An intelligent dialogue system for temporal memory management and task distribution

Abstract

The present invention discloses an intelligent dialogue system for temporal memory management and task distribution. Based on the multi-agent collaboration of CrewAI, it combines time-series memory updating and multi-agent collaborative work. In order to solve the problem of low memory management efficiency, an independent memory unit is introduced, and an undirected graph structure is used to store the learned information, and each piece of information is stored as a node in the graph; at the same time, memory is added, modified and forgotten according to the time sequence, which alleviates the problem of forgetting important information to a certain extent; in order to solve the problem of insufficient task interaction, a multi-agent collaborative mechanism is introduced, and user demand understanding, task execution and task verification are respectively handed over to three different intelligent agents to work independently. Through the independent work and complete information exchange between the three intelligent agents, information synchronization and status sharing between modules are achieved, information islands are reduced, and decision-making accuracy is improved.

Description

Technical Field

[0001] The present invention relates to a multi-agent collaborative dialogue system in the field of artificial intelligence, and in particular to an intelligent dialogue system for temporal memory management and task distribution based on the multi-agent framework CrewAI. Background Art

[0002] Multi-agent dialogue is one of the key solutions for complex task processing and long-term conversations in the field of artificial intelligence. This involves assigning knowledge extraction and learning, task decomposition, task execution, and task verification to different agents independently, with independent memory management systems responsible for memory, forgetting, and updating. However, current multi-agent dialogue systems suffer from the following issues:

[0003] 1. Inefficient memory management: Inefficient memory management is a common problem in AI dialogue systems. Existing dialogue system memory management requires not only storing conversation content but also maintaining multi-dimensional information such as contextual relationships, user status, and task progress. Compared to traditional information storage systems, this consumes significant computing resources and storage space. Furthermore, as the number of conversation rounds increases, the system must balance information integrity and processing efficiency within limited computing resources. Simple sliding window mechanisms can lead to the loss of important historical information, while full storage results in inefficient retrieval and wasteful resources. Furthermore, as the conversation scenario changes, memory management strategies must be adjusted accordingly, further increasing the difficulty of system maintenance. Consequently, multi-agent dialogues face challenges such as heavy storage burdens and a difficult trade-off between resources and efficiency.

[0004] 2. Insufficient Task Interaction: Existing dialogue systems often separate functions such as understanding user needs, memory management, and task execution. While this modular design facilitates management and maintenance in engineering implementation, it also leads to serious functional fragmentation. Modules are connected only through parameter sharing or simple message passing. This shallow information exchange cannot meet the requirements of complex dialogue scenarios. The challenges faced by systems in multi-task collaborative processing far exceed those faced by single-task processing. First, complex dependencies between tasks must be accurately identified and handled. For example, the output of one task may be the necessary input for another. Second, when multiple tasks run simultaneously, resource competition arises. The system must rationally allocate limited resources such as computing resources and memory space, which increases the complexity of task scheduling. When handling complex multi-turn dialogues, the information silo effect between task modules becomes increasingly apparent. For example, the natural language understanding module may understand some of the user's implicit intent, but due to the lack of an effective information sharing mechanism, this important information cannot be promptly transmitted to the task planning or execution modules. Similarly, the experience and contextual information accumulated by the task execution module during processing is difficult to feed back to the needs understanding module, resulting in the inability of the system to perform global optimization and adjustments. This information silo effect ultimately leads to biased system decisions.

[0005] Current status of the problem and solution: Updating the agent's memory in a temporal manner is currently a relatively mature method for solving the problem of forgetting in long-term conversations. The problem of insufficient agent-task interaction has also been addressed by relevant multi-agent collaborative work models, but the industry has not yet found a case that combines temporal memory updating and multi-agent collaborative work. Summary of the Invention

