Multi-system cooperative task response method and device, equipment and storage medium

Abstract

This application discloses a multi-system collaborative task response method, apparatus, device, and storage medium. The method includes: receiving a current task instruction and determining the current operation intent corresponding to the current task instruction; matching the target interface path of the target business system required by the current operation intent in a preset model context protocol registry center, wherein the preset model context protocol registry center stores interface paths of different business systems; determining the interface input parameters according to the current operation intent, and calling the target business interface in the target business system according to the target interface path based on the interface input parameters to obtain the task response result.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a multi-system collaborative task response method, apparatus, device and storage medium. Background Technology

[0002] As enterprises continue to advance their digital transformation, various business systems, such as Enterprise Resource Planning (ERP), Customer Relationship Management (CRM), and Human Resource Management (HR), are widely used within enterprises. These systems are typically provided by different vendors, employing heterogeneous technical architectures and interface standards, each supporting specific business processes. Due to the lack of unified interaction standards between systems, data cannot flow efficiently between different systems, resulting in information silos.

[0003] Meanwhile, the development of artificial intelligence (AI) technology, especially large language models, has provided new technological paths for enterprises to improve customer service efficiency. More and more enterprises are introducing AI systems for task response to optimize user experience. However, due to the aforementioned information silo problem, AI systems cannot directly connect to the real-time data sources of various business systems. They usually obtain task-related data indirectly: data from different business systems is periodically exported, cleaned, and integrated into a pre-built knowledge base, and the AI ​​system responds to tasks based on the data in this knowledge base.

[0004] However, due to the continuous dynamic changes in data across various business systems, the synchronization cycle of the knowledge base causes its content to lag behind the actual business status. When responding to tasks, the AI ​​system relies on outdated knowledge base data, which in turn leads to insufficient reliability and accuracy in the final generated response results. Summary of the Invention

[0005] The main purpose of this application is to provide a multi-system collaborative task response method, apparatus, device and storage medium, which aims to solve the technical problem that the response results obtained by existing AI systems based on pre-built knowledge bases have insufficient reliability and accuracy.

[0006] To achieve the above objectives, this application proposes a multi-system collaborative task response method, which includes: Receive the current task instruction and determine the current operation intent corresponding to the current task instruction; Based on the current operation intent, the target interface path of the target business system required by the current operation intent is matched in the preset model context protocol registry center. The preset model context protocol registry center stores the interface paths of different business systems. Determine the interface input parameters based on the current operation intent, and call the target business interface in the target business system according to the target interface path based on the interface input parameters to obtain the task response result.

[0007] Furthermore, to achieve the above objectives, this application also proposes a multi-system collaborative task response device, which includes: The intent recognition module is used to receive the current task instruction and determine the current operation intent corresponding to the current task instruction; The interface matching module is used to match the target interface path of the target business system required by the current operation intent in the preset model context protocol registry center according to the current operation intent. The preset model context protocol registry center stores the interface paths of different business systems. The task execution module is used to determine the interface input parameters based on the current operation intention, and call the target business interface in the target business system according to the target interface path based on the interface input parameters to obtain the task response result.

[0008] In addition, to achieve the above objectives, this application also proposes a multi-system collaborative task response device, which includes: a memory, a processor, and a multi-system collaborative task response program stored in the memory and executable on the processor. The multi-system collaborative task response program is configured to implement the steps of the multi-system collaborative task response method described above.

[0009] In addition, to achieve the above objectives, this application also proposes a storage medium that is a computer-readable storage medium, on which a multi-system collaborative task response program is stored, which, when executed by a processor, implements the steps of the multi-system collaborative task response method described above. Attached Figure Description

[0010] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0011] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 This is a flowchart illustrating the first embodiment of the multi-system collaborative task response method of this application; Figure 2 This is a schematic diagram of the operation intent recognition process in this application; Figure 3This is a flowchart illustrating the second embodiment of the multi-system collaborative task response method of this application; Figure 4 This is a schematic diagram illustrating the calling process of the target business interface in this application; Figure 5 This is a flowchart illustrating the third embodiment of the multi-system collaborative task response method of this application; Figure 6 This is a schematic diagram illustrating the entire process of the multi-system collaborative task response method of this application; Figure 7 This is a schematic diagram of the modular structure of the multi-system collaborative task response device of this application; Figure 8 This is a schematic diagram of the multi-system collaborative task response device of this application.

[0013] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0014] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0015] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0016] The main solution of this application embodiment is: receiving the current task instruction and determining the current operation intention corresponding to the current task instruction; matching the target interface path of the target business system required by the current operation intention in the preset model context protocol registry center, the preset model context protocol registry center stores the interface paths of different business systems; determining the interface input parameters according to the current operation intention, and calling the target business interface in the target business system according to the target interface path based on the interface input parameters to obtain the task response result.

[0017] The task response process of an AI system requires it to retrieve relevant data from multiple different business systems. However, due to the lack of standardized interfaces among these business systems, the AI ​​system struggles to interact efficiently with them, resulting in a data silo effect. Current methods typically involve periodically exporting, cleaning, and integrating data from different business systems into a pre-built knowledge base, allowing the AI ​​system to retrieve relevant data only from this knowledge base during task response. However, because the knowledge base is not synchronized with the various business systems in real time—meaning updates to the knowledge base often lag behind changes in the business systems—the timeliness of the data retrieved from the knowledge base during task response is insufficient, thus affecting the reliability and accuracy of the final response result.

