Interaction method and device and computer readable storage medium
By scheduling multiple agents to cooperate, the problem that a single agent is difficult to meet the needs of complex users is solved, the performance and adaptability of AI assistants are improved, the cost is reduced, and more accurate and natural answers are achieved.
Patent Information
- Application Number
- CN202411440940.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-15
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-10-15
AI Technical Summary
A single agent is difficult to meet the needs of complex and diverse users, and its performance bottlenecks and high costs are limited, which limits the development and application of AI assistants.
By responding to user input information, determine user interaction intentions, split complex tasks, schedule combinations of multiple agents to perform different types of tasks, and use different types of agents to collaborate to improve performance and adaptability.
It can handle complex and variable user needs, improve the performance and adaptability of AI assistants, reduce overall costs, and ensure the accuracy and naturalness of answers.
Smart Images

Figure CN118964688A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of artificial intelligence, and in particular to an interaction method and device, and a computer-readable storage medium. Background Art
[0002] Artificial Intelligence (AI agent) refers to an intelligent entity that can perceive the environment, make decisions, perform actions, and interact with the environment. Artificial Intelligence is an important concept in the field of artificial intelligence and plays a key role in achieving automated decision-making and task execution.
[0003] With the continuous development and popularization of artificial intelligence technology, artificial intelligence entities will play an increasingly important role in people's lives and work. Summary of the invention
[0004] According to the first aspect of the present disclosure, an interaction method is provided, comprising: in response to receiving user input information, determining the user's interaction intention; determining a plurality of tasks and their execution order according to the interaction intention; for each of the plurality of tasks, determining an agent corresponding to the each task to obtain a combination of a plurality of agents, wherein different types of agents perform different types of tasks; and scheduling the combination of the plurality of agents to perform the plurality of tasks according to the execution order.
[0005] According to a second aspect of the present disclosure, an interaction device is provided, comprising: an intention determination module, configured to determine a user's interaction intention in response to receiving input information from the user; a task determination module, configured to determine a plurality of tasks and an execution order thereof according to the interaction intention; an agent determination module, configured to determine an agent for each of the plurality of tasks, and obtain a combination of a plurality of agents, wherein different types of agents perform different types of tasks; and a scheduling module, configured to schedule the combination of the plurality of agents to perform the plurality of tasks according to the execution order.
[0006] According to a third aspect of the present disclosure, an interaction device is provided, comprising: a memory; and a processor coupled to the memory, wherein the processor is configured to execute an interaction method according to any of some embodiments of the present disclosure based on instructions stored in the memory.
[0007] According to a fourth aspect of the present disclosure, there is provided an interactive system, comprising: an interactive device according to any of the embodiments of the present disclosure; and a plurality of intelligent agents.
[0008] According to a fifth aspect of the present disclosure, a computer-readable storage medium is provided, on which computer program instructions are stored. When the instructions are executed by a processor, the interaction method described in any embodiment of the present disclosure is implemented.
[0009] According to a sixth aspect of the present disclosure, a computer program product is provided, comprising computer program instructions, which, when executed by a processor, implement the interaction method according to any of the embodiments of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS
[0010] The preferred embodiments of the present disclosure are described below with reference to the accompanying drawings. The drawings described herein are used to provide a further understanding of the present disclosure. Each of the drawings together with the following detailed description is included in this specification and forms a part of the specification, and is used to explain the present disclosure. It should be understood that the drawings described below only relate to some embodiments of the present disclosure and do not constitute a limitation on the present disclosure. In the drawings: Figure 1 A schematic diagram of a flow chart of an interaction method according to some embodiments of the present disclosure is shown; Figure 2 A schematic diagram of a process for determining an intelligent agent according to some embodiments of the present disclosure is shown; Figure 3 A schematic diagram showing an interaction method according to some embodiments of the present disclosure is shown; Figure 4 A block diagram of an interaction device according to some embodiments of the present disclosure is shown; Figure 5 A block diagram of an interaction device according to some other embodiments of the present disclosure is shown; Figure 6 A block diagram of an electronic device according to some embodiments of the present disclosure is shown.
[0011] It should be understood that, for ease of description, the sizes of the various parts shown in the drawings are not necessarily drawn according to the actual proportional relationship. The same or similar reference numerals are used in the various drawings to represent the same or similar parts. Therefore, once an item is defined in one drawing, it may not be further discussed in subsequent drawings. DETAILED DESCRIPTION
[0012] The following will be combined with the drawings in the embodiments of the present disclosure to clearly and completely describe the technical solutions in the embodiments of the present disclosure, but it is obvious that the described embodiments are only part of the embodiments of the present disclosure, rather than all the embodiments. The following description of the embodiments is actually only illustrative and is by no means intended to limit the present disclosure and its application or use. It should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments set forth herein.
