A multi-field task processing method, system, device and medium based on knowledge fusion and agent cooperation

Abstract

The application provides a multi-field task processing method, system, device and medium based on knowledge fusion and Agent cooperation, and belongs to the technical field of cross-field knowledge fusion. The application generates a sub-task set by decomposing a user input task; constructs a specialized Agent set according to a skill mapping table and a role template library; performs cross-field knowledge retrieval and fusion on the Agent set, monitors output consistency and classifies collaboration conflicts; triggers corresponding expansion mechanisms for different conflicts, updates a knowledge graph or generates a new Agent; and simultaneously monitors field knowledge updates, adjusts knowledge weights and synchronizes to related modules. The application realizes accurate decomposition and specialized processing of complex tasks, improves cross-field knowledge fusion efficiency, solves collaboration conflicts, enhances system adaptability, ensures processing timeliness through real-time knowledge updates, and improves the task processing quality and efficiency of a multi-Agent system.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of cross-domain knowledge fusion, and particularly relates to a multi-domain task processing method, system, device and medium based on knowledge fusion and Agent cooperation. BACKGROUND

[0002] In the current environment of accelerating digital transformation, the above-mentioned technology is particularly suitable for industries that require highly specialized knowledge and multidisciplinary collaboration, such as medical health, financial technology, intelligent manufacturing, scientific research and education.

[0003] In related technologies, when facing complex cross-domain tasks, the professional fields involved cannot be accurately identified, resulting in chaotic sub-task decomposition, overlap or omission between sub-tasks, and affecting processing efficiency. At present, although Agents are used, the problem of mismatch between Agent capabilities and sub-task requirements still exists. The generation of Agents often lacks precise association with sub-tasks, and the skills, tool interfaces and reasoning strategies do not match the requirements of sub-tasks, resulting in insufficient professional processing of sub-tasks. When multiple Agents cooperate, conflicts are easily produced due to inconsistent output semantics, but there is a lack of real-time monitoring and effective classification mechanism, resulting in conflicts that cannot be discovered in time or are difficult to solve specifically due to unclear conflict types. SUMMARY

[0004] The application provides a multi-domain task processing method based on knowledge fusion and Agent cooperation, which improves the processing efficiency and accuracy of complex tasks and enhances the adaptability of the system to cross-domain and changing tasks.

[0005] The method comprises the following steps:

[0006] Step S101: performing domain recognition and sub-task decomposition on the task input by the user, identifying the semantic elements involved in the task, and generating a sub-task set;

[0007] Step S102: according to the sub-task set, calling a skill mapping table to obtain the corresponding relationship between the professional skills and tool interfaces required by the sub-tasks, combining the Agent behavior patterns and reasoning strategies in the role template library to generate specialized Agent instances, and forming an Agent set;

[0008] Step S103: performing cross-domain knowledge retrieval and fusion on the Agent set;

[0009] Step S104: based on the multi-source knowledge fusion results, monitoring the output consistency of the Agent set, analyzing the semantic consistency of the output results of different Agents in real time, and classifying the types of cooperation conflicts according to the semantic consistency analysis results;

[0010] Step S105: If a collaboration conflict is detected and the confidence level of the Agent output is lower than a first preset threshold, trigger vertical knowledge expansion to update the knowledge graph of the corresponding domain;

[0011] When the collaboration conflict rate exceeds a second preset threshold, trigger horizontal Agent expansion to generate a new specialized Agent instance;

[0012] When the new concept recognition frequency exceeds a third preset threshold, trigger knowledge graph update to supplement new concepts and their associated relationships;

[0013] Step S106: Monitor new domain knowledge updates, update the knowledge graph of the corresponding domain, calculate the comprehensive weights of the timeliness, relevance, and credibility of each knowledge segment in the knowledge graph, and reload the updated knowledge weights and knowledge graph information.

[0014] Preferably, the execution of cross-domain knowledge retrieval and fusion in step S103 includes: a first layer domain knowledge retriever retrieves domain-specific information from a professional knowledge base based on the specialized skill requirements of the Agent, a second layer collaboration memory retriever retrieves solutions for similar tasks from a historical collaboration case library, and a knowledge weighting fusioner uses a comprehensive weighting algorithm of timeliness, relevance, and credibility to fuse the retrieval results of the two layers to generate a multi-source knowledge fusion result.

[0015] Preferably, step S102 specifically includes:

[0016] For each subtask in the subtask set generated in step S101, the text description is parsed through natural language processing technology to extract the domain label, functional requirement, and input-output requirement of the subtask;

[0017] The association between the domain label and the professional skill, tool interface is configured in the skill mapping table, and by querying the skill mapping table, a set of professional skills and a list of tool interfaces that completely match the domain label of the subtask are obtained;

[0018] The behavior patterns and reasoning strategies of the Agent are configured in the role template library according to the domain classification, and the closest combination of behavior patterns and reasoning strategies is selected through a semantic matching algorithm according to the functional requirements of the subtask;

[0019] The obtained skill set, tool interface, and selected behavior patterns, reasoning strategies are structured and encapsulated, and an instantiation engine is used to generate an independent Agent instance for each subtask.

[0020] Preferably, step S103 specifically includes:

[0021] The first layer domain knowledge retrieval result and the second layer collaboration memory retrieval result are collected respectively, and the timeliness tag, relevance label, and credibility identifier of each piece of knowledge are extracted;

[0022] According to the preset mapping rule, the timeliness label is converted into a timeliness quantitative value, the relevance label is converted into a relevance quantitative value, and the credibility identifier is converted into a credibility quantitative value;

[0023] For each piece of knowledge, the fusion weight of a single piece of knowledge is calculated in a manner of timeliness quantitative value x a + relevance quantitative value x b + credibility quantitative value x g, and the weight result is marked; a, b, and g are predefined weight distribution coefficients, and a + b + g = 1;

[0024] According to the knowledge fusion weight from high to low, the knowledge whose weight reaches a threshold is selected, and the knowledge is aggregated according to the sub-task logic to generate a knowledge fusion result.

[0025] Preferably, step S104 specifically comprises:

[0026] The output results of the professionalized Agent instances in step S103 are collected and converted into a unified intermediate representation format;

[0027] For each standardized output content, a natural language processing technique is used to extract key semantic elements;

[0028] Based on the generated semantic vector, the following methods are used to calculate the consistency of the outputs of different Agents under the same task:

[0029] The cosine similarity of the semantic vectors of the outputs of any two Agents is calculated;

[0030] The proportion of the number of completely matched key entities in the two outputs in the total number of entities is counted;

[0031] It is judged whether the intention descriptions of the two outputs belong to the same task sub-target;

[0032] The scores of the above three dimensions are integrated to generate an overall consistency score of all Agent outputs;

[0033] A preset conflict classification rule library is provided, and according to the semantic feature difference, the conflicts are classified into data inconsistency, logical inconsistency, and target inconsistency.