[0006] In response to the problems existing in the above-mentioned prior art, the purpose of the present invention is to combine time-series memory updates and multi-agent collaborative work. In order to solve the problem of low memory efficiency management, an independent memory unit is introduced, and an undirected graph structure is used to store the learned information, and each piece of information is stored as a node in the graph. At the same time, memory is added, modified and forgotten according to the time sequence, which alleviates the problem of forgetting important information to a certain extent. In order to solve the problem of insufficient task interaction, a multi-agent collaborative mechanism is introduced, and user demand understanding, task execution and task verification are respectively handed over to three different intelligent agents to work independently. Through the independent work and complete information exchange between the three intelligent agents, information synchronization and status sharing between modules are achieved, information islands are reduced, and decision-making accuracy is improved.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: Based on the multi-agent collaborative framework CrewAI, the present invention includes three agents: AI user, AI assistant, and decider, and a memory unit. After the user input information is processed by the AI ​​user, basic features such as user profile, semantic information, and task requirements are obtained. This basic feature information is stored in the temporal memory unit of the memory system for management. In the temporal memory unit, each time a round of tasks is completed, it is updated according to the memory update formula, compensating for the problems of memory redundancy and memory forgetting caused by traditional single storage. After being decomposed by the AI ​​user, the task is entered into the task queue to solve the problems of task dependency and resource competition. Based on the multi-agent collaborative mechanism, the AI ​​assistant is responsible for executing the task, and the execution results are passed to the decider for judgment of the execution result. At the same time, the information learned from the task execution is passed to the memory unit for memory update. If the result is correct, the information is passed to the AI ​​user to continue to execute the next task. If the result is incorrect, the task is re-executed. By repeatedly executing the task learning-task distribution-task execution-task feedback cycle mechanism, the system is optimized to reduce system resource consumption while improving task processing performance. The task distribution, execution verification, and memory update processes are repeated until the goal of long-term dialogue coherence is achieved.

[0008] The system includes a user interaction and information processing module, a memory unit, a task execution module and a result verification module;

[0009] The user interaction and information processing module: first, processes the information input by the user through AI User; second, performs task decomposition and knowledge expansion; third, prepares for task distribution;

[0010] The memory unit: after the AI ​​user extracts and classifies user information, it constructs a knowledge graph and updates the formula based on the memory after each round of tasks is completed;

[0011] The task execution module: After being decomposed by the AI ​​user, the task is input into the task queue. The AI ​​assistant is then responsible for executing the task, passing the execution result to the result verification module, and at the same time, passing the information learned from the task execution to the memory unit for memory update;

[0012] The result verification module uses a decider to determine the correctness of the execution result. If the result is correct, the information is transmitted to the AI ​​user to continue to perform the next task. If the result is incorrect, the task is re-executed.

[0013] As a preferred embodiment of the present invention, in the user interaction and information processing module:

[0014] AI user is responsible for extracting information from the user side. The system will parse the user input, obtain basic feature information, and store this feature information in the memory unit for management. It then performs task decomposition and knowledge expansion. The system decomposes complex tasks into executable subtasks and assigns a priority to each subtask. For these subtasks, the system uses RAG technology to expand knowledge, match similar cases and solutions in the knowledge graph, add the newly acquired knowledge as leaf nodes to the knowledge graph, and update the memory and integrate it with the existing knowledge along with the original knowledge graph. Finally, it prepares for task distribution. The system will retrieve relevant information from the memory unit based on the current task context, extract detailed information, execution requirements, and knowledge support for the subtasks, and build these complete task information packages into a priority task queue and pass them to the subtask queue.

[0015] As a preferred embodiment of the present invention, the construction of the knowledge graph includes:

[0016] User portraits and task contents are extracted from the user interaction and information processing module as two core nodes in the knowledge graph, and they are connected by an edge to establish an initial relationship. Then, the AI ​​user learns relevant knowledge through RAG technology, and adds each new knowledge point as a leaf node to the graph. These leaf nodes are connected to the task content nodes. In the subsequent processing, when the AI ​​user retrieves information from the memory unit, a search path is formed between different nodes. Whenever a node is searched from another node, a new edge is established between the two nodes, so that the graph can dynamically reflect the relationship between the nodes. Finally, when generating a specific task, the system will identify all the knowledge nodes used, and connect these nodes actually used in the task generation process in sequence through edges to form a complete task-related subgraph structure. At the same time, in the process of the AI ​​assistant executing each round of tasks, the AI ​​assistant will insert the learned knowledge as a leaf node into the subgraph related to this task in the knowledge graph, and update the entire knowledge graph after each round.

[0017] The forgetting probability of a node is calculated by the following formula:

[0018] ,

[0019] Where S is the node importance and T is the time decay factor.