[0013] It should be understood that the various steps recorded in the method embodiments of the present disclosure can be performed in different orders, and / or performed in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown in the execution. The scope of the present disclosure is not limited in this respect. Unless otherwise specifically stated, the relative arrangement, numerical expressions and numerical values of the parts and steps set forth in these embodiments should be interpreted as being merely exemplary and do not limit the scope of the present disclosure.
[0014] The term "include" and its variations used in the present disclosure mean an open term that includes at least the following elements / features but does not exclude other elements / features, that is, "including but not limited to". In addition, the term "include" and its variations used in the present disclosure mean an open term that includes at least the following elements / features but does not exclude other elements / features, that is, "including but not limited to". Therefore, include is synonymous with include. The term "based on" means "based at least in part on".
[0015] References throughout this specification to "one embodiment," "some embodiments," or "an embodiment" mean that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the present invention. For example, the term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; and the term "some embodiments" means "at least some embodiments." Furthermore, the appearances of the phrases "in one embodiment," "in some embodiments," or "in an embodiment" in various places throughout this specification do not necessarily all refer to the same embodiment, but may do so.
[0016] It should be noted that the concepts of "first", "second", etc. mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units. Unless otherwise specified, the concepts of "first", "second", etc. are not intended to imply that the objects described in this way must be in a given order in time, space, ranking, or any other manner.
[0017] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".
[0018] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.
[0019] The embodiments of the present disclosure are described in detail below in conjunction with the accompanying drawings, but the present disclosure is not limited to these specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. In addition, in one or more embodiments, specific features, structures or characteristics can be combined in any suitable manner that will be clear from the present disclosure by a person of ordinary skill in the art.
[0020] In the field of AI assistants, as user needs become increasingly complex and diverse, the limitations of a single model are becoming increasingly prominent. These limitations are mainly manifested in: single functions, which make it difficult to simultaneously meet the diverse needs of users in different scenarios; performance bottlenecks, where a single agent may perform poorly when handling certain specific tasks; and high costs, where larger and more expensive models are often required to improve overall performance. The above shortcomings limit the further development and application of AI assistants.
[0021] An embodiment of the present disclosure provides an interaction method, comprising: determining the user's interaction intention in response to receiving the user's input information; determining a plurality of tasks and their execution order according to the interaction intention; for each of the plurality of tasks, determining an agent corresponding to the each task to obtain a combination of a plurality of agents, wherein different types of agents execute different types of tasks; and scheduling the combination of the plurality of agents to execute the plurality of tasks according to the execution order.
[0022] Compared with a single model, the combination of multiple agents and scheduling different types of agents to perform different tasks enables the combination of agents to handle more complex and changing user needs, improving the performance and adaptability of AI assistants. The collaboration of multiple specialized agents can also make the answers more accurate and natural.
[0023] Figure 1 A flowchart of an interaction method according to some embodiments of the present disclosure is shown.
[0024] like Figure 1 As shown, the interaction method includes: step S1, in response to receiving the user's input information, determining the user's interaction intention; step S2, determining multiple tasks and their execution order according to the interaction intention; step S3, for each of the multiple tasks, determining the intelligent agent corresponding to each task, and obtaining a combination of multiple intelligent agents, wherein different types of intelligent agents perform different types of tasks; step S4, scheduling the combination of multiple intelligent agents to perform multiple tasks according to the execution order.
[0025] The user's input information may be, for example, text, URL, picture, video, etc.
[0026] User interaction intention refers to the potential purpose or desired goal of user interaction behavior.
[0027] For example, multiple tasks include multiple steps that need to be performed to satisfy the user's interaction intention. Multiple tasks can be performed in parallel or in series. Alternatively, some tasks can be performed in parallel and some tasks can be performed in series.
[0028] An agent is, for example, an agent that performs a specific task, also called a dedicated agent. Each type of agent has a specific expertise, thereby refining the business division of the agent. During the scheduling of a combination of multiple agents to perform multiple tasks, each agent performs the task it is good at, and the combination of multiple agents cooperates with each other.
[0029] The interactive method of this embodiment can be performed by an interactive device. The interactive device is, for example, a scheduler, and the above steps S1-S4 can be performed by the scheduler. By introducing the "scheduler" role, intelligent task allocation can be achieved.
[0030] The interactive method of this embodiment may be executed on the client side, or may be partially executed on the server side.
[0031] Combine the following Figure 2 , first introduce the process of determining an intelligent agent according to some embodiments of the present disclosure.
[0032] like Figure 2 As shown, step S3 determines the intelligent agent corresponding to each task for each of the multiple tasks, including: step S31, determining the type and requirements of each task; step S32, determining one or more candidate intelligent agents matching the type of each task; step S33, selecting the intelligent agent corresponding to each task from one or more candidate intelligent agents according to the requirements of each task.