[0018] This application's method can dynamically match operational intents with the interface paths of target business systems based on a pre-defined model context protocol registry center. It eliminates the need to pre-build a unified knowledge base for different business systems, avoiding data lag issues caused by periodic data export, cleaning, and integration processes. Furthermore, during task response, data is obtained by directly calling the real-time interface of the target business system, ensuring that the acquired business data remains consistent with the current state of the business system. This effectively improves the timeliness of the data used for task response, thereby ensuring the reliability and accuracy of the obtained task response results.

[0019] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, a mainframe computer, a server cluster, etc., or a system server of a task response system, which may integrate AI technology. The following description uses the system server of the task response system (hereinafter referred to as the "AI system") as an example to illustrate this embodiment and the subsequent embodiments.

[0020] Based on this, this application proposes a multi-system collaborative task response method, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the multi-system collaborative task response method of this application. In this embodiment, the multi-system collaborative task response method includes steps S10 to S30: Step S10: Receive the current task instruction and determine the current operation intent corresponding to the current task instruction.

[0021] It should be noted that the current task instruction can be a natural language description issued by the user, requesting the AI ​​system to complete a certain task. It can contain a clear operational intent, and therefore the task can be further broken down into several sub-tasks with operational intents.

[0022] For example, the current task instruction could be: "Help me create a Jira task for technical research, assign it to myself, with a task timeframe from January 1, 2026 to February 1, 2026. After creating the Jira task, create the research document in the wiki system and update the description in the Jira task, including the address of the research document in the wiki system in the description."

[0023] It should be understood that AI systems can integrate open-source, user-interactive Large Language Models (LLMs) as the foundational model. Therefore, AI systems can provide user-facing interfaces that guide users to input current task commands within the interface's dialog box.

[0024] When the basic model receives the current task instruction, it first performs text cleaning and standardization to obtain a clean and standardized task instruction text. Text cleaning involves removing noise from the original text. Since user-input natural language instructions may contain various irrelevant content, text cleaning can include: removing meaningless symbols, error correction, removing interjections, and standardizing formatting. Standardization involves converting the text into a unified and standardized form to ensure that the same meaning expressed in different ways can be consistently recognized. Standardization can include: standardizing date formats, standardizing system names, normalizing synonyms, and standardizing terminology.

[0025] Specifically, the basic model first identifies and removes meaningless symbols, interjections, and emoticons from the current task instructions; then, it standardizes the dates in the current task instructions to the same format, for example, unifying February 1, 2026 as 2026-02-01; it standardizes the system names, for example, unifying the wiki system as Wiki system with the first letter capitalized; it then injects the user's context role and historical session state, for example, the current role is the user who initiated the instruction, and there is no historical session state; finally, it identifies domain keywords, such as the business system names Jira, Wiki, etc.

[0026] It should be understood that the task instruction text can be clean, standardized text content obtained after cleaning and standardization, which is different from the original current task instruction. The task instruction text has removed noisy text and standardized the format, which is beneficial for subsequent intent recognition.

[0027] For example, if the current task instruction is "Help me create a Jira task from January 1st to 2nd, 2026", then the task instruction text obtained after the above text cleaning and standardization process can be "Create Jira task, time: 2026-01-01 to 2026-02-01".

[0028] It should also be noted that the preset intent recognition model can also be an LLM integrated into the AI ​​system. However, unlike the aforementioned basic model, this preset intent recognition model can utilize open-source LLM networks, such as the Bidirectional Encoder Representation (BERT) model, which have superior performance in semantic text understanding tasks, and is trained using a large amount of enterprise corpus data (natural language samples and intent category labels). This ensures that the obtained preset intent recognition model can identify the user's current operational intent from natural language text.

[0029] The current operation intent is one or more sub-tasks that the user wants to perform, which are identified from the task instruction text by the preset intent recognition model. Therefore, the current operation intent can contain multiple intent fragments.

[0030] In the specific implementation, the pre-defined intent recognition model receives task instruction text as input. It first performs a macroscopic understanding of the entire text to identify which intent category it belongs to. The intent category can be a coarse-grained classification of user needs, such as "task creation", "information query", "data update", "approval operation", etc.

[0031] Next, the pre-defined intent recognition model identifies natural segmentation boundaries from the task instruction text based on pre-set segmentation words, such as conjunctions, temporal words, or logical relational words like "then," "after," "and," "simultaneously," and "after completion," to divide the continuous text into multiple relatively semantically independent segments. For example, "Help me create a Jira task, then create a Wiki document, and then update the Wiki address in the Jira task" can be segmented into: segment 1: "Create a Jira task"; segment 2: "Create a Wiki document"; segment 3: "Update the Wiki address in the Jira task".

