agent set according to the input information, comprising: Obtaining user portraits and historical behavior data as auxiliary information for the first agent; The first agent performs intent recognition on the input information based on the auxiliary information to obtain an intent information set, wherein the intent information set includes at least one user intent; The one or more second agents are determined from the set of agents according to the user intention.

13. The method according to claim 1, wherein: Before the first agent determines one or more second agents from the agent set according to the input information, the method further includes: Determine the current target business scenario according to the input information; Obtaining an agent associated with the target business scenario; According to the agents associated with the target business scenario, a set of agents corresponding to the target business scenario is determined.

14. The method according to claim 1, wherein: Before the first agent determines one or more second agents from the agent set according to the input information, the method further includes: Obtain the industry to which the intelligent service belongs and the industry of the user corresponding to the input information; According to the industry to which the intelligent service belongs and the industry of the user corresponding to the input information, a set of intelligent agents is obtained.

15. According to the method described in any one of claims 1-5 and 12-14, the agent set includes at least one of the following agents: a speech generation agent, a retrieval enhancement generation agent, an active dialogue agent, a dialogue image agent, a dialogue video agent, an image recognition agent and an information search agent.

16. A multi-agent based information processing device, wherein: The device comprises: A receiving module, configured to receive an information processing request, wherein the information processing request includes input information; A first generating module, configured to input the input information to a first agent, and the first agent determines one or more second agents from an agent set according to the input information, and obtains output information of the first agent; The second generating module is used to obtain response information corresponding to the input information according to the output information of the first intelligent agent and the second intelligent agent.

17. The device according to claim 16, wherein: The second generating module is further used for: Determining auxiliary information required by the second agent, the auxiliary information comprising at least one of a pre-configured dialogue strategy, a user profile, and a historical dialogue; generating a prompt word for the second agent according to the auxiliary information and the output information of the first agent; Acquire response information corresponding to the input information according to the prompt word of the second agent.

18. The device according to claim 16, wherein: The second generating module is further used for: Determining, according to the output information of the first agent, the execution order of the plurality of second agents and the dependency relationship between the second agents; Invoke the plurality of second agents in the execution order, and determine input information of the second agent currently being invoked according to the dependency relationship; Generate a prompt word for the currently called second agent according to the auxiliary information of the second agent and the corresponding input information; The second output information corresponding to the second agent currently called is obtained according to the prompt word of the second agent, until the sequential execution of multiple second agents is completed, and the response information corresponding to the input information is obtained.

19. The device according to claim 18, wherein: The second generating module is further used for: According to the dependency relationship, determining an upstream agent that the second agent currently called depends on, wherein the upstream agent that is dependent on is at least one of the plurality of second agents; The output information of the dependent upstream agent is obtained and determined as the input information of the second agent currently called.

20. The device according to claim 16, wherein: The second generating module is further used for: For each second agent, determining the auxiliary information required by the second agent; Determining associated output information matching the second agent from the output information; generating a prompt word for each second agent according to the auxiliary information of the second agent and the matched associated output information; Acquire first output information corresponding to the associated output information according to the prompt word of the second agent; According to the first output information of each of the second agents, response information corresponding to the input information is generated.

21. The device according to any one of claims 17 to 20, wherein: The device also includes: An acquisition module, used to start at least one third agent for dialogue monitoring and acquire dialogue content during the dialogue process; An analysis module is used to perform intelligent analysis on the conversation content through the third agent.

22. The device according to claim 21, wherein The analysis module is further used for: Extracting information from the conversation content by the third agent to obtain key information, wherein the key information is used to optimize subsequent conversations; or The third agent clarifies and reflects on the content of the conversation to obtain a conversation reflection result.

23. The device according to claim 22, wherein: The analysis module is further used for: Determine the key information and the dialogue reflection result as the analysis result of the dialogue analysis agent; Performing structured processing on the analysis result of the dialogue analysis agent to obtain first structured data; The first structured data is determined to be shared data and stored in a data center, and the first structured data can be called by at least one of the first agent and the second agent.

24. The device according to any one of claims 17 to 20, wherein: The analysis module is further used for: The third agent performs hallucination monitoring and compliance monitoring on the conversation content to obtain quality detection results.

25. The device according to claim 24, wherein: The analysis module is further used for: Based on the quality detection results, abnormal conversation content is determined, and the abnormal conversation content is collected as sample data for fine-tuning the intelligent agent, and the sample data is used to perform fine-tuning training on the relevant intelligent agent.

26. The device according to any one of claims 17 to 20, wherein: The second generating module is further used for: Obtaining a target business corresponding to the second agent; The configuration information of the target service corresponding to the second agent is obtained, and the auxiliary information required by the second agent is obtained from the configuration information.

27. The device according to claim 16, wherein: The first generating module is further used for: Obtaining user portraits and historical behavior data as auxiliary information for the first agent; The first agent performs intent recognition on the input information based on the auxiliary information to obtain an intent information set, wherein the intent information set includes at least one user intent; The one or more second agents are determined from the set of agents according to the user intention.

28. The device according to claim 16, wherein: The first generating module is further used for: Determine the current target business scenario according to the input information; Obtaining an agent associated with the target business scenario; According to the agents associated with the target business scenario, a set of agents corresponding to the target business scenario is determined.

29. The device according to claim 16, wherein: The first generating module is further used for: Obtain the industry to which the intelligent service belongs and the industry of the user corresponding to the input information; According to the industry to which the intelligent service belongs and the industry of the user corresponding to the input information, a set of intelligent agents is obtained.

30. According to the device described in any one of claims 16-20 and 27-29, the agent set includes at least one of the following agents: a speech generation agent, a retrieval enhancement generation agent, an active dialogue agent, a dialogue image agent, a dialogue video agent, an image recognition agent and an information search agent.

31. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 15.

32. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-15.

33. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the method according to any one of claims 1 to 15 is implemented.

Images