[0033] For example, if the task type is chatting, the corresponding agent type is a chat agent (e.g., a role-playing agent). One or more candidate agents for a task may be agents of the same type but with different tones and styles. For example, different chat agents have different tones and styles, such as friendly, humorous, calm and rational, literary and emotional, etc.
[0034] The requirements of the task can be determined based on the user's intention. For example, if the user wants to be comforted, the task requirement is to be warm and friendly when chatting, and then an agent with a friendly style is determined from the chat-type agents to perform the chat task.
[0035] In some embodiments, in response to receiving input information from the user, determining the interaction intent of the user includes: in response to receiving input information from the user, determining the type of the input information; and determining the interaction intent based on the type of the input information.
[0036] The type of input information refers to the carrier of the input information, such as text, URL, picture, video, etc. Different types of input information may have different interaction intentions.
[0037] If it is detected that the user's input contains an image, it is judged that the user's intention may be image processing, or image-generated text, etc. If it is detected that the user's input is in text form, natural language processing technology can be used to analyze what the user's intention is.
[0038] In some embodiments, in response to receiving user input information, determining the user's interaction intention includes: in response to determining that the input information does not meet a second specified condition, outputting a second guidance question to the user; and determining the interaction intention based on the user's feedback on the second guidance question.
[0039] For example, if the user's question is not clear enough and it is difficult to determine the user's interaction intention, the input information does not meet the second specified condition. To this end, it is necessary to output a second guiding question to the user to guide the user to clarify his intention.
[0040] In some embodiments, multiple tasks and their execution order are determined based on the interaction intention, including: determining a complex task corresponding to the interaction intention; splitting the complex task into multiple tasks; and determining the execution order based on the relationship between the multiple tasks.
[0041] For example, the scheduler determines how many intentions are in a sentence input by the user, what tasks need to be performed to achieve these intentions, and determines the execution order of these tasks. If the user wants to create an article that meets the user's needs, he needs to search the Internet first, and then write based on the searched content. Therefore, the execution order of tasks is to search first, then create.
[0042] The execution order of tasks can be determined first, and then the agent corresponding to each task can be determined based on the tasks and their execution order. Alternatively, the agent corresponding to each task can be determined first, and then the execution order of tasks can be determined based on the characteristics of the agent, user intention, task requirements, etc. The relationship between multiple tasks is, for example, an input-output relationship. For example, if the processing result of task A is required to execute task B, task A should be executed first, and then task B.
[0043] By splitting up complex tasks and refining the business division of labor and collaboration of intelligent agents, efficient collaboration can be achieved. At the same time, by fine-tuning the task allocation of complex tasks, a basic model can solve a basic task split from a complex task, reducing the excessive use of models with large parameters and high costs.
[0044] In some embodiments, the interaction method further comprises: providing task information to at least one agent of the plurality of agents based on the input information.
[0045] For example, the scheduler assigns tasks to selected agents, processes user input information, splits it into context information required to execute the task, and provides it to the corresponding agent.
[0046] The scheduler can provide task information to all agents or only to some of them. For example, if the tasks are executed in series and the next agent only depends on the output of the previous agent to execute the task, then only the task information needs to be provided to the first agent in the series. If the first several tasks are executed in parallel, then the task information is provided to all the first several agents.
[0047] In some embodiments, the interaction method further comprises: determining a base model for at least one agent of the plurality of agents based on the input information.
[0048] The base model and the agent do not need to be bound together, and the base model of the agent can be flexibly adjusted according to the task requirements of different scenarios.
[0049] For example, the base model includes models that are good at tool calling, models that are good at language communication, models that are good at text processing, and models that are good at mathematical reasoning. For tasks that require interaction with external systems, choose a model that is good at tool calling; for tasks that require natural and fluent conversations, choose a model that is good at language communication; for tasks that process and generate large amounts of text, choose a model that is good at text processing; for scientific computing and logical analysis tasks, choose a model that is good at mathematical reasoning.
[0050] If the user input is in text form, the base can also be selected based on the token of the text. For example, if the text input by the user is long, a base model with a long context window is selected for the agent.
[0051] The multiple agents include at least one of the following: a role-playing agent, an information retrieval agent, a creative agent, and a question-answering agent. The language styles of different creative agents are different. The characteristics of information retrieval agents are, for example, rigorous and professional, good at finding and organizing information, and capable of network search, database query, and information aggregation. The base model of information retrieval agents is, for example, a model with a wide range of knowledge.
[0052] The characteristics of the creative agent are, for example, being imaginative, good at content creation, and being able to write, create stories, create poetry, etc. The base model of the creative agent is, for example, a model with strong creativity.
[0053] The question-answering agent is also called a problem-solving agent, and its characteristics include strong logic, good at analyzing and solving complex problems, and the ability to perform mathematical calculations, logical reasoning, and solution design. The base model of the question-answering agent, for example, adopts a model with strong reasoning ability.