[0032] Finally, the pre-defined intent recognition model binds the aforementioned intent categories to each segment, thereby ensuring that each segment corresponds to a clear intent label, resulting in several intent segments. Each intent segment contains two parts: first, the specific text content (e.g., "create a Jira task"), and second, the intent category corresponding to that text content (e.g., "create task"). The current operation intent is then a set composed of the aforementioned intent segments, with each segment representing a subtask to be executed.

[0033] Furthermore, this can be referenced here. Figure 2 This paper describes the operation intent recognition process implemented by the AI ​​system of this application based on a preset intent recognition model. Figure 2 This is a schematic diagram of the operation intent recognition process in this application.

[0034] Depend on Figure 2 It can be seen that the AI ​​system first receives the current task instructions described by the user in natural language based on the interactive interface; Next, the AI ​​system, based on a fundamental large language model, performs text cleaning and standardization operations on the current task instruction. This includes: removing meaningless symbols, interjections, and emoticons from the current task instruction; standardizing the date format and system name; injecting user context roles; recognizing historical conversation states; and enhancing domain keywords. The result is a clean, standardized text content—the task instruction text—that has undergone cleaning and standardization.

[0035] Finally, the AI ​​system can call the built-in preset intent recognition model. This preset intent recognition model first receives the task instruction text as input, identifies one or more corresponding intent categories, then segments the task instruction text according to preset segmentation words, and then binds the aforementioned intent categories with the fragments obtained from each segment to obtain several intent fragments, which can then be used as the current operation intent corresponding to the current task instruction.

[0036] Step S20: Match the target interface path of the target business system required by the current operation intent in the preset model context protocol registry center. The preset model context protocol registry center stores the interface paths of different business systems.

[0037] It should be noted that the default Model Context Protocol (MCP) registry can be a registry pre-deployed in the AI ​​system to store and manage all business systems associated with the AI ​​system.

[0038] The MCP protocol is an open standard protocol that enables LLM integration with external data sources, tools, and services. Each business system can provide several business interfaces, each with corresponding interface metadata. This metadata may include: interface path, function description, request parameter definitions, and permission requirements. Therefore, administrators can pre-write the interface metadata of each business system into the MCP registry center based on the requirements of the MCP protocol, thereby completing the registration of the business systems (business interfaces).

[0039] It should be understood that the AI ​​system can match the target business system in the MCP registry center with the current operational intent as determined above. For example, if the current operational intent is "create a Jira task", then the matched target business system should be the task creation interface in the Jira system.

[0040] Furthermore, since the current operation intent may include several intent fragments, the number of target business systems matched by the current operation intent can also be one or more. To match the current operation intent to the target business system, the AI ​​system can also integrate a preset business matching model. To specifically illustrate how to perform target business interface matching based on the preset business matching model, step S20 specifically includes: steps S201~S203: Step S201: Obtain the business interface list of each business system from the preset model context protocol registry center. The business interface list includes several business interfaces and corresponding interface metadata. Each interface metadata is pre-configured by the management user and includes at least the interface path and function description.

[0041] It should be noted that the MCP registry center can maintain different business interfaces for different business systems in the form of a business interface list: each business system corresponds to a business interface list, and each business interface list contains all the business interfaces under that business system and their corresponding interface metadata. Each piece of interface metadata includes the interface path, function description, request parameter definition, and permission requirements corresponding to that business interface.

[0042] The interface path can be the URL address required to call the business interface; the function description can be the function that the interface can perform, described in natural language, which can be used as the interface feature for subsequent intent matching; the request parameter definition can be used to check whether the user has provided required parameters during subsequent parameter validation; and the permission requirements can be used to implement security control for subsequent interface calls.

[0043] Step S202: Input the current operation intent and metadata of each interface into the preset business matching model to obtain the target business interface. The preset business matching model is used to match the functional description of each business interface with the current operation intent, and determine the business interface with the highest matching degree as the target business interface.

[0044] It should be noted that this preset business matching model can also be an LLM integrated into the AI ​​system. Unlike the aforementioned basic model, this preset business matching model can also use an open-source LLM network that performs well in semantic text understanding tasks, and is trained using a large amount of enterprise corpus data (intent fragment samples and corresponding business interface labels). This ensures that the obtained preset business matching model can match the most suitable target business interface from several business interfaces in the MCP registry center.

[0045] Furthermore, since both the preset business matching big model and the aforementioned intent recognition big model can use LLMs that perform well in semantic text understanding tasks, the two big models can use the same open-source LLM or different open-source LLMs. This embodiment does not impose any restrictions on this.

[0046] In its implementation, the AI ​​system can input the current operational intent (containing several intent fragments) and the interface metadata of each business interface into a pre-defined business matching model. The model then extracts the functional description fields of each business interface from the interface metadata and calculates the semantic similarity between each functional description field and each intent fragment. Finally, the business interface with the highest similarity score to each intent fragment is identified as the target business interface. Since each intent fragment corresponds to a business interface with the highest similarity score, the number of target business interfaces can also be one or more.

[0047] For example, when the intent fragment is "create a Jira task", the business interface with the highest similarity score can be the Jira system's task creation interface; when the intent fragment is "query Wiki documents", the business interface with the highest similarity score can be the Wiki system's document query interface.