[0034] Preferably, step S105 specifically comprises:

[0035] The conflict type, severity, associated Agent ID, and conflict content information output in step S104 are collected and converted into related data to form an inconsistency event library;

[0036] Based on the conflict events, a preset extension strategy rule library is called for matching:

[0037] If the conflict type is data inconsistency and the severity level is high, match the vertical knowledge expansion strategy;

[0038] If the conflict type is logical inconsistency and the severity level is medium, match the horizontal Agent expansion strategy;

[0039] If the conflict type is target inconsistency or the new concept recognition frequency exceeds θ3, match the knowledge graph update strategy;

[0040] If multiple types of conflicts are met simultaneously, according to the priority, data inconsistency > logical inconsistency > target inconsistency, select the main expansion strategy;

[0041] According to the matched strategy, perform vertical and horizontal expansion operations; verify the expanded knowledge graph and new Agent instances.

[0042] Preferably, step S106 specifically comprises:

[0043] When the knowledge base has new, modified or deleted operations, the knowledge base pushes knowledge update events;

[0044] According to the monitored update content, modify the knowledge graph to ensure the atomicity and traceability of the update operation:

[0045] If the update content is a new concept not included, create a new node in the knowledge graph, and mark the type, definition, source and creation time;

[0046] If the update content is a new association of an existing concept, add a bidirectional edge in the knowledge graph, and mark the association strength and basis;

[0047] If the update content is the attribute adjustment of an existing node, modify the attribute value of the corresponding node, and keep the historical version record;

[0048] Based on the updated knowledge graph, recalculate the comprehensive weight of timeliness, relevance and credibility for each knowledge fragment;

[0049] Synchronize the update content and weight change of the knowledge graph to the Agent set of step S102 and step S103.

[0050] The application also provides a multi-field task processing system based on knowledge fusion and Agent cooperation, which comprises:

[0051] A task recognition module is configured to perform field recognition and sub-task decomposition on the user input task, recognize semantic elements involved in the task, and generate a sub-task set;

[0052] An instantiation module is configured to obtain a correspondence between professional skills and tool interfaces required by the subtasks according to the subtask set, and generate professional Agent instances by combining Agent behavior patterns and reasoning strategies in the role template library, thereby forming an Agent set.

[0053] A cross-domain fusion module is configured to perform cross-domain knowledge retrieval and fusion on the Agent set.

[0054] A collaboration monitoring module is configured to monitor output consistency of the Agent set based on the multi-source knowledge fusion result, analyze semantic consistency of different Agent output results in real time, and classify types of collaboration conflicts according to the semantic consistency analysis result.

[0055] An extension triggering module is configured to trigger vertical knowledge extension to update the knowledge graph of the corresponding domain if the collaboration conflict is monitored and the confidence of the Agent output result is lower than a first preset threshold.

[0056] When the collaboration conflict rate exceeds a second preset threshold, horizontal Agent extension is triggered to generate new professional Agent instances.

[0057] When the new concept recognition frequency exceeds a third preset threshold, knowledge graph update is triggered to supplement new concepts and their associated relationships.

[0058] A graph continuous update module is configured to monitor new domain knowledge updates, update the knowledge graph of the corresponding domain, calculate comprehensive weights of timeliness, relevance and credibility of each knowledge segment in the knowledge graph, and reload updated knowledge weights and knowledge graph information.

[0059] According to another embodiment of the present application, an electronic device is provided, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the multi-domain task processing method based on knowledge fusion and Agent collaboration when executing the program.

[0060] According to another embodiment of the present application, an electronic device is provided, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the multi-domain task processing method based on knowledge fusion and Agent collaboration when executing the program.

[0061] From the above technical solutions, the present application has the following advantages:

[0062] The multi-field task processing method based on knowledge fusion and Agent cooperation provided by the application can grasp the key of the task by identifying the professional field involved in the task and extracting core semantic elements, and the independent sub-task set generated based on the field ontology and the template library reduces the processing difficulty of complex tasks, provides a clear target for Agent allocation, and reduces information omission. According to the sub-task skill mapping table and the role template library, the generated specialized Agent has matching skills, tool interfaces and reasoning strategies, which improves the professionalism and accuracy of sub-task processing, and the formation of Agent set enables complex tasks to be efficiently promoted through collaborative mode. Multi-level knowledge retrieval takes into account field-specific information and historical cases, making knowledge acquisition more comprehensive; the comprehensive weighted fusion mechanism balances the timeliness, relevance and reliability of knowledge, making the fusion result more reliable, providing high-quality knowledge support for task processing, and reducing the impact of one-sided information. Real-time monitoring of the semantic consistency of Agent output can timely detect collaboration conflicts; classification of conflict types and severity provides a clear direction for subsequent conflict resolution, avoiding conflict spread affecting the overall progress and quality of the task. The extension mechanism triggered by different conflicts, vertical knowledge extension can deepen the knowledge graph, horizontal Agent extension can enhance the processing capacity, and the supplement of new concepts and associated can perfect the knowledge system, ensuring that the system can adapt to changing task requirements and improve the accuracy of conflict resolution. Monitor and update domain knowledge to ensure the timeliness and accuracy of the knowledge graph; recalculate the knowledge weight and synchronize it to the related modules, so that the Agent and the knowledge routing engine work based on the latest information, maintain long-term efficient operation of the system, and avoid processing deviation caused by knowledge lag. BRIEF DESCRIPTION OF DRAWINGS

[0063] In order to more clearly illustrate the technical solutions of the present application, the drawings needed to be used in the description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0064] Figure 1 The flow chart of the multi-field task processing method based on knowledge fusion and Agent cooperation;

[0065] Figure 2 The schematic diagram of the multi-field task processing system based on knowledge fusion and Agent cooperation;

[0066] Figure 3 The schematic diagram of the electronic device. DETAILED DESCRIPTION

[0067] The application provides a multi-field task processing method based on knowledge fusion and agent cooperation, which comprises a task adaptive decomposer, a field-aware agent constructor, a multi-level knowledge routing engine and an intelligent conflict mediator.