[0054] The multiple agents may also include a small talk agent. The small talk agent may be good at daily conversations, and may be able to handle daily greetings, small talk, and simple questions and answers. The base model of the small talk agent may be a model that is good at natural conversations.
[0055] At the same time, multiple agents can also include translation agents, text expansion agents, image expansion agents, text-image agents, etc.
[0056] In some embodiments, the interaction method also includes: when the output of multiple agents does not satisfy the first specified condition, redetermining the agent for each task to obtain a redetermined combination of multiple agents; scheduling the redetermined combination of multiple agents according to the execution order to re-execute multiple tasks.
[0057] The first specified condition is, for example, that the outputs of multiple agents are consistent, or that the final result generated based on the outputs of multiple agents satisfies the user's interaction intention.
[0058] If the first specified condition is not met, the scheduler can reselect the agent, change the type of the selected agent, or select a different agent of the same type. For example, if the outputs of agent A performing task 1 and agent B performing task 2 are inconsistent, then agent C of the same type but different style as agent B can be used to perform task 2, or agent D of a different type from agent B can be used to perform task 2.
[0059] That is to say, at runtime, the scheduler can dynamically adjust the combination and working mode of the agents, and flexibly configure multiple agents to work together according to different business scenarios. The working mode includes, for example, sensing the environment, making planning decisions, and executing actions. At the same time, it can also iteratively optimize and continuously adjust the collaboration mode of the agents based on user feedback.
[0060] An additional detector may be added to detect whether the tasks, task sequence, and selected intelligent agents determined by the scheduler are reasonable. If not, the scheduler will rearrange the tasks, task sequence, and selected intelligent agents.
[0061] In some embodiments, the interaction method also includes: when the outputs of multiple agents do not meet the first specified condition, redetermining multiple tasks and their execution order, determining the agent corresponding to each task, and scheduling a combination of multiple agents to perform multiple tasks.
[0062] For example, if the outputs of multiple agents do not meet the first specified condition, it may not only be that the combination of agents is unreasonable, but also that the task arrangement is unreasonable. Therefore, multiple tasks and their execution order can be re-determined and the agents can be reallocated.
[0063] In some embodiments, the interaction method also includes: when the output of multiple agents does not meet the first specified condition, outputting a guiding question to the user; redetermining the user's interaction intention based on the user's feedback on the guiding question; redetermining multiple tasks and their execution order, determining the agent corresponding to each task, and scheduling a combination of multiple agents to perform multiple tasks.
[0064] For example, if the outputs of multiple agents do not meet the first specified condition, it may also be that the user's intention is not accurately determined. The user can be asked to clarify his intention through a counter-question, and the tasks can be rearranged and assigned.
[0065] In some embodiments, the interaction method further includes: in the event that there is a conflict in the outputs of multiple agents, redetermining multiple tasks and their execution order, determining the agent corresponding to each task, and scheduling a combination of multiple agents to perform multiple tasks.
[0066] For example, if the outputs of multiple agents conflict, it may be because the combination of agents is not appropriate or the task arrangement is unreasonable. Therefore, the conflict can be resolved by rearranging and assigning tasks. The agent responsible for the final integration can determine whether the outputs of all agents conflict, or other detection modules can determine whether the outputs of all agents conflict.
[0067] If the outputs of multiple agents conflict, the combination of agents can be changed first. If the conflict is resolved, the result can be generated and output to the user. If the conflict is not resolved and the number of times the agents are recombined reaches a threshold, it is possible to start by re-determining multiple tasks and select a combination of agents for the re-determined tasks.
[0068] In some embodiments, the interaction method also includes: when there is a conflict in the outputs of multiple agents, outputting a first guiding question to the user; redetermining the user's interaction intention based on the user's feedback on the first guiding question; redetermining multiple tasks and their execution order, determining the agent corresponding to each task, and scheduling a combination of multiple agents to perform multiple tasks.
[0069] For example, if the outputs of multiple agents conflict, the user can be asked to decide which part he needs.
[0070] After obtaining user feedback, the part of the conflicting content that the user needs can be output to the user. After obtaining user feedback, the user's interaction intention can also be re-determined based on the user's feedback, and the task can be re-determined and assigned. In addition, when assigning tasks to agents, the agent that produces the part of the content that the user does not need can be excluded.
[0071] If the outputs of multiple agents conflict, the combination of agents can be changed first. If the conflict is resolved, the result can be generated and output to the user. If the conflict is not resolved and the number of times the agents are recombined reaches a threshold, it can start by re-determining multiple tasks and selecting a combination of agents for the re-determined tasks. If the number of times the tasks are re-determined reaches a threshold and the conflict is still not resolved, the first guiding question is output to the user to seek user help.