[0048] Step S203: Determine the interface path corresponding to the target business interface as the target interface path of the target business system required by the current operation intent.

[0049] It should be noted that after the target business interface is determined, the AI ​​system can directly extract the interface path from the interface metadata corresponding to the target business interface, and use it as the target interface path for subsequent execution of the current operation intent.

[0050] Step S30: Determine the interface input parameters according to the current operation intention, and call the target business interface in the target business system according to the target interface path based on the interface input parameters to obtain the task response result.

[0051] It should be noted that the AI ​​system can determine the parameters that need to be passed to the target interface to be called, i.e., the interface input parameters, based on the recognized current operation intention. These interface input parameters may include parameter values ​​directly extracted from the text content of the current operation intention, field values ​​mapped from colloquial expressions, and automatically generated default parameter values.

[0052] Specifically, the AI ​​system can first perform parameter extraction: directly extract parameter values ​​from the text content of the current operation intent, such as extracting "technical research" from "the task title is technical research"; then perform parameter mapping: map colloquial expressions to field values ​​that the system can recognize, such as mapping "assign to myself" to the user's login ID in the AI ​​system; finally, perform parameter completion: automatically generate default values ​​for required parameters in the request parameter definition of the business interface that the user has not provided, such as the creation time being set to the current time.

[0053] It should also be noted that the interface input parameters can be represented as a set of key-value pairs, where the key is the field name required in the request parameter definition of the business interface, and the value is the specific data obtained from the text content of the current operation intent or the dialog context.

[0054] It should be understood that after determining the interface input parameters, the AI ​​system can initiate a request to retrieve the target business interface based on the target interface path, send the interface input parameters to the target business system to which the target business interface belongs, and wait for the business interface to respond.

[0055] After the API call is completed, the AI ​​system will receive API response data from the target business system. This response data may include: operation confirmation information, generated data identifiers, queried data content, and error information. This API response data can then be directly displayed to the user as the task response result.

[0056] This embodiment enables dynamic matching of operational intents with target business system interface paths based on a pre-defined model context protocol registry center. This eliminates the need to pre-build a unified knowledge base for different business systems, avoiding data lag issues caused by periodic data export, cleaning, and integration processes. Furthermore, during task response, data is obtained by directly calling the target business system's real-time interface, ensuring consistency between the acquired business data and the current state of the business system. This effectively improves the timeliness of the data used for task response, thereby ensuring the reliability and accuracy of the obtained task response results. Specifically, by introducing a large model to semantically match operational intents with interface function descriptions, intelligent dynamic mapping between operational intents and business interfaces is achieved. There is no need to pre-define fixed trigger keywords for each business interface. When a business interface changes or is added in the business system, only the function description in the interface metadata needs to be updated; no modification to the matching logic is required, effectively improving the flexibility and scalability of system integration.

[0057] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 3 , Figure 3 This is a flowchart illustrating the second embodiment of the multi-system collaborative task response method of this application.

[0058] In this embodiment, considering that multiple target service interfaces need to be called when the current operation intent contains multiple intent fragments, in order to ensure the sequential calling of multiple target service interfaces and retain the control capability of manual intervention, step S30 further includes: steps S301~S305: Step S301: Extract the corresponding interface input parameters from each intent segment, and determine the parameter dependency relationship between each intent segment based on each interface input parameter.

[0059] It should be noted that the AI ​​system can determine the corresponding target business interface based on the target interface path matched by each intent fragment, and then obtain the corresponding request parameter definition from the interface metadata of the target business interface.

[0060] For example, for the Jira task creation interface, the request parameter definition may include: Title: string, required, maximum length 255; Description: string, optional; Assignor: string, optional, valid AI system login ID; Deadline: date format YYYY-MM-DD, optional.

[0061] Specifically, the AI ​​system can first extract the corresponding interface parameters from each intent fragment, and then determine whether the extracted interface parameters meet the request parameter definitions of their respective target business interfaces, including mandatory field checks, data type checks, format checks, and value range checks; then, when each interface parameter meets the corresponding request parameter definition, the system determines the parameter dependency relationship between each intent fragment based on each interface parameter.

[0062] It should be understood that parameter dependencies can be relationships that exist between different intent segments, where the output of one segment serves as the input of another. By comparing the interface input and output parameters of each intent segment, the AI ​​system can determine the parameter dependencies between them.

[0063] Parameter dependencies can include: output dependencies, where the execution result of fragment A (such as a Jira task ID) is an input parameter of fragment B (such as when creating a Wiki document, this ID needs to be referenced); sequence dependencies, where fragment B must be executed after fragment A; and condition dependencies, where the execution of fragment C depends on the execution result of fragment A or B satisfying a specific condition.

[0064] For example, consider fragment 1: "Create a Jira task"; fragment 2: "Create a Wiki document"; fragment 3: "Update the Wiki address in the Jira task". The corresponding parameter dependency relationship can be fragment 1-fragment 2-fragment 3, that is, the input parameters of the interface corresponding to fragment 3 depend on the output parameters of fragment 2, and the input parameters of the interface corresponding to fragment 2 depend on the output parameters of fragment 1.