[0068] The multi-field task processing method based on knowledge fusion and agent cooperation will be described in detail below. To illustrate but not to limit, specific details such as specific system structures and technologies are given to help a thorough understanding of the embodiments of the application. However, it should be clear to those skilled in the art that the application can be implemented in other embodiments without these specific details.

[0069] To clearly describe the technical solutions of the application, the same or similar items with basically the same functions and effects are distinguished by the first, second, etc. The skilled in the art can understand that the first, second, etc. do not limit the quantity and execution order, and the first, second, etc. do not necessarily mean different.

[0070] The phrase one or some embodiments described in the application means that the specific features, structures or characteristics described in the embodiment are included in one or more embodiments of the application. Therefore, the phrases one embodiment, some embodiments, other embodiments, additional embodiments, etc. appearing in different places in the application do not necessarily refer to the same embodiment, but mean one or more but not all embodiments, unless otherwise specifically emphasized.

[0071] The technical solutions in the embodiments of the application will be described clearly and completely in combination with the drawings in the embodiments of the application. Obviously, the described embodiments are only some of the embodiments of the application, not all. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the application.

[0072] Please refer to Figure 1 The flowchart of the multi-field task processing method based on knowledge fusion and agent cooperation in a specific embodiment is shown in FIG. 1, and the method comprises the following steps.

[0073] Step S101: Field recognition and sub-task decomposition are performed on the task input by the user, semantic elements involved in the task are recognized, and a sub-task set is generated.

[0074] In some embodiments, the text of the user input task is vector converted by a pre-trained BERT domain classification model, matched with each domain feature vector library, and one or more professional fields involved are identified; the semantic analysis module uses dependency syntax analysis and named entity recognition technology to extract core entities, actions and constraints in the task; the domain ontology is a concept hierarchical system for the domain. For example, the concept hierarchical system in the computer field includes hardware and software sub-concepts.

[0075] The task template library stores the decomposition cases of historical similar tasks, and the original task is decomposed into independent sub-tasks by combining the two, which cover all core elements of the original task without overlapping.

[0076] The embodiment is based on natural language processing technology to analyze task text, determine the professional field by matching domain features, and then decompose complex tasks into independent sub-tasks according to the domain knowledge structure and historical task decomposition experience, ensuring the integrity and independence of the sub-tasks. Improve the pertinence of subsequent processing; sub-task decomposition reduces task complexity, facilitates specialized processing, reduces information omission, and provides clear targets for agent allocation.

[0077] Step S102: According to the sub-task set, the skill mapping table is called to obtain the corresponding relationship between the professional skills and tool interfaces required by the sub-tasks, and the professional agent instances are generated by combining the behavior patterns and reasoning strategies of the agent in the role template library, forming an agent set.

[0078] In some embodiments, the skill mapping table stores the mapping relationship between the sub-task type and the required professional skills and tool interfaces in the form of key-value pairs; the role template library includes the behavior patterns and reasoning strategies of agents in each domain. Optionally, the reasoning strategy can be a combination of rule-based reasoning and probabilistic reasoning for medical domain agents.

[0079] According to the sub-tasks, the matching skills and tool interfaces are obtained from the skill mapping table, the templates of the corresponding domain in the role template library are called, and the agents with specific skills, tool calling permissions and reasoning capabilities are instantiated and generated, multiple agents form an agent set, and each agent corresponds to one or more sub-tasks.

[0080] As can be seen, according to the skill requirements and tool dependencies of the sub-tasks, the corresponding resources and templates are matched to generate specialized agent instances, so that each agent has the ability to handle specific sub-tasks, forming a collaborative processing system. The agent has specific skills and tool calling capabilities, improving the professionalism and efficiency of sub-task processing; based on the template generation, the agent behavior conforms to the domain specification.

[0081] Step S103: Perform cross-domain knowledge retrieval and fusion for the Agent set.

[0082] In some embodiments, cross-domain knowledge retrieval and fusion are performed through a multi-layered knowledge routing engine. The first-layer domain knowledge retrieval engine retrieves domain-specific information from a professional knowledge base based on the agent's specialized skill requirements. The second-layer collaborative memory retrieval engine retrieves solutions to similar tasks from a historical collaborative case library. The knowledge weighted fusion engine uses a comprehensive weighting algorithm based on timeliness, relevance, and credibility to fuse the retrieval results from the two layers, generating a multi-source knowledge fusion result.

[0083] Specifically, the first-layer domain knowledge retrieval unit receives the agent's specialized skill requirements and retrieves domain-specific information from the professional knowledge base using semantic vector retrieval technology. The second-layer collaborative memory retrieval unit uses case attribute matching to retrieve task solutions with similarity ≥ a threshold from the historical collaborative case library. In the knowledge weighted fusion unit, timeliness is calculated by the difference between the information release time and the current time, relevance is calculated by semantic overlap, and credibility is calculated by the authority level of the information source. The three factors are combined to weight and merge the retrieval results from the two layers to generate a fused result.

[0084] It can be seen that by acquiring professional knowledge and historical processing experience in the field through hierarchical retrieval, and then balancing the value of multi-source information through weighted fusion, comprehensive and prioritized knowledge support can be provided to the Agent.

[0085] Step S104: Based on the results of multi-source knowledge fusion, monitor the output consistency of the Agent set, analyze the semantic consistency of the output results of different Agents in real time, and classify the types of collaboration conflicts according to the semantic consistency analysis results.

[0086] In some embodiments, the collaborative conflict intelligent mediator collects the output results of the Agent set in real time and uses a semantic similarity calculation method to analyze the consistency of the results; when the similarity is lower than the threshold, the conflict type is classified by rule base, the severity is determined by combining the criticality of the tasks involved in the conflict, and a conflict analysis report is generated.

[0087] This embodiment uses semantic analysis technology to monitor the consistency of Agent output, identifies the nature and impact of conflicts based on preset rules, promptly detects inconsistencies in Agent collaboration, and avoids erroneous output results.

[0088] Step S105: If a collaboration conflict is detected, an expansion mechanism is triggered according to the conflict type and the preset threshold. When the confidence of the Agent output is lower than the preset threshold θ1, vertical knowledge expansion is triggered to update the knowledge graph in the corresponding field. When the collaboration conflict rate exceeds the preset threshold θ2, horizontal Agent expansion is triggered to generate a new specialized Agent instance. When the frequency of new concept recognition exceeds the preset threshold θ3, knowledge graph updating is triggered to supplement new concepts and their associated relationships.