[0072] After multiple agents output their results, the outputs of the multiple agents can be integrated to obtain an integrated result, and the integrated result is output. The following is an introduction to the integration process.
[0073] For example, the first target agent is determined by the scheduler, and the scheduler schedules the first target agent to collect and integrate the outputs of multiple agents. The first target agent can be one of the multiple agents performing the task, or an agent specifically used for integration.
[0074] The scheduler can also determine a second target agent and schedule the second target agent to output the integrated result. The second target agent can be one of the aforementioned multiple agents or an agent specifically used for output. The second target agent can be the same agent as the first target agent or a different agent.
[0075] The scheduler can determine the first target agent and the second target agent based on the tasks and their execution order. For example, the last agent that executes the task can be designated as the first target agent and the second target agent, which are responsible for integration and output.
[0076] The integrated results may also be subjected to a consistency check, and when the consistency check passes, the results that pass the consistency check are output.
[0077] The consistency check can be used to check whether the result of the integration is consistent in logic and tone. For example, if the result of the integration is an article, it can be checked whether there are contradictions between the previous and next paragraphs of the article, or whether the writing style is different.
[0078] The scheduler schedules the third target agent to perform a consistency check on the integrated result, and if the consistency check passes, schedules the second target agent to output the result that passes the consistency check.
[0079] If the consistency check fails, the outputs of multiple agents are reintegrated and the consistency check is performed on the reintegrated results.
[0080] For example, if the consistency check fails, there may be a problem in the integration process. The outputs of multiple intelligent agents can be reintegrated, and only consistent content can be selected for integration.
[0081] If the consistency check fails, you can first reintegrate. If the consistency problem can be solved after reintegration, the integrated result will be output to the user. If the number of reintegrations reaches a threshold and the consistency problem is not solved, you can change the combination of agents. If the re-combination of agents can solve the consistency problem, you can generate a result and output it to the user. If the consistency problem is still not solved and the number of agent re-combinations reaches a threshold, you can start by re-determining multiple tasks and select a combination of agents for the re-determined tasks. If the number of re-determined tasks reaches a threshold and the consistency problem is still not solved, you can output guidance questions to the user to seek user help.
[0082] Multiple agents can exchange information through a shared knowledge base. For example, multiple agents can share some common knowledge through a shared knowledge base. By centrally storing common knowledge and providing it to multiple agents, it is beneficial to update and maintain common knowledge in real time.
[0083] According to some embodiments of the present disclosure, by determining the user's interaction intention, multiple tasks and their execution order are determined, and a corresponding intelligent agent is determined for each task, a combination of multiple intelligent agents is obtained, and the combination of multiple intelligent agents is scheduled to perform multiple tasks, thereby satisfying the user's interaction intention.
[0084] Compared with a single model, the combination of multiple agents and the way different types of agents perform different tasks enable the combination of agents to handle more complex and changing user needs, improving the performance and adaptability of AI assistants. The collaboration of multiple specialized agents can also make the answers more accurate and natural.
[0085] Multi-agent collaboration can give full play to the advantages of different models and allow sub-agents to be optimized, improving the performance of a certain agent in a specific task, thereby improving overall performance and reducing costs.
[0086] Figure 3 A schematic diagram of an interaction method according to some embodiments of the present disclosure is shown.
[0087] like Figure 3As shown, when the user inputs prompt information, the system can use a preset persona to prompt the user and assist the user in inputting information. At the same time, the user's identity can also be verified to improve security.
[0088] Determine whether the carrier input by the user is a long text, image, video, or link. If it is a long text, select a model that is suitable for its token length. If it is an image, video, or link, you can also determine the user's intention based on the carrier type and select the corresponding model.
[0089] If the carrier is not of the above type but a text of normal length, then determine whether the user's meaning is clear. If not, use rhetorical questions to help the user clarify his intention. If the user's intention can be clearly judged, determine the tasks that need to be performed and their corresponding agents.
[0090] Figure 4 A block diagram of an interaction device according to some embodiments of the present disclosure is shown.
[0091] like Figure 4 As shown, the interaction device 4 includes: an intention determination module 41, configured to determine the user's interaction intention in response to receiving the user's input information; a task determination module 42, configured to determine multiple tasks and their execution order according to the interaction intention; an agent determination module 43, configured to determine the agent of each task for each of the multiple tasks, and obtain a combination of multiple agents, wherein different types of agents perform different types of tasks; a scheduling module 44, configured to schedule a combination of multiple agents to perform multiple tasks according to the execution order.
[0092] The intention determination module 41 of the interaction device 4 can be used to perform Figure 1 Step S1. The task determination module 42 of the interactive device 4 can be used to perform Figure 1 The agent determination module 43 can be used to perform Figure 1 The scheduling module 44 can be used to perform Figure 1 Step S4.
[0093] In some embodiments, the interaction device 4 further includes: a providing module configured to provide task information to at least one of the multiple agents based on the input information.