[0020] The node importance S is calculated by the following formula:

[0021] ,

[0022] Among them, C is the node centrality, and the calculation formula is:

[0023] ,

[0024] Where DC is the degree centrality of the node, BC is the betweenness centrality of the node, and EC is the eigenvector centrality of the node. The more connections a node has with its neighbors and the closer the relationship, the greater the degree centrality value of the node, the greater the betweenness centrality value of the node, the greater the eigenvector centrality value of the node, and thus the greater the node centrality value.

[0025] Where F is the access frequency of the node in the last three rounds. R is the weight of the most recent access time, and the calculation formula is:

[0026] ,

[0027] The closer the time the node was accessed is to the current time, the smaller the difference between the current time and the last access time, and thus the greater the time weight. D is the calculation formula for node dependency:

[0028] ,

[0029] The more degrees the node has and the more active nodes are around it, the greater the probability that the node will be visited and the higher the node dependency. In the formula, The sum of the nodes is 1. By calculating the importance of nodes, we can filter out which nodes are more likely to be forgotten.

[0030] T is the time decay factor, and the calculation formula is:

[0031] ,

[0032] Where k is the decay rate parameter, m is the number of rounds that the node has not been visited. The more rounds the node has not been visited, the larger the value of m and the smaller the value of T. Therefore, when a node is less important, the lower the node importance, the higher the time decay factor, and the greater the probability of being forgotten. For core nodes, if the relevance to the current task is greater than 0.5, forgetting is temporarily prohibited. If the degree of the core node is greater than the average node degree, forgetting is stopped.

[0033] As a preferred embodiment of the present invention, in the task execution module:

[0034] First, when the system receives an execution command, the AI ​​assistant will extract key information from the subtask queue. This information includes not only specific subtask requirements and constraints, but also relevant knowledge support retrieved from the memory unit, including historical execution experience, solutions, and precautions. Then, the AI ​​assistant completes the task processing independently. Finally, the AI ​​assistant will pass the complete execution results to the decider for quality assessment and validity verification. At the same time, the AI ​​assistant will pass the new knowledge and experience gained during the task execution (successful solutions, lessons from failure, and optimization strategies, etc.) to the memory unit for knowledge updating.

[0035] As a preferred embodiment of the present invention, in the judgment module:

[0036] Decider is a pre-trained model used to determine whether the execution results of the AI ​​assistant meet the task requirements. After receiving the execution results submitted by the AI ​​assistant, Decider will first extract the key features of the original task requirements and the execution results, and then perform a multi-dimensional quality assessment of the execution results based on preset evaluation criteria. If the execution results meet the requirements, the information will be transmitted to the AI ​​user to enable it to continue to generate the next task. If the execution results do not meet the requirements, the information will be transmitted to the AI ​​user to enable it to regenerate the current task based on the newly learned information.

[0037] The multi-dimensional quality assessment includes functional completeness, logical correctness, efficiency rationality and matching degree with user needs.

[0038] Compared with the prior art, the main advantages of the present invention are:

[0039] (1) This invention adopts a multi-agent collaborative architecture, assigning user interaction, task execution, and result verification to dedicated agents, respectively, and achieving knowledge sharing through memory units constructed using graph neural networks. This design enables the system to accurately understand user needs, efficiently execute tasks, and ensure the quality of execution results. In particular, the RAG technology is used to process user input and update the knowledge graph, significantly improving the system's knowledge acquisition and utilization capabilities.

[0040] (2) This invention introduces a task queue management mechanism, which uses AI users to decompose and distribute tasks. Combined with the historical experience in the memory unit, it can decompose complex tasks into manageable subtasks. This design not only improves the efficiency and accuracy of task execution, but also ensures the quality of execution results through the Decider verification mechanism, effectively solving the limitations of traditional dialogue systems in handling complex tasks.

[0041] (3) The present invention introduces a memory management mechanism based on a graph structure. On the one hand, through the node importance scoring and dynamic forgetting mechanism, the system can adaptively retain key information while timely cleaning up redundant data, thus reducing the resource consumption of memory storage; on the other hand, through the hierarchical structure and relational modeling of the memory graph, effective organization and rapid retrieval of knowledge are achieved, solving the problem of long-term dependency and difficulty in maintaining knowledge association in traditional memory systems;

[0042] (4) This invention adopts a flexible verification feedback mechanism, which uses a Decider to evaluate the task execution results and decide whether to re-execute the task based on the evaluation results. This mechanism not only ensures the output quality, but also continuously optimizes the task execution strategy through the feedback loop, improving the overall performance of the system. At the same time, new knowledge acquired during the execution process will be updated to the memory unit in a timely manner, realizing the continuous learning and evolution of the system.