[0089] In some embodiments, the preset thresholds θ1 (such as 0.6), θ2 (such as 30%), and θ3 (such as 5 times / hour). When the confidence of the Agent output is lower than θ1, the vertical knowledge expansion module calls the domain knowledge crawler to grab the latest literature from authoritative databases, extracts the depth information of the knowledge, and updates the knowledge graph. When the collaboration conflict rate exceeds θ2, the horizontal Agent expansion module generates a new Agent instance in the same field based on the role template library to supplement the skill gaps. When the frequency of new concept recognition exceeds θ3, the knowledge graph updating module analyzes the definition and context of the new concept, adds new nodes and their associated relationships with existing nodes.

[0090] As can be seen, by comparing the confidence of the conflict, the conflict rate, and the frequency of new concepts with the preset thresholds, the corresponding expansion strategy is triggered to adjust the system resources from the depth of knowledge, the number of Agents, and the breadth of knowledge, thereby solving the collaboration conflict and avoiding excessive expansion of resources. By adjusting the knowledge and Agents, the system's ability to handle complex tasks is improved, and the adaptability is enhanced.

[0091] Step S106: Monitor new domain knowledge updates, update the knowledge graph in the corresponding field, calculate the comprehensive weight of the timeliness, relevance, and credibility of each knowledge segment in the knowledge graph, and reload the updated knowledge weight and knowledge graph information.

[0092] In some embodiments, new domain knowledge is monitored through RSS subscription, API interface listening, and other methods. Entity linking and relationship extraction techniques are used to add entities and relationships in the new knowledge to the knowledge graph and update the attributes of the original nodes. By traversing each knowledge segment in the knowledge graph, the timeliness, relevance, and credibility are recalculated, and the new weight value is obtained by synthesizing the three. The knowledge updating notification is sent to the Agent set through the message queue, and the Agent receives and reloads the knowledge weight and graph information, and sends an update instruction to the knowledge routing engine, so that the retrieval strategy is adapted to the new knowledge weight.

[0093] This embodiment maintains the timeliness and accuracy of the knowledge graph, avoids system dependence on outdated knowledge leading to errors, and adjusts the knowledge weight to prioritize key information, improving the processing quality of the Agent and the knowledge routing engine, and enhancing the long-term effectiveness of the system.

[0094] In one embodiment of the present invention, based on step S152, the following will provide a possible embodiment and its specific implementation will be described in a non-limiting manner. Step S102 specifically includes:

[0095] Step S1021: For each subtask in the subtask set generated in step S101, parse the text description using natural language processing technology to extract the subtask's domain labels, functional requirements, and input / output requirements.

[0096] It should be noted that, to ensure accurate matching between the skill set and the requirements of the sub-tasks, an ontology-based semantic similarity algorithm is used to calculate the similarity S between the sub-task domain labels and the domain labels in the skill mapping table. S = |C_task|C_skill| |C_task|C_skill|.

[0097] Here, Task C is the set of domain concepts involved in the subtask, and Skill C is the set of domain concepts in the skill mapping table. A match is considered successful when S ≥ θ4.

[0098] Step S1022: Configure the association between domain tags and professional skills and tool interfaces in the skill mapping table. By querying the skill mapping table, obtain the set of professional skills and the list of tool interfaces that completely match the domain tags of the subtask.

[0099] Step S1023: Configure the Agent's behavior patterns and inference strategies by domain in the role template library, and select the closest combination of behavior patterns and inference strategies based on the functional requirements of the sub-tasks using a semantic matching algorithm.

[0100] Step S1024: The skill set and tool interface obtained in step S1022 are structurally encapsulated with the behavior patterns and inference strategies selected in step S1023. An independent Agent instance is generated for each subtask through the instantiation engine. Each instance includes: a skill execution module, a tool call interface, a behavior control module, and an inference engine.

[0101] Below is a specific example of an annual development plan for a company that includes market analysis, financial budgeting, and legal advice.

[0102] Step S1021: For the market analysis subtask decomposed from S101, the natural language processing technology parses its text description and extracts the domain label as marketing, the functional requirement is to analyze the target market size, competitor situation and consumer preference, the input requirements include enterprise existing market data, industry report, and the output requirements are market share prediction table and competition strategy suggestion. Then calculate the similarity of the domain label in the skill mapping table of this subtask, assuming that C task is {market size, competitor, consumer preference}, C skill is {market size, competitor, consumer behavior, market share}, then S=0.75, if θ4 is set to 0.6, because 0.75≥0.6, it is determined that the matching is successful.

[0103] Step S1022: The professional skill set associated with the marketing domain label in the skill mapping table includes market research skills, data statistical analysis skills, and competitive situation analysis skills, and the tool interface list has market data query interface, statistical analysis software interface, and industry report database interface. By querying the table, the above-mentioned professional skill set and tool interface list matched with the market analysis subtask are obtained.

[0104] Step S1023: The Agent behavior mode of the marketing domain in the role template library has the mode of periodically collecting market data, analyzing data trends, generating preliminary reports, and adjusting according to feedback, and the reasoning strategy includes trend prediction method based on historical data and comparative analysis reasoning method. According to the functional requirement of analyzing the target market size and giving strategy suggestions in the market analysis subtask, the closest behavior mode and reasoning strategy combination are selected through the semantic matching algorithm, that is, the above-mentioned behavior mode and two reasoning strategies.

[0105] Step S1024: The market research skills and market data query interface obtained in step S1022 are structurally encapsulated with the behavior mode and reasoning strategy selected in step S1023, and the market analysis Agent instance is generated by instantiating the engine. In this instance, the skill execution module is responsible for processing data using market research skills, the tool calling interface connects market data query tools, the behavior control module operates according to the mode of collecting data, analyzing trends, generating reports, and adjusting, and the reasoning engine uses the trend prediction method and the comparative analysis reasoning method for analysis.

[0106] Through precise semantic analysis and similarity matching, the matching accuracy of subtasks, professional skills and tool interfaces is ensured, and the problem of mismatch between skills and task requirements is avoided; based on the role template library, the appropriate behavior mode and reasoning strategy are selected, so that the generated Agent instance has a working mode that meets the characteristics of the domain, and the professionalism and efficiency of task processing are improved.

[0107] In an embodiment of the present application, based on step S103, a possible embodiment will be given below to specifically and non-limitingly illustrate the specific implementation thereof. Step S103 specifically comprises:

[0108] Step S1031: Collect the first layer domain knowledge retrieval result and the second layer collaborative memory retrieval result respectively, and extract the timeliness label, the relevance label and the credibility identifier of each piece of knowledge.