[0094] In some embodiments, the interaction device 4 further includes: a base determination module configured to determine a base model for at least one agent of the plurality of agents based on input information.
[0095] In some embodiments, the interactive device 4 also includes: a loop module, which is configured to redetermine the agent for each task when the output of multiple agents does not meet the first specified condition, and obtain a redetermined combination of multiple agents; and schedule the redetermined combination of multiple agents according to the execution order to re-execute multiple tasks.
[0096] In some embodiments, the interactive device 4 also includes: a loop module, which is configured to redetermine multiple tasks and their execution order, determine the agent corresponding to each task, and schedule a combination of multiple agents to perform multiple tasks when there is a conflict in the outputs of multiple agents.
[0097] In some embodiments, the interaction device 4 also includes: a loop module, which is configured to output a first guiding question to the user when there is a conflict in the outputs of multiple intelligent agents; redetermine the user's interaction intention based on the user's feedback on the first guiding question; redetermine multiple tasks and their execution order, determine the intelligent agent corresponding to each task, and schedule a combination of multiple intelligent agents to perform multiple tasks.
[0098] In some embodiments, the interaction device 4 further includes: an integration module configured to integrate the outputs of multiple agents to obtain an integrated result; and output the integrated result.
[0099] In some embodiments, the integration module is further configured to perform a consistency check on the integration result; if the consistency check passes, output the result that passes the consistency check.
[0100] In some embodiments, the integration module is further configured to reintegrate the outputs of the multiple agents if the consistency check fails, and perform a consistency check on the reintegrated results.
[0101] Figure 5 A block diagram of an interaction device according to some other embodiments of the present disclosure is shown.
[0102] like Figure 5 As shown, the interaction device 5 includes: a memory 51 ; and a processor 52 coupled to the memory 51 , and the processor 52 is configured to execute the interaction method of any of the aforementioned embodiments based on the instructions stored in the memory 51 .
[0103] The memory 51 is used to store one or more computer-readable instructions. The memory 51 may include any combination of various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory, including but not limited to random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), read-only memory (ROM), flash memory. The memory 51 may store, for example, an operating system, an application, a boot loader (BootLoader), a database, and other programs, and may also store various applications and various data.
[0104] The processor 52 is used to run computer-readable instructions to implement the interaction method of any of the above embodiments. The specific implementation of each step of the interaction method can be referred to the above embodiments, and the repeated parts will not be repeated here.
[0105] The processor 52 may be configured to execute Figure 1 The processor 52 may be embodied as various processing devices, such as a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. The central processing unit (CPU) may be an X86 or ARM architecture, etc.
[0106] The processor 52 and the memory 51 may communicate with each other directly or indirectly. For example, the processor 52 and the memory 51 may communicate with each other via a network. The network may include a wireless network, a wired network, and / or any combination of a wireless network and a wired network. The processor 52 and the memory 51 may also communicate with each other via a system bus, which is not limited in the present disclosure.
[0107] It should be noted that Figure 5 The components of the interactive device 5 shown are only exemplary and non-restrictive. The interactive device 5 may also have other components according to actual application requirements. The processor 52 may control other components in the interactive device 5 to perform desired functions.
[0108] The interactive device may be implemented by software, firmware and / or hardware, and may be integrated into an electronic device installed with relevant application programs.
[0109] The present disclosure provides an interactive system, comprising: an interactive device according to any of the embodiments of the present disclosure; and multiple agents. The interactive system according to the present disclosure implements multi-agent collaboration and provides a flexible and efficient multi-agent collaboration framework.
[0110] In some embodiments, the interactive system further includes a user interface configured to: receive user input; and display output of the interactive device.
[0111] In some embodiments, the interactive system further includes a knowledge base configured to store knowledge and information shared by the agents.
[0112] In some embodiments, the interactive system further includes an external API interface configured to connect to external services and data sources.
[0113] According to some embodiments of the present disclosure, a multi-agent AI assistant system can be constructed, and the multi-agent AI assistant system includes a scheduler and multiple agents for selection to provide services to users.
[0114] The system framework of the multi-agent AI assistant can adopt an extensible architecture, such as a plug-in system, to support the flexible addition of new agents and base models.
[0115] Each agent can be used as an independent module, which is easy to add, remove or update. At the same time, standardized interfaces can be used to define a unified data exchange format to facilitate the integration of new agents and models.
[0116] Developers can create their own agents. In addition, developers can add their own agents to the AI assistant system by talking to the AI assistant. For example, if a developer creates an agent for searching news and adds it to the system, when the user talks to the AI assistant, the user's own agent can be called to process the user's intention related to news. By adopting an extensible architecture, the scalability of the system can be enhanced, new models and functions can be easily integrated, and it can adapt to future technological developments.