[0065] Step S302: Generate a task flowchart based on parameter dependencies and the target interface path matched by each intent fragment.

[0066] It should be noted that a task graph can be a structured task execution plan, which can contain several task nodes, node information, and edges. Each task node corresponds to an intent fragment, i.e., its corresponding execution step. The node information includes the target business interface called by the execution step, the interface input parameters, and the expected output parameters, all marked on the task node. The edges are the connecting lines between task nodes, used to represent parameter dependencies and the execution order of each task node.

[0067] Based on fragments 1, 2, and 3 in the above example, the generated task flowchart can be represented as follows: Node 1: Call the Jira task creation interface - output task ID; Node 2: Call the Wiki document creation interface (depending on task ID) - output document URL; Node 3: Call the Jira task update interface (depending on document URL) - output update status.

[0068] Furthermore, to prevent the existence of unexecutable loop processes, a circular dependency detection can be performed before generating the task flowchart. Therefore, step S302 specifically includes: detecting whether there is a circular dependency in the parameter dependency relationship and obtaining the detection result; when the detection result is that there is no circular dependency, determining the task execution order based on the parameter dependency relationship, and binding the task execution order with the target interface path corresponding to each intent fragment to obtain the task flowchart.

[0069] It's important to note that circular dependencies occur when two or more intent fragments depend on each other, forming a closed loop, making it impossible to determine the order of execution. For example, fragment A requires the output of fragment B, fragment B requires the output of fragment C, and fragment C requires the output of fragment A. In this case, any execution order cannot satisfy the parameter dependency relationship, and the AI ​​system can generate an error message and prompt the user to correct it by updating the current task instructions.

[0070] If there are no circular dependencies, it means that at least one execution order satisfies the above parameter dependencies, and the AI ​​system can directly generate the task execution order based on the above parameter dependencies. If there are multiple feasible task execution orders, the AI ​​system can determine the optimal task execution order based on minimizing the total energy consumption of calling each target business interface, or it can determine it based on other optimization objectives. This embodiment does not impose any restrictions on this.

[0071] Finally, the task execution order is associated with the target interface path corresponding to the aforementioned intent fragment, thereby generating the aforementioned task flowchart composed of task nodes, which includes the execution order (execution steps) and the target interface path (target business interface).

[0072] Step S303: Display the task flowchart to the user and receive feedback operation instructions from the user on the current task flowchart.

[0073] Step S304: Update the task flowchart according to the feedback operation instructions, and determine the current execution flow based on the update results.

[0074] It should be understood that the AI ​​system can also use the aforementioned interactive interface to display the generated task flowchart to the user through a combination of nodes and arrows.

[0075] The interactive interface containing the task flowchart can also provide different feedback operation buttons, such as editing operation buttons for modifying parameters, adjusting order, deleting or removing nodes, setting conditions, etc., as well as confirmation operation buttons. Feedback operation commands can include editing commands and confirmation commands.

[0076] After viewing the task flowchart, users can modify the interface parameters of a task node, adjust the order of execution steps of different task nodes, add or delete a task node, and add execution conditions to the execution steps of a task node, thereby generating corresponding editing operation instructions.

[0077] In practice, the AI ​​system updates the task flowchart based on the user's editing instructions, generates a new version of the task flowchart, and re-displays it on the interactive interface until it receives the user's confirmation instruction, at which point the latest version of the task flowchart is used to determine the current execution flow.

[0078] Step S305: According to the current execution flow, call the target business interfaces in the target business system sequentially based on the input parameters of each interface and the target interface path to obtain the task response results.

[0079] In its implementation, the AI ​​system executes the aforementioned current execution process based on the scheduling function built into the basic large model: following the order of the current execution process, different target business interfaces are called sequentially, and corresponding interface parameters are injected to obtain the interface response data returned by each target business interface; finally, the interface response data are integrated to obtain the task response result.

[0080] Furthermore, this can be referenced here. Figure 4 This application describes the process by which the AI ​​system calls the target business interface to obtain the task response result. Figure 4 This is a schematic diagram illustrating the calling process of the target business interface in this application.

[0081] Depend on Figure 4 As can be seen, the above calling process includes: the flowchart generation stage and the interface calling stage.

[0082] In the flowchart generation phase, after determining the target business interface that matches the current operation intent (which includes several intent fragments), the corresponding interface input parameters can be extracted from each intent fragment and the parameter dependency relationship between each intent fragment can be identified. Then, the task flowchart is initialized based on the parameter dependency relationship. Next, circular dependency detection can be performed on parameter dependencies. If a circular dependency is detected, an error will be generated and the user will be prompted to correct it by updating the current task command. If no circular dependencies are detected, the final task graph can be determined. This task graph includes several task nodes (corresponding intent fragments), node information (target business interface, interface input parameters and expected output parameters), and edges (parameter dependencies).

[0083] Finally, the task graph can be visualized in the AI ​​system's interactive interface.

[0084] During the API call phase, the AI ​​system can first receive confirmation or editing commands triggered by the user through the interactive interface. If it is an editing command, it can perform editing operations such as modifying parameters, adjusting order, adding or deleting nodes, and setting conditions to update the task flowchart. If it is a confirmation command, it can determine the current execution flow based on the latest version of the task flowchart.