[0109] Step S1032: According to the preset mapping rule, convert the timeliness label into a timeliness quantitative value; convert the relevance label into a relevance quantitative value; and convert the credibility identifier into a credibility quantitative value.

[0110] Step S1033: For each piece of knowledge, calculate the fusion weight of a single piece of knowledge in the manner of timeliness quantitative value x a + relevance quantitative value x b + credibility quantitative value x g, and mark the weight result. a, b, g are pre-defined weight distribution coefficients, and a + b + g = 1.

[0111] Step S1034: Sort according to the knowledge fusion weight from high to low, select the knowledge whose weight reaches the threshold, aggregate according to the sub-task logic, and generate the knowledge fusion result.

[0112] The embodiment can more scientifically evaluate the value of each piece of knowledge by quantifying the timeliness, relevance and credibility of the knowledge and comprehensively calculating the weight, so that the screened knowledge is more in line with the task requirements. According to the weight order and the selected qualified knowledge, it is ensured that the final fusion result contains knowledge with higher value, and the effectiveness of the knowledge is improved. According to the sub-task logic, the aggregated knowledge can closely revolve around the sub-task, enhancing the relevance of the knowledge and the sub-task and improving the efficiency and quality of the task processing.

[0113] In an embodiment of the present application, based on step S104, a possible embodiment will be given below to specifically and non-limitingly illustrate the specific implementation thereof. Step S104 specifically comprises:

[0114] Step S1041: Collect the output results of each specialized Agent instance in step S103, and convert them into a unified intermediate representation format, including task ID, Agent ID, output content, timestamp, associated sub-task identifier and the like, to ensure that the outputs from different sources are comparable.

[0115] Step S1042: For each standardized output content, extract key semantic elements using natural language processing technology.

[0116] Exemplarily, the extraction of semantic features can employ a pre-trained BERT language model to encode the output content and capture implicit semantic information. For example, a patient age of 30 years and a patient age of thirty years will be identified by the model as the same entity value; and a drug A suitable for type 1 diabetes and a drug A for type 1 diabetes treatment will be identified as the same intent.

[0117] Step S1043: Based on the semantic vector generated in step S1042, the consistency of the outputs of different Agents under the same task is calculated by the following method:

[0118] Calculate the cosine similarity of the semantic vectors of any two Agent outputs;

[0119] Statistically, the proportion of the number of completely matched key entities in the total number of entities in the two outputs;

[0120] Determine whether the intent descriptions of the two outputs belong to the same task sub-goal.

[0121] Integrate the scores of the above three dimensions to generate the overall consistency score of all Agent outputs, ranging from 0 to 1, and the higher the value, the lower the conflict risk.

[0122] Step S1044: A preset conflict classification rule library includes the following rules:

[0123] According to the semantic feature difference, it is divided into data inconsistency, logical inconsistency and goal inconsistency;

[0124] Severity determination: based on the consistency score and the conflict type, set the threshold:

[0125] High severity: consistency score < 0.3, or containing goal inconsistency;

[0126] Medium severity: 0.3 ≤ consistency score < 0.6, and data inconsistency or logical inconsistency;

[0127] Low severity: consistency score ≥ 0.6, or only minor entity value difference.

[0128] Since different Agents use different output formats, a format conversion module is needed to unify them into an intermediate representation, eliminating the interference of format differences on conflict detection. For example, the output of patient data compliance may be in the form of GDPR compliant: yes text or {'gdpr_compliant': True}, and after standardization, both are converted into structured objects containing compliance status, verification basis, etc.

[0129] It should be noted that the inconsistency of the embodiment refers to the difference or contradiction of information from different sources in the semantic, logical or target level in the multi-agent collaboration output or knowledge fusion result. It is an inevitable phenomenon in the multi-source knowledge fusion process and a problem that needs to be monitored and solved by the system. The data inconsistency refers to the ambiguity of the entity definition in the knowledge graph; the logical inconsistency exposes the defects of the Agent behavior pattern or reasoning strategy; and the target inconsistency reflects the deviation of the Agent behavior pattern in the role template library.

[0130] The step S104 unifies the intermediate representation format to solve the comparison difficulty problem caused by the difference of the output formats of different Agents, ensures the accuracy of the conflict detection, extracts the key semantic elements to capture the deep semantics, such as identifying the same entity or intention in different expressions, the multi-dimensional consistency score calculation makes the conflict evaluation more comprehensive, reduces the deviation of single-dimensional judgment, and classifies the conflict type and severity according to the rule library, which provides clear basis for subsequent targeted conflict resolution, facilitates quick positioning of problems and taking measures, and improves the coordination of multi-agent collaboration and the reliability of task processing.

[0131] In an embodiment of the present application, based on step S105, a possible embodiment will be given below to illustrate the specific implementation scheme. The step S105 specifically includes:

[0132] Step S1051: Collect the conflict type, severity, associated Agent ID, and conflict content information output in step S104, convert them into related data, and form an inconsistency event library to facilitate subsequent matching and expansion strategies. The related data includes a conflict ID, a type label, a severity level, an associated subtask ID, and a conflict description field JSON object.

[0133] Step S1052: Based on the structured conflict event of step S1051, call the preset expansion strategy rule library for matching:

[0134] If the conflict type is data inconsistency and the severity level is high, match the vertical knowledge expansion strategy;

[0135] If the conflict type is logical inconsistency and the severity level is medium, match the horizontal Agent expansion strategy;

[0136] If the conflict type is target inconsistency or the new concept recognition frequency exceeds θ3, match the knowledge graph update strategy;

[0137] If multiple types of conflicts are met at the same time, according to the priority, data inconsistency > logical inconsistency > target inconsistency, select the main expansion strategy.

[0138] Step S1053: According to the strategy matched in step S1052, perform vertical and horizontal expansion operations.

[0139] Specifically, vertical knowledge expansion addresses data inconsistency by obtaining the latest definitions or revised rules for conflicting entities from authoritative knowledge sources and adding the new rules to the knowledge graph of the corresponding domain.

[0140] Horizontal agent expansion addresses logical inconsistencies by extracting new skills involved in the conflict from the skill mapping table and combining them with the behavioral patterns of verification agents in the role template library to generate new specialized agent instances.

[0141] Knowledge graph updates involve identifying new concepts, adding the new concepts and their relationships to the knowledge graph, and supplementing related entities.

[0142] Step S1054: Verify the expanded knowledge graph and the newly added Agent instance.