[0043] (5) The present invention adopts a modular and extensible design concept as a whole. On the one hand, the responsibilities of each intelligent agent are clear and the interface is standardized, which facilitates model replacement and function expansion according to actual needs. On the other hand, through a unified communication protocol and memory sharing mechanism, efficient collaboration between intelligent agents is achieved, which improves the overall performance and maintainability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 Schematic diagram of the structure of the intelligent dialogue system in this embodiment. DETAILED DESCRIPTION

[0045] In order to better understand the present invention, the technical solution of the present invention is further explained below in conjunction with the accompanying drawings and specific embodiments.

[0046] like Figure 1 As shown, this embodiment proposes an intelligent dialogue system for temporal memory management and task distribution based on multi-agent collaboration of CrewAI. The system constructs memory units through the collaborative work of multiple agents and combines the properties of graphs to achieve intelligent processing of complex tasks, continuous accumulation of knowledge and effective control of task execution quality. The system uses GPT-3.5-turbo as the interaction and execution agent, Mistral-7B as the decision-making agent, uses RAG technology for knowledge learning and expansion, and realizes knowledge updating through the properties of graphs to improve the overall performance and reliability of the system. The intelligent dialogue system for temporal memory management and task distribution based on multi-agent collaboration of CrewAI mainly includes a user interaction and information processing module, a memory unit, a task execution module, and a result verification module.

[0047] User interaction and information processing module: In the first step, the AI ​​user is responsible for extracting information from the user side. The system will parse the user input and divide the input information into categories such as user portraits, task requirements, constraints, etc. The second step is task decomposition and knowledge expansion. The system decomposes complex tasks into executable subtasks and assigns a priority to each subtask. For these subtasks, the system uses RAG technology to expand knowledge, match similar cases and solutions in the knowledge graph, add the newly acquired knowledge as leaf nodes to the knowledge graph, and update the memory and integrate the existing knowledge together with the original knowledge graph. The third step is task distribution preparation. The system will retrieve relevant information from the memory unit according to the current task context, extract detailed information, execution requirements and related knowledge support of the subtask, and build these complete task information packages into a priority task queue and pass them to the subtask queue.

[0048] Memory Unit: After the AI ​​user extracts and classifies user information, a knowledge graph is constructed. The user profile and task content are extracted from the user input as two core nodes in the knowledge graph, and an edge connects them to establish an initial relationship. The AI ​​user then learns relevant knowledge using RAG technology. Each new knowledge point is added to the graph as a leaf node, primarily connected to the task content node. During subsequent processing, when the AI ​​user retrieves information from the memory unit, a search path is formed between different nodes. Whenever a node is searched from another, a new edge is established between the two nodes, allowing the graph to dynamically reflect the connections between nodes. Finally, when generating a specific task, the system identifies all knowledge nodes used and connects them sequentially through edges, forming a complete task-related subgraph structure. Furthermore, during each round of task execution, the AI ​​assistant inserts the learned knowledge as leaf nodes into the subgraph of the knowledge graph relevant to that task. The entire knowledge graph is updated after each round.

[0049] The forgetting probability of a node is calculated by the following formula:

[0050] ,

[0051] Where S is the node importance and T is the time decay factor.

[0052] The node importance is calculated by the following formula:

[0053] ,

[0054] Among them, C is the node centrality, and the calculation formula is:

[0055] ,

[0056] Where DC is the degree centrality of the node, BC is the betweenness centrality of the node, and EC is the eigenvector centrality of the node. The more connections a node has with its neighbors and the closer the relationship, the greater the degree centrality value of the node, the greater the betweenness centrality value of the node, the greater the eigenvector centrality value of the node, and thus the greater the node centrality value.

[0057] Where F is the access frequency of the node in the last three rounds. R is the weight of the most recent access time, and the calculation formula is:

[0058] ,

[0059] The closer the time the node was accessed is to the current time, the smaller the difference between the current time and the last access time, and thus the greater the time weight. D is the calculation formula for node dependency:

[0060] ,

[0061] The more degrees the node has and the more active nodes are around it, the greater the probability that the node will be visited and the higher the node dependency. In the formula, The sum of the nodes is 1. By calculating the importance of nodes, we can filter out which nodes are more likely to be forgotten.