[0117] Figure 6 A block diagram of an electronic device according to some embodiments of the present disclosure is shown.
[0118] Figure 6 The electronic device 6 shown may be a computer system with a dedicated hardware structure, which can execute corresponding functions when a relevant application program is installed.
[0119] Electronic devices include, but are not limited to, mobile terminals such as smart phones, laptops, personal digital assistants (PDA), tablet personal computers (Tablet PC), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), wearable devices, etc., and fixed terminals such as digital televisions, desktop computers, etc.
[0120] like Figure 6 As shown, the central processing unit (CPU) 61 performs various processes according to the program stored in the read-only memory (ROM) 62 or the program loaded from the storage part 68 to the random access memory (RAM) 63. In the RAM 63, data required when the CPU 61 performs various processes, etc. is stored as needed. The central processing unit is only exemplary, and it can also be other types of processors, such as the various processors mentioned above. The ROM 62, RAM 63 and the storage part 68 can be various forms of computer-readable storage media. It should be noted that although Figure 6 ROM 62, RAM 63 and storage section 68 are shown separately in FIG. 1 , but one or more of them may be combined or located in the same or different memory or storage modules.
[0121] The CPU 61, the ROM 62, and the RAM 63 are connected to one another via a bus 64. To the bus 64, an input / output interface 65 is also connected.
[0122] The following components are connected to the input / output interface 65: an input portion 66, such as a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output portion 67, including a display, such as a cathode ray tube (CRT), a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage portion 68, including a hard disk, a magnetic tape, etc.; and a communication portion 69, including a network interface card such as a LAN card, a modem, etc. The communication portion 69 allows communication processing to be performed via a network such as the Internet. It is easy to understand that although Figure 6 The various devices or modules in the electronic device 6 are shown to communicate via a bus 64, but they may also communicate via a network or other means, wherein the network may include a wireless network, a wired network, and / or any combination of a wireless network and a wired network.
[0123] A drive 610 is also connected to the input / output interface 65 as needed. A removable medium 611 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc. is mounted on the drive 610 as needed so that a computer program read therefrom is installed into the storage section 68 as needed.
[0124] When the above-described series of processing is implemented by software, a program constituting the software can be installed from a network such as the Internet or a storage medium such as the removable medium 611 .
[0125] According to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, some embodiments of the present disclosure include a computer program product, which, when running on a computer, enables the computer to implement the interactive method of any of the aforementioned embodiments. The computer program product includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 69, or installed from the storage part 68, or installed from the ROM 62. When the computer program is executed by the CPU 61, the interactive method of the embodiment of the present disclosure is executed.
[0126] It should be noted that, in the context of the present disclosure, a computer-readable medium may be a tangible medium that may contain or store a program for use by an instruction execution system, apparatus, or device or for use in conjunction with an instruction execution system, apparatus, or device.
[0127] The computer readable medium may be a computer readable storage medium, or a computer readable signal medium, or any combination of the two.
[0128] Computer-readable storage media include, but are not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or components, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to, electrical connections with one or more wires, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROMs), optical storage devices, magnetic storage devices, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, device, or device. A computer program is stored on a computer-readable storage medium that, when executed by a processor, implements the interactive method of any of the foregoing embodiments.
[0129] Computer readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer readable program codes. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer readable signal media may also be any computer readable medium other than a computer readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, device, or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.
[0130] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.
[0131] In some embodiments, a computer program is further provided, comprising: instructions, which, when executed by a processor, cause the processor to execute the interaction method of any of the above embodiments. For example, the instructions may be embodied as computer program codes.
[0132] In embodiments of the present disclosure, computer program codes for performing the operations of the present disclosure may be written in one or more programming languages or combinations thereof, including but not limited to object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" language or similar programming languages. The program code may be executed entirely on a user's computer, partially on a user's computer, as an independent software package, partially on a user's computer, partially on a remote computer, or entirely on a remote computer or server. In situations involving a remote computer, the remote computer may be connected to the user's computer via any type of network (including a local area network (LAN) or a wide area network (WAN)), or may be connected to an external computer (e.g., using an Internet service provider to connect via the Internet).
[0133] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.
[0134] The functions described above may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.
[0135] Although some specific embodiments of the present disclosure have been described in detail by way of example, it should be understood by those skilled in the art that the above examples are for illustration only and are not intended to limit the scope of the present disclosure. It should be understood by those skilled in the art that the above embodiments may be modified without departing from the scope and spirit of the present disclosure. The scope of the present disclosure is defined by the appended claims.
Claims
1. An interactive method, comprising: In response to receiving input information from a user, determining an interaction intent of the user; Determine multiple tasks and their execution order according to the interaction intention; For each of the plurality of tasks, determining an agent corresponding to each of the tasks to obtain a combination of a plurality of agents, wherein different types of agents perform different types of tasks; According to the execution order, a combination of the plurality of agents is scheduled to execute the plurality of tasks.