[0143] Specifically, knowledge graph verification involves verifying, through domain experts or historical case libraries, whether the updated knowledge fragments resolve the original conflicts.

[0144] Agent instance verification is a simulation of task scenarios to test the effectiveness of newly added agents' skills;

[0145] After successful verification, the skill mapping table update is pushed to the Agent set in step S102, and the knowledge graph version number and newly added Agent information are sent to the knowledge routing engine in step S103 to ensure that subsequent tasks call the latest resources.

[0146] This embodiment structures conflict information into an event database, facilitating rapid querying and matching of extension strategies, thereby improving the efficiency of conflict handling. Based on a rule base and priority matching of extension strategies, it ensures that the most suitable solution is adopted for different conflicts, avoiding strategy misuse. For example, high-severity data inconsistencies are prioritized for resolution through vertical knowledge expansion. Specific extension operations are highly targeted: vertical expansion supplements knowledge to resolve data inconsistencies, horizontal expansion adds agents to resolve logical inconsistencies, and updating the knowledge graph can address target inconsistencies and new concept issues. The verification process ensures the effectiveness of the expanded resources, preventing invalid expansion from affecting system operation. Pushing update information ensures that relevant modules use the new resources synchronously, maintaining the stability of multi-agent collaboration and improving the continuity and accuracy of task processing.

[0147] In one embodiment of the present invention, based on step S106, the following will provide a possible embodiment and its specific implementation will be described in a non-limiting manner. Step S106 specifically includes:

[0148] Step S1061: When a knowledge base is added, modified, or deleted, the knowledge base pushes a knowledge update event, which includes information such as update type, update content, and update timestamp.

[0149] Step S1062: According to the updated content monitored in step S1061, modify the knowledge graph to ensure the atomicity and traceability of the update operation:

[0150] If the update content is a new concept not included, create a new node in the knowledge graph, and mark the type, definition, source and creation time;

[0151] If the update content is a new association of an existing concept, add a bidirectional edge in the knowledge graph, and mark the association strength and basis;

[0152] If the update content is an attribute adjustment of an existing node, modify the attribute value of the corresponding node and keep the historical version record.

[0153] Step S1063: Based on the updated knowledge graph, recalculate the comprehensive weight of timeliness, relevance and credibility of each knowledge fragment to ensure that the weight reflects the current value of the knowledge:

[0154] Specifically, according to the difference Δt between the last update time of the knowledge fragment and the current time, it is divided into four levels;

[0155] According to the matching degree of the knowledge fragment and the current task field, the score is calculated through the pre-defined field label matching rule;

[0156] According to the source authority and historical verification result of the knowledge fragment, take the weighted average of the two;

[0157] The scores of the above dimensions are weighted and summed according to timeliness α, relevance β and credibility γ to obtain the comprehensive weight W of the knowledge fragment = α ⋅ T + β ⋅ R + γ ⋅ C.

[0158] Step S1064: Synchronize the update content and weight change of the knowledge graph to the Agent set of step S102 and step S103.

[0159] Specifically, use the message queue to send the knowledge graph update event, and the event content includes the knowledge graph version number, the added / modified knowledge fragment ID, the comprehensive weight change value, the update timestamp and other information;

[0160] Each Agent instance in the Agent set listens to the message queue, and after receiving the update event, updates the local knowledge graph cache to ensure that the latest knowledge is used in subsequent task execution;

[0161] After receiving the update event, refresh its knowledge retrieval index and reload the weight of the latest knowledge fragment to ensure that the retrieval result of the new task is based on the latest knowledge.

[0162] It can be seen that the step S106 receives the knowledge base update event in real time, can timely capture the knowledge change, avoids the system from relying on the out-of-date knowledge; the targeted modification of the knowledge graph ensures that the new knowledge is accurately integrated into the existing system, retains the historical version for tracing back, such as creating a node for a newly added concept, adding an associated edge and labeling the basis; the recalculation of the comprehensive weight makes the knowledge value evaluation conform to the current state, ensures that the important knowledge is used preferentially; the information is updated through the message queue, ensures that the Agent set and the knowledge routing engine timely load the latest knowledge and weight, maintains the knowledge consistency of each module of the system, improves the accuracy of multi-Agent cooperation and the reliability of task processing, enables the system to continuously adapt to the changes of the domain knowledge.

[0163] The technical solutions of the present application will be described in detail below with specific embodiments:

[0164] Embodiment 1: Perform medical compliance consultation.

[0165] Step 1: Task input and decomposition User input task: "Develop a personalized diabetes treatment plan for patients in a certain region that meets GDPR requirements".

[0166] Task adaptive decomposer workflow:

[0167] The domain classification module identifies two main domains: "medical diagnosis" and "data compliance";

[0168] The semantic analysis module extracts key entities: [certain region, GDPR, diabetes, personalized treatment];

[0169] The subtask generation module outputs: T1="diabetes treatment plan development", T2="patient data GDPR compliance review".

[0170] Step 2: Agent construction The Agent constructor constructs a domain-aware Agent according to the subtask requirements:

[0171] For T1, map skills: [endocrinology expertise, drug interaction analysis, patient data interpretation], generate "endocrine treatment specialist Agent";

[0172] For T2, map skills: [GDPR clause interpretation, data anonymization technology, cross-border data transmission compliance], generate "data compliance review Agent".

[0173] Step 3: Knowledge retrieval and fusion The multi-level knowledge routing engine executes:

[0174] First layer retrieval: The endocrine treatment specialist Agent retrieves the latest diabetes treatment guidelines from the medical literature database;

[0175] Second layer retrieval: Obtain "certain region medical data processing" historical cases from the collaborative memory bank;

[0176] Knowledge weighted fusion: Calculate the weight of each knowledge source, and give priority to high-weight information to generate a preliminary solution.

[0177] Step 4: Conflict detection and resolution Collaborative conflict intelligent mediator finds that the gene detection data contained in the treatment plan conflicts with the GDPR sensitive data processing requirements.

[0178] Conflict type identification: conflict between data privacy and medical needs;

[0179] Extension: Generate a "medical data ethics coordination agent";

[0180] Conflict resolution: Redesign the data processing process and use differential privacy technology to process genetic data.

[0181] Embodiment 2: System adaptive optimization method.

[0182] When the system detects the release of new medical guidelines during operation, the extension module:

[0183] Update the medical knowledge graph and add new diagnosis and treatment nodes;

[0184] Recalculate the timeliness weight of related knowledge;

[0185] Notify the relevant Agent to reload the updated knowledge weight.