[0085] Next, the AI ​​system can sequentially call different target business interfaces according to the current execution flow and inject the corresponding interface parameters to obtain the interface response data returned by each target business interface. Ultimately, the AI ​​system can integrate the response data from various interfaces and generate task response text in natural language as the task response result.

[0086] In addition, the AI ​​system can also log and archive the response data of each interface during the above-mentioned interface call phase and the final generated task response text for subsequent task review or task reuse.

[0087] This embodiment, after completing multi-intent decomposition and interface matching, further identifies the dependencies between parameters required for the execution of each intent fragment, and automatically generates an executable task flowchart based on these dependencies. This integrates multiple isolated interface calls into a logically related collaborative process, avoiding execution order errors or call failures caused by ignoring parameter dependencies. Simultaneously, by displaying the flowchart to the user and receiving feedback operation instructions, the user can confirm and adjust the process before execution. This leverages the efficiency advantages of AI automatic orchestration while retaining the control capability of human intervention, improving the accuracy of complex task execution and user trust in the execution process. Finally, the interfaces are called sequentially according to the user-confirmed process, ensuring correct data transmission between steps and achieving a complete and accurate response to complex instructions containing multiple subtasks.

[0088] Based on the first and second embodiments of this application, in the third embodiment of this application, the content that is the same as or similar to the first and second embodiments described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 5 , Figure 5This is a flowchart illustrating the third embodiment of the multi-system collaborative task response method of this application.

[0089] In this embodiment, to avoid call failure due to insufficient user permissions, the identity of the user initiating the call can be securely verified before executing the interface call. Therefore, step S305 also includes steps S3051 to S3054: Step S3051: Determine the permission requirements of each target business interface based on the target interface path matched by each intent fragment.

[0090] It should be noted that the AI ​​system can determine the corresponding target business interface based on the target interface path matched by each intent fragment, and then obtain the corresponding permission requirements from the interface metadata of the target business interface.

[0091] It should be understood that the permission requirements may include the scope of permissions required to call the target business interface, such as permission to create tasks, permission to query documents, and permission to approve orders; it may also include pre-defined system roles that can call the target business interface, such as "only project managers can call" or "administrators and ordinary employees can call".

[0092] Step S3052: Obtain the user identity information to which the current task instruction belongs, and determine whether the user identity information meets the permission requirements.

[0093] It should be understood that the user identity information associated with the current task instruction may include: the user's login ID in the AI ​​system, which can be the user's unique identifier in the AI ​​system; the user's role, such as "project manager", "developer", "ordinary employee", etc.; the permission list, that is, the full range of permissions that the user has; and authentication credentials, such as a token, which is used to carry when calling the target business interface.

[0094] In practice, the AI ​​system can compare user identity information with the aforementioned permission requirements item by item for each target business interface. Only when the permission requirements of all target business interfaces are met will it be determined that the user identity information meets the permission requirements.

[0095] Step S3053: When the user's identity information meets the permission requirements, inject the input parameters of each interface into each target business interface in sequence according to the current execution flow to call each target business interface and obtain the interface response data returned by each target business interface.

[0096] It should be understood that the AI ​​system will only execute the current execution process when the user's identity information meets the permission requirements of each target business interface, and will sequentially call each target business interface to obtain the corresponding interface response data.

[0097] The interface response data may include: operation confirmation information, such as "Jira task created successfully"; generated data identifiers, such as the ID of the newly created task or the URL of the newly created document; queried data content, such as inventory quantity or task list; and error information, such as the error code returned when the call fails.

[0098] It should also be noted that during the process of calling each target business interface in sequence, the AI ​​system can store and record the interface response data corresponding to each target business interface, so as to use it as the interface input parameter for calling subsequent target business interfaces.

[0099] Step S3054: Integrate the response data from each interface, and generate a task response text based on the integrated response data as the task response result.

[0100] It should be understood that after completing all the calls to the target business interfaces and obtaining the corresponding interface response data, the AI ​​system can then integrate the response data of each interface according to the call order in the current execution flow to obtain a sorted structured data set.

[0101] Finally, to improve the readability of the task response results, the AI ​​system can generate a natural language description, i.e., the task response text, based on the organized structured data set. This task response text is then fed back to the user through the interactive interface to inform the user of the execution result of the task described in the current task instruction.

[0102] For example, for fragments 1, 2, and 3 in the aforementioned example, the interface response data that needs to be integrated may include: the "task ID" and "task key" returned by the Jira task creation interface, the "document URL" returned by the Wiki document creation interface, and the "update success status" returned by the Jira task update interface.

[0103] The final generated task response text could be: "Your task is complete. Jira task JIRA-123 and its corresponding Wiki document have been created for you, and the Wiki link has been updated in the Jira task description."

[0104] In addition, the AI ​​system can receive user feedback on task response text, and archive the feedback, task response text, and stored interface response data to generate task operation logs for subsequent continuous optimization of the aforementioned basic big model, preset intent recognition big model, and preset business matching big model.