2. The interactive method according to claim 1, wherein: For each of the plurality of tasks, determining an agent corresponding to each task includes: Determine the type and requirements of each of the tasks described; Determining one or more candidate agents that match the type of each of the tasks; According to the requirements of each task, an agent corresponding to each task is selected from the one or more candidate agents.
3. The interactive method according to claim 1, wherein: Determining a plurality of tasks and their execution order according to the interaction intention includes: Determine a complex task corresponding to the interaction intention; splitting the complex task into the multiple tasks; The execution order is determined according to the relationship between the multiple tasks.
4. The interactive method according to claim 1, further comprising: Based on the input information, task information is provided to at least one of the plurality of agents.
5. The interactive method according to claim 1, further comprising: Based on the input information, a base model is determined for at least one agent of the plurality of agents.
6. The interactive method according to claim 1, further comprising: In the case where the outputs of the plurality of agents do not satisfy the first specified condition, re-determining the agent for each task to obtain a re-determined combination of the plurality of agents; According to the execution order, the re-determined combination of the multiple agents is scheduled to re-execute the multiple tasks.
7. The interactive method according to claim 1, further comprising: In the event that there is a conflict in the outputs of the multiple agents, the multiple tasks and their execution order are re-determined, the agent corresponding to each task is determined, and the combination of the multiple agents is scheduled to execute the multiple tasks.
8. The interactive method according to claim 1, further comprising: When there is a conflict between the outputs of the plurality of agents, outputting a first guiding question to the user; Re-determining the user's interaction intention based on the user's feedback on the first guiding question; Re-determine a plurality of tasks and their execution order, determine an intelligent agent corresponding to each task, and schedule a combination of the plurality of intelligent agents to execute the plurality of tasks.
9. The interactive method according to claim 1, further comprising: Integrating the outputs of the multiple intelligent agents to obtain an integrated result; The result of the integration is output.
10. The interactive method according to claim 9, wherein: The outputting of the integrated result includes: Performing consistency check on the results of the integration; When the consistency check passes, a result of passing the consistency check is output.
11. The interactive method according to claim 10, further comprising: In the case where the consistency check fails, the outputs of the multiple agents are reintegrated, and a consistency check is performed on the reintegrated result.
12. The interactive method according to claim 1, wherein: The step of determining the user's interaction intention in response to receiving the user's input information includes: In response to determining that the input information does not satisfy a second specified condition, outputting a second guiding question to the user; The interaction intention is determined according to the user's feedback on the second guiding question.
13. The interactive method according to claim 1, wherein: The step of determining the user's interaction intention in response to receiving the user's input information includes: In response to receiving the input information, determining a type of the input information; The interaction intention is determined according to the type of the input information.
14. The interactive method according to claim 1, wherein: The plurality of agents include at least one of the following: role playing agents; Information retrieval agents; Creative agents, where different creative agents have different language styles; Question answering agent.
15. The interactive method according to any one of claims 1 to 14, wherein: The multiple agents exchange information via a shared knowledge base.
16. An interactive device, comprising: an intention determination module, configured to determine the user's interaction intention in response to receiving the user's input information; A task determination module, configured to determine a plurality of tasks and their execution order according to the interaction intention; An agent determination module is configured to determine an agent for each of the multiple tasks to obtain a combination of multiple agents, wherein different types of agents perform different types of tasks; The scheduling module is configured to schedule the combination of the multiple agents to execute the multiple tasks according to the execution order.
17. An interactive device, comprising: Memory; as well as A processor coupled to the memory, wherein the processor is configured to execute the interaction method according to any one of claims 1 to 15 based on instructions stored in the memory.
18. An interactive system comprising: An interactive device according to claim 16 or 17; as well as Multiple agents.
19. A computer-readable storage medium having computer program instructions stored thereon, which, when executed by a processor, implement the interactive method according to any one of claims 1 to 15.
20. A computer program product, comprising computer program instructions, which, when executed by a processor, implement the interaction method according to any one of claims 1 to 15.
Citation Information
Patent Citations
Ship electric-power system reconstruction method based on BDI (belief-desire-intention theory) multi-agent
CN103745120A
Intelligent agent action determination method and device, electronic equipment and medium
CN113807460A
Task allocation method for clustering multi-agent network
CN114723402A
Intelligent agent control method and device, storage medium, electronic equipment and server
CN115220885A
Multi-agent interaction system construction method and device, computer equipment and storage medium
CN115469981A
Cited By
Task processing method and device, equipment and storage medium
CN120610823A
Interaction method and device based on agent cooperation, agent and storage medium
CN120654731A
User demand response method and device based on intelligent agent
CN121563549A
Method and system for executing user request by AI agent
CN121967407A
Interaction method and apparatus, and computer-readable storage medium
WO2026081624A1