[0186] As can be seen, through the task-driven Agent generation mechanism, the system can construct a specialized Agent combination according to the characteristics of the task, avoiding the problem of role solidification in traditional systems. Through real-time monitoring and extension, the system can identify and solve the collaboration conflicts between Agents, improving the stability and reliability of the system. With the functions of knowledge base updating and Agent capability expansion, the system maintains continuous optimization and adaptability.

[0187] It should be understood that the size of the serial number of each step in the above embodiments does not mean the order of execution, and the execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0188] The following is an embodiment of a multi-field task processing system based on knowledge fusion and Agent collaboration provided by the embodiments of the present disclosure. The system and the above-mentioned multi-field task processing method based on knowledge fusion and Agent collaboration belong to the same inventive concept. Details not described in the embodiment of the multi-field task processing system based on knowledge fusion and Agent collaboration can be referred to the above-mentioned embodiment of the multi-field task processing method based on knowledge fusion and Agent collaboration.

[0189] AsFigure 2 As shown, the system includes:

[0190] The task identification module 201 is used to perform domain identification and subtask decomposition on the task input by the user, identify the semantic elements involved in the task, and generate a set of subtasks.

[0191] The instantiation module 202 is used to obtain the correspondence between the professional skills and tool interfaces required by the subtasks by calling the skill mapping table based on the subtask set, and generate professional agent instances by combining the agent behavior patterns and inference strategies in the role template library to form an agent set.

[0192] The cross-domain fusion module 203 is used to perform cross-domain knowledge retrieval and fusion for a set of Agents.

[0193] The collaboration monitoring module 204 monitors the output consistency of the Agent set based on the results of multi-source knowledge fusion, analyzes the semantic consistency of the output results of different Agents in real time, and classifies the types of collaboration conflicts according to the semantic consistency analysis results.

[0194] The extended trigger module 205 is used to trigger vertical knowledge expansion to update the knowledge graph of the corresponding domain when a collaboration conflict is detected and the confidence level of the Agent output result is lower than the first preset threshold.

[0195] When the collaboration conflict rate exceeds the second preset threshold, horizontal agent expansion is triggered to generate new specialized agent instances.

[0196] When the frequency of new concept recognition exceeds the third preset threshold, the knowledge graph is updated to supplement the new concepts and their related relationships.

[0197] The knowledge graph continuous update module 206 is used to monitor new domain knowledge updates, update the knowledge graph of the corresponding domain, calculate the comprehensive weight of the timeliness, relevance and credibility of each knowledge fragment in the knowledge graph, and reload the updated knowledge weights and knowledge graph information.

[0198] like Figure 3 As shown, this application also provides an electronic device, including a display module 103, a memory 102, a processor 101, and a computer program stored in the memory and executable on the processor 101. When the processor 101 executes the program, it implements the steps of the method for parsing and loading a GIM model of a power transmission project.

[0199] In embodiments of the application, electronic devices include, but are not limited to, laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other suitable computers. Electronic devices can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices, and other similar computing devices. The components, their connections, and relationships, and their functions, as described herein, are meant to be examples only, and are not intended to limit implementations of the present application described and / or claimed herein.

[0200] In embodiments of the present application, the processor 101 can be implemented by using at least one of application specific integrated circuits, programmable logic devices, field programmable gate arrays, processors, controllers, micro-controllers, microprocessors, electronic units designed to perform the functions described herein, and in some cases such an implementation can be implemented in a controller. For software implementation, embodiments of the processes or functions can be implemented with separate software modules, which allow at least one function or operation to be performed, by a software application (or program) written in any suitable programming language to be executed by a controller. Software code can be implemented by the software application (or program) written in any suitable programming language to be executed by a controller, and can be stored in the memory.

[0201] The display module 103 is configured to display information input by a user or information provided to the user. The display module 103 can include a display panel, which can be configured in the form of a liquid crystal display, an organic light emitting diode, or the like.

[0202] The memory 102 can be used to store software programs as well as various data. The memory 102 can include a high-speed random access memory, and can also include a non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other volatile solid-state memory device.

[0203] The present application also provides a storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the steps of the multi-field task processing method based on knowledge fusion and Agent cooperation.

[0204] The storage medium can take any combination of one or more of a readable medium. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium may, for example, be but is not limited to an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or apparatus, or any suitable combination of the above. More specific examples (non-exhaustive list) of readable storage medium include an electrical connection having one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0205] The foregoing description of the disclosed embodiments enables a person skilled in the art to make or use the application. Numerous modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein can be applied to other embodiments without the use of the inventive faculty. Therefore, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A multi-domain task processing method based on knowledge fusion and agent collaboration, characterized in that, The methods include: S101: Perform domain identification and subtask decomposition on the user-input task, identify the semantic elements involved in the task, and generate a set of subtasks; S102: Based on the set of subtasks, call the skill mapping table to obtain the correspondence between the professional skills and tool interfaces required by the subtasks, and combine the Agent behavior patterns and reasoning strategies in the role template library to generate professional Agent instances and form an Agent set; Step S102 specifically includes: For each subtask in the subtask set generated in step S101, the text description is parsed using natural language processing technology to extract the subtask's domain labels, functional requirements, and input / output requirements. Configure the association between domain tags and professional skills and tool interfaces in the skill mapping table. By querying the skill mapping table, obtain the set of professional skills and the list of tool interfaces that perfectly match the domain tags of the subtask. In the role template library, the behavioral patterns and inference strategies of the Agent are configured according to the domain. Based on the functional requirements of the sub-task, the closest combination of behavioral patterns and inference strategies is selected through semantic matching algorithm. The acquired skill sets, tool interfaces, and filtered behavioral patterns and inference strategies are structurally encapsulated, and an independent Agent instance is generated for each subtask through an instantiation engine; S103: Perform cross-domain knowledge retrieval and fusion for the Agent set; Step S103 specifically includes: The first-layer domain knowledge retrieval results and the second-layer collaborative memory retrieval results are collected respectively, and the timeliness marker, relevance tag, and credibility identifier of each knowledge are extracted. Based on preset mapping rules, timeliness markers are converted into timeliness quantification values; relevance tags are converted into relevance quantification values; and credibility identifiers are converted into credibility quantification values. For each piece of knowledge, the fusion weight of a single piece of knowledge is calculated according to the method of timeliness quantification value × α + relevance quantification value × β + credibility quantification value × γ, and the weight result is marked; α, β, and γ are predefined weight allocation coefficients, and α + β + γ = 1; Sort the knowledge fusion weights from high to low, select the knowledge whose weights reach the threshold, aggregate them according to the sub-task logic, and generate the knowledge fusion result; S104: Based on the results of multi-source knowledge fusion, monitor the output consistency of the Agent set, analyze the semantic consistency of the output results of different Agents in real time, and classify the types of collaboration conflicts according to the semantic consistency analysis results; Step S104 specifically includes: Step S1041: Collect the output results of each specialized Agent instance in step S103 and convert them into a unified intermediate representation format; Step S1042: For each standardized output content, extract key semantic elements using natural language processing techniques; Step S1043: Based on the semantic vector generated in step S1042, the consistency of outputs from different agents under the same task is calculated using the following method: Calculate the cosine similarity of the semantic vectors output by any two agents; Calculate the proportion of exact matches of key entities in the two outputs to the total number of entities; Determine whether the intent descriptions of the two outputs belong to the same task sub-goal; By combining the scores from the three dimensions, an overall consistency score is generated for all Agent outputs, ranging from 0 to 1, with higher values ​​indicating lower conflict risk. Step S1044: Preset conflict classification rule base, including the following rules: Based on semantic feature differences, they are divided into data inconsistency, logical inconsistency, and goal inconsistency. Severity determination: Based on consistency score and conflict type, a threshold is set: High severity: Consistency score < 0.3, or contains goal inconsistency; Medium severity: 0.3 ≤ consistency score < 0.6, and it is either data inconsistency or logical inconsistency; Low severity: Consistency score ≥ 0.6, or only minor differences in entity values; S105: If a collaboration conflict is detected and the confidence level of the Agent output result is lower than the first preset threshold, vertical knowledge expansion is triggered to update the knowledge graph of the corresponding domain. When the collaboration conflict rate exceeds the second preset threshold, horizontal agent expansion is triggered to generate new specialized agent instances; When the frequency of new concept recognition exceeds the third preset threshold, the knowledge graph is updated to supplement new concepts and their related relationships. S106: Monitor new domain knowledge updates, update the knowledge graph of the corresponding domain, calculate the comprehensive weight of the timeliness, relevance and credibility of each knowledge fragment in the knowledge graph, and reload the updated knowledge weights and knowledge graph information.