[0105] This embodiment ensures that users have legitimate access rights to all interfaces before executing multi-step tasks by uniformly verifying the permission requirements of each interface. This avoids execution failures or security risks due to insufficient permissions, thus improving the security and success rate of multi-system collaborative calls. After permission verification, each interface is called sequentially according to the user-confirmed process, and parameters are dynamically passed during the call to ensure that the data dependencies between each step are correctly satisfied, achieving complete execution of complex multi-step tasks. Finally, by integrating the response data returned by each interface and generating task response text in natural language format, the scattered technical return results can be transformed into a unified response that is easy for users to understand, improving the convenience and interactive experience for users to obtain task execution results.

[0106] Furthermore, you can also refer to this section. Figure 6 This application provides a complete description of the task response method for multi-system collaboration. Figure 6 This is a schematic diagram of the entire process of the multi-system collaborative task response method in this application.

[0107] Depend on Figure 6 It is known that the AI ​​system first receives the current task instruction input by the user, which is: "Create a technical research Jira task (from January 1, 2026 to February 1, 2026), and create a Wiki document and update Jira after completion."

[0108] The AI ​​system invokes a pre-defined intent recognition model to semantically parse the current task command, identifying three intent fragments as the current operation intent: Jira creation, Wiki creation, and Jira update.

[0109] The AI ​​system calls a pre-defined business matching model to match target business systems and interfaces in the MCP registry center, including: the Jira creation interface in the Jira system, the Wiki creation interface in the Wiki system, and the Jira update interface in the Jira system.

[0110] The AI ​​system first calls the Jira creation interface in the Jira system to obtain the task ID returned by the Jira creation interface; Next, the Wiki creation interface in the Wiki system is called to obtain the document URL returned by the Wiki creation interface; Finally, the Jira update interface in the Jira system is called to add the document URL to the task description corresponding to the aforementioned task ID, and the update success status is returned.

[0111] The AI ​​system receives and integrates the interface response data (task ID, document URL, and update success status) from each of the aforementioned target business interfaces to generate a task response result. This task response result can be in the form of a Jira task link plus a Wiki document URL link.

[0112] In addition, the AI ​​system can also maintain a log library module to store operation logs of the entire process of task instruction response. The operation logs may include the user's login ID in the AI ​​system, the time when the user inputs the current task instruction, the target business interface corresponding to each intent fragment, and the interface response data, etc.

[0113] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the task response method of multi-system collaboration in this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0114] This application also provides a multi-system collaborative task response device; please refer to... Figure 7 , Figure 7 This is a schematic diagram of the module structure of the multi-system collaborative task response device of this application. The multi-system collaborative task response device includes: The intent recognition module 701 is used to receive the current task instruction and determine the current operation intent corresponding to the current task instruction; The interface matching module 702 is used to match the target interface path of the target business system required by the current operation intention in the preset model context protocol registry center according to the current operation intention. The preset model context protocol registry center stores the interface paths of different business systems. The task execution module 703 is used to determine the interface input parameters according to the current operation intention, and call the target business interface in the target business system according to the target interface path based on the interface input parameters to obtain the task response result.

[0115] The multi-system collaborative task response device provided in this application, employing the multi-system collaborative task response method described in the above embodiments, can solve the technical problem that existing multi-system collaborative task response methods suffer from insufficient reliability and accuracy in task response results. Compared with the prior art, the beneficial effects of the multi-system collaborative task response device provided in this application are the same as those of the multi-system collaborative task response method described in the above embodiments, and other technical features in the multi-system collaborative task response device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0116] This application also provides a multi-system collaborative task response device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the multi-system collaborative task response method in Embodiment 1 above.

[0117] The following is for reference. Figure 8 , Figure 8 This is a schematic diagram of the multi-system collaborative task response device of this application. The multi-system collaborative task response device in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, personal digital assistants (PDAs), tablet computers (PADs), portable media players (PMPs), etc., as well as fixed terminals such as digital TVs, desktop computers, etc. Figure 8 The multi-system collaborative task response device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0118] like Figure 8As shown, a multi-system collaborative task response device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.) that can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 1002 or a program loaded from storage device 1003 into random access memory (RAM) 1004. RAM 1004 also stores various programs and data required for the operation of the multi-system collaborative task response device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows a multi-system collaborative task response device to wirelessly or wiredly communicate with other devices to exchange data. While a multi-system collaborative task response device with various systems is shown in the figure, it should be understood that it is not required to implement or possess all of the systems shown. More or fewer systems may be implemented alternatively.

[0119] The multi-system collaborative task response device provided in this application, employing the multi-system collaborative task response method described in the above embodiments, can solve the technical problem of multi-system collaborative task response. Compared with the prior art, the beneficial effects of the multi-system collaborative task response device provided in this application are the same as those of the multi-system collaborative task response method described in the above embodiments, and other technical features of this multi-system collaborative task response device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0120] This application also provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the multi-system collaborative task response method described in the above embodiments.

[0121] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, radio frequency (RF), etc., or any suitable combination thereof.

[0122] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described multi-system collaborative task response method, and is capable of solving the technical problems of the multi-system collaborative task response method. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the multi-system collaborative task response method provided in the above embodiments, and will not be repeated here.