[0062] T is the time decay factor, and the calculation formula is:

[0063] ,

[0064] Where k is the decay rate parameter, and m is the number of rounds the node has not been visited. The more rounds the node has not been visited, the larger the value of m and the smaller the value of T. Therefore, the less important a node is, the lower its node importance, the higher its time decay factor, and the greater its probability of being forgotten. For core nodes, if their relevance to the current task is greater than 0.5, forgetting is temporarily prohibited. If the degree of a core node is greater than the average node degree, forgetting is stopped. By implementing memory management, the system can more efficiently store information and improve information utilization.

[0065] Task execution module. First, when the system receives an execution command, the AI ​​assistant will extract key information from the subtask queue. This information includes not only specific subtask requirements and constraints, but also relevant knowledge support retrieved from the memory unit, such as historical execution experience, solutions, and precautions. In the second step, the AI ​​assistant completes the task processing independently. In the third step, the AI ​​assistant will pass the complete execution results to the decider for quality assessment and validity verification. At the same time, the AI ​​assistant will pass the new knowledge and experience gained during the task execution process, such as successful solutions, lessons from failure, and optimized strategies, to the memory unit for knowledge updating, thereby continuously improving the system's execution capabilities and adaptability.

[0066] Judgment module. Decider is a pre-trained model used to determine whether the execution results of the AI ​​assistant meet the task requirements. After receiving the execution results submitted by the AI ​​assistant, Decider will first extract the key features of the original task requirements and the execution results. Then, based on the preset evaluation criteria, it will conduct a multi-dimensional quality assessment of the execution results, including functional completeness, logical correctness, efficiency rationality, and matching with user needs. If the execution result meets the requirements, the information will be passed to the AI ​​user to continue generating the next task. If the execution result does not meet the requirements, the information will be passed to the AI ​​user to regenerate the current task based on the newly learned information.

[0067] The above embodiments are intended only to illustrate the technical concepts and features of the present invention. Their purpose is to enable those skilled in the art to understand the contents of the present invention and implement them accordingly. They are not intended to limit the scope of protection of the present invention. Any equivalent changes or modifications made in accordance with the spirit of the present invention are intended to be covered by the scope of protection of the present invention.

Claims

1. An intelligent dialogue system for temporal memory management and task distribution, based on CrewAI's multi-agent collaboration, featuring: It includes user interaction and information processing module, memory unit, task execution module and result verification module; The user interaction and information processing module: first, processes the information input by the user through AI User; second, performs task decomposition and knowledge expansion; third, prepares for task distribution; The memory unit: after the AI ​​user extracts and classifies user information, it constructs a knowledge graph and updates the formula based on the memory after each round of tasks is completed; The task execution module: After being decomposed by the AI ​​user, the task is input into the task queue. The AI ​​assistant is then responsible for executing the task, passing the execution result to the result verification module, and at the same time, passing the information learned from the task execution to the memory unit for memory update; The result verification module: determines the correctness of the execution result through the decider. If the result is correct, the information is transmitted to the AI ​​user to continue to perform the next task. If the result is incorrect, the task is re-executed; The construction of the knowledge graph includes: User portraits and task content are extracted from the user interaction and information processing module as two core nodes in the knowledge graph, and an edge is used to connect them to establish an initial relationship. Then, the AI ​​user learns relevant knowledge through RAG technology, and each time a new knowledge point is acquired, it is added to the graph as a leaf node. These leaf nodes are connected to the task content node. In the subsequent processing, when the AI ​​user retrieves information from the memory unit, a search path is formed between different nodes. Whenever a node is searched from another node, a new edge is established between the two nodes, so that the graph can dynamically reflect the relationship between nodes. Finally, when generating a specific task, the system will identify all the knowledge nodes used and connect these nodes actually used in the task generation process in sequence through edges to form a complete task-related subgraph structure. At the same time, when the AI ​​assistant executes each round of tasks, the AI ​​assistant will insert the learned knowledge as leaf nodes into the subgraph related to this task in the knowledge graph, and the entire knowledge graph will be updated after each round. The forgetting probability of a node is calculated by the following formula: ， Where S is the node importance and T is the time decay factor; The node importance S is calculated by the following formula: ， Among them, C is the node centrality, and the calculation formula is: ， Among them, DC is the degree centrality of the node, BC is the betweenness centrality of the node, and EC is the eigenvector centrality of the node. The more connections a node has with its adjacent nodes in the graph, and the closer the relationship, the greater the degree centrality value of the node, the greater the betweenness centrality value of the node, the greater the eigenvector centrality value of the node, and thus the greater the node centrality value; Among them, F is the access frequency of the node in the last three rounds, R is the weight of the most recent access time, and the calculation formula is: ， The closer the time the node is accessed is to the current time, the smaller the difference between the current time and the last access time is, and thus the greater the time weight is. D is the node dependency calculation formula: ， The more degrees the node has and the more active nodes are around it, the greater the probability that the node will be visited and the higher the node dependence. In the formula, The sum of is 1. By calculating the importance of nodes, we can filter out which nodes are more likely to be forgotten. T is the time decay factor, and the calculation formula is: ， Where k is the decay rate parameter, m is the number of rounds that the node has not been visited. The more rounds the node has not been visited, the larger the value of m and the smaller the value of T. Therefore, when a node is less important, the lower the node importance, the higher the time decay factor, and the greater the probability of being forgotten. For core nodes, if the relevance to the current task is greater than 0.5, forgetting is temporarily prohibited. If the degree of the core node is greater than the average node degree, forgetting is stopped.