2. The multi-domain task processing method based on knowledge fusion and agent collaboration according to claim 1, characterized in that, The execution of cross-domain knowledge retrieval and fusion in step S103 includes: a first-layer domain knowledge retrieval engine retrieves domain-specific information from a professional knowledge base based on the agent's specialized skill requirements; a second-layer collaborative memory retrieval engine retrieves solutions to similar tasks from a historical collaborative case library; and a knowledge weighted fusion engine uses a comprehensive weighting algorithm based on timeliness, relevance, and credibility to fuse the retrieval results from the two layers, generating a multi-source knowledge fusion result.

3. The multi-domain task processing method based on knowledge fusion and agent collaboration according to claim 1, characterized in that, Step S105 specifically includes: Collect the conflict type, severity, associated AgentID, and conflict content information output in step S104, convert them into relevant data, and form an inconsistency event database; Based on the conflict event, the preset extended policy rule base is invoked for matching: If the conflict type is data inconsistency and the severity level is high, match the vertical knowledge expansion strategy; If the conflict type is logical inconsistency and the severity level is medium, match the horizontal agent expansion strategy; If the conflict type is goal inconsistency or the frequency of new concept recognition exceeds θ3, match the knowledge graph update strategy; If multiple types of conflicts are met simultaneously, the main expansion strategy is selected in descending order of priority: data inconsistency, logical inconsistency, and goal inconsistency. Based on the matching strategy, perform vertical and horizontal expansion operations; verify the expanded knowledge graph and the newly added Agent instances.

4. The multi-domain task processing method based on knowledge fusion and agent collaboration according to claim 1, characterized in that, Step S106 specifically includes: When a knowledge base is updated, modified, or deleted, the knowledge base pushes a knowledge update event. Based on the monitored updates, the knowledge graph is modified to ensure the atomicity and traceability of the update operations: If the updated content is a new concept that has not been included in the knowledge graph, a new node will be created in the knowledge graph, and the type, definition, source and creation time will be labeled. If the updated content is a new association of an existing concept, then add a bidirectional edge to the knowledge graph and label the association strength and basis. If the update involves adjusting the attributes of an existing node, then modify the attribute values ​​of the corresponding node and retain the historical version record; Based on the updated knowledge graph, the comprehensive weights of timeliness, relevance, and credibility are recalculated for each knowledge fragment; The updated content and weight changes of the knowledge graph are synchronized to the Agent set in step S102 and step S103.

5. A multi-domain task processing system based on knowledge fusion and agent collaboration, characterized in that, The system is used to implement the multi-domain task processing method based on knowledge fusion and agent collaboration as described in any one of claims 1 to 4; The system includes: The task recognition module is used to perform domain recognition and subtask decomposition on the task input by the user, identify the semantic elements involved in the task, and generate a set of subtasks. The instantiation module is used to obtain the correspondence between the professional skills and tool interfaces required by the subtasks by calling the skill mapping table based on the subtask set, and generate professional agent instances by combining the agent behavior patterns and inference strategies in the role template library to form an agent set. The cross-domain fusion module is used to perform cross-domain knowledge retrieval and fusion for a set of Agents; The collaboration monitoring module monitors the output consistency of the Agent set based on the results of multi-source knowledge fusion, analyzes the semantic consistency of the output results of different Agents in real time, and classifies the types of collaboration conflicts according to the semantic consistency analysis results. An extended triggering module is used to trigger vertical knowledge expansion to update the knowledge graph of the corresponding domain when a collaboration conflict is detected and the confidence level of the Agent output result is lower than the first preset threshold. When the collaboration conflict rate exceeds the second preset threshold, horizontal agent expansion is triggered to generate new specialized agent instances; When the frequency of new concept recognition exceeds the third preset threshold, the knowledge graph is updated to supplement new concepts and their related relationships. The knowledge graph continuous update module is used to monitor new domain knowledge updates, update the corresponding domain knowledge graph, calculate the comprehensive weight of the timeliness, relevance and credibility of each knowledge fragment in the knowledge graph, and reload the updated knowledge weights and knowledge graph information.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the multi-domain task processing method based on knowledge fusion and agent collaboration as described in any one of claims 1 to 4.

7. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the multi-domain task processing method based on knowledge fusion and agent collaboration as described in any one of claims 1 to 4.

Images