[0123] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other elements in the process, method, article, or system that includes that element.

[0124] The above embodiment numbers are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments. They are only some embodiments of this application and do not limit the scope of this application. All equivalent structural transformations made based on the technical concept of this application and the content of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included within the protection scope of this application.

Claims

1. A multi-system coordinated task response method, characterized in that, The method includes: Receive the current task instruction and determine the current operation intent corresponding to the current task instruction; According to the current operation intention, the target interface path of the target business system required by the current operation intention is matched in the preset model context protocol registry center. The preset model context protocol registry center stores the interface paths of different business systems. Based on the current operation intent, determine the interface input parameters, and based on the interface input parameters, call the target business interface in the target business system according to the target interface path to obtain the task response result.

2. The method of claim 1, wherein, The step of receiving the current task instruction and determining the current operation intent corresponding to the current task instruction includes: The current task instruction is cleaned and standardized to obtain the task instruction text; The task instruction text is input into a preset intent recognition model to determine the current operation intent; The preset intent recognition model is used to identify intent categories based on the task instruction text, divide the task instruction text into several segments using preset segmentation words, and bind each segment to the intent category to obtain several intent segments as the current operation intent.

3. The method of claim 1, wherein, The step of matching the target interface path of the target business system required by the current operation intent in the preset model context protocol registry center according to the current operation intent includes: The business interface list of each business system is obtained from the preset model context protocol registry center. The business interface list includes several business interfaces and corresponding interface metadata. Each interface metadata is pre-configured by the management user and includes at least the interface path and function description. The current operation intent and the metadata of each interface are input into a preset business matching model to obtain the target business interface. The preset business matching model is used to match the functional description of each business interface with the current operation intent, and to determine the business interface with the highest matching degree as the target business interface. The interface path corresponding to the target business interface is determined as the target interface path of the target business system required by the current operation intent.

4. The method as described in claim 1, characterized in that, The current operation intent includes several intent fragments; the step of determining the interface input parameters based on the current operation intent, and calling the target business interface in the target business system according to the target interface path based on the interface input parameters to obtain the task response result includes: Extract the corresponding interface input parameters from each intent segment, and determine the parameter dependency relationship between each intent segment based on each interface input parameter; A task flowchart is generated based on the parameter dependencies and the target interface paths matched by each intent fragment. Display the task flowchart to the user and receive feedback operation instructions from the user on the current task flowchart; The task flowchart is updated according to the feedback operation instructions, and the current execution flow is determined based on the update result; According to the current execution flow, the target business interfaces in the target business system are called sequentially based on the input parameters of each interface and the target interface path to obtain the task response result.

5. The method as described in claim 4, characterized in that, The step of extracting the corresponding interface input parameters from each of the intent segments and determining the parameter dependencies between the intent segments based on the interface input parameters includes: The request parameter definitions for each target business interface are determined based on the target interface path matched by each intent fragment; Extract the corresponding interface input parameters from each intent fragment, and determine whether each interface input parameter satisfies the corresponding request parameter definition; When all the input parameters of each interface satisfy the corresponding request parameter definition, the parameter dependency relationship between each intent fragment is determined according to the input parameters of each interface.

6. The method as described in claim 4, characterized in that, The step of generating a task flowchart based on the parameter dependencies and the target interface paths matched by each intent fragment includes: Detect whether there are circular dependencies in the parameter dependencies and obtain the detection results; When the detection result indicates that there is no circular dependency, the task execution order is determined based on the parameter dependency relationship, and the task execution order is bound to the target interface path corresponding to each intent fragment to obtain a task flowchart.

7. The method as described in claim 4, characterized in that, The step of sequentially calling the target business interfaces in the target business system according to the current execution flow based on the input parameters of each interface and the target interface path to obtain the task response result includes: The permission requirements of each target business interface are determined based on the target interface path matched by each intent fragment; Obtain the user identity information to which the current task instruction belongs, and determine whether the user identity information meets the permission requirements. When the user identity information meets the permission requirements, the interface parameters are sequentially injected into the target business interface according to the current execution flow to call the target business interface and obtain the interface response data returned by the target business interface. The response data from each of the interfaces is integrated, and a task response text is generated based on the integrated response data as the task response result.

8. A multi-system collaborative task response device, characterized in that, The device includes: An intent recognition module is used to receive the current task instruction and determine the current operation intent corresponding to the current task instruction; The interface matching module is used to match the target interface path of the target business system required by the current operation intention in the preset model context protocol registry center according to the current operation intention. The preset model context protocol registry center stores the interface paths of different business systems. The task execution module is used to determine the interface input parameters according to the current operation intention, and call the target business interface in the target business system according to the target interface path based on the interface input parameters to obtain the task response result.

9. A multi-system collaborative task response device, characterized in that, The device includes: a memory, a processor, and a multi-system collaborative task response program stored in the memory and executable on the processor, the multi-system collaborative task response program being configured to implement the steps of the multi-system collaborative task response method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and the storage medium stores a multi-system collaborative task response program, which, when executed by a processor, implements the steps of the multi-system collaborative task response method as described in any one of claims 1 to 7.

Images