2. The intelligent dialogue system for temporal memory management and task distribution according to claim 1, characterized in that: The processing of user input information by AI User is specifically as follows: AI user is responsible for extracting information from the user side. The system will parse the user input, obtain basic feature information, and store this basic feature information in the memory unit for management. The basic feature information includes user portraits, task requirements, semantic information and constraints.

3. The intelligent dialogue system for temporal memory management and task distribution according to claim 1, characterized in that: The task decomposition and knowledge expansion are specifically as follows: The system breaks down complex tasks into executable subtasks and assigns a priority to each subtask. For these subtasks, the system uses RAG technology to expand knowledge, match similar cases and solutions in the knowledge graph, add the newly acquired knowledge as leaf nodes to the knowledge graph, and update memory and integrate existing knowledge along with the original knowledge graph.

4. The intelligent dialogue system for temporal memory management and task distribution according to claim 1, characterized in that: The task distribution preparation is specifically as follows: The system retrieves relevant information from the memory unit based on the current task context, extracts detailed information, execution requirements, and knowledge support for subtasks, and constructs these complete task information packages into priority task queues and passes them to subtask queues.

5. The intelligent dialogue system for temporal memory management and task distribution according to claim 1, characterized in that: In the task execution module: First, when the system receives an execution command, the AI ​​assistant extracts key information from the subtask queue. This information includes not only specific subtask requirements and constraints, but also relevant knowledge support retrieved from the memory unit. Then, the AI ​​assistant independently completes the task processing. Finally, the AI ​​assistant passes the complete execution results to the decider for quality assessment and validity verification. At the same time, the AI ​​assistant passes the new knowledge and experience gained during the task execution to the memory unit for knowledge updating.

6. The intelligent dialogue system for temporal memory management and task distribution according to claim 5, characterized in that: The relevant knowledge support includes historical execution experience, solutions and precautions; the new knowledge and experience include successful solutions, lessons from failure and optimization strategies.

7. The intelligent dialogue system for temporal memory management and task distribution according to claim 1, characterized in that: In the judgment module: Decider is a pre-trained model used to determine whether the execution results of the AI ​​assistant meet the task requirements. After receiving the execution results submitted by the AI ​​assistant, Decider will first extract the key features of the original task requirements and the execution results, and then perform a multi-dimensional quality assessment of the execution results based on preset evaluation criteria. If the execution results meet the requirements, the information will be transmitted to the AI ​​user to enable it to continue to generate the next task. If the execution results do not meet the requirements, the information will be transmitted to the AI ​​user to enable it to regenerate the current task based on the newly learned information.

8. The intelligent dialogue system for temporal memory management and task distribution according to claim 5, characterized in that: The multi-dimensional quality assessment includes functional completeness, logical correctness, efficiency rationality and matching degree with user needs.

9. The intelligent dialogue system for temporal memory management and task distribution according to claim 1, characterized in that: The system works through collaboration between multiple intelligent agents, including AI User based on GPT-3.5-turbo, AI Assistant based on GPT-3.5-turbo, and Decider based on Mistral-7B.

Images