Data processing method and system and storage medium

Abstract

The invention provides a data processing method and system and a storage medium. In the data processing method, the data processing system can firstly receive demand information which is input by a user and is used for requesting to execute a calculation task of a target field; then, the data processing system calls an agent system based on the demand information to obtain a calculation result corresponding to the calculation task output by the agent system, the agent system comprises a planning agent and a plurality of execution agents, the planning agent is used for planning the calculation task on the basis of the demand information, and the execution agents are used for executing the calculation task on the basis of the planning agent. A calculation process corresponding to the calculation task is generated, target agents in the execution agents are called based on the calculation process to execute the calculation process, a calculation result is obtained, and decentralized data interaction can be carried out between the target agents based on the calculation process. And finally, the data processing system can output the measurement and calculation result generated by the intelligent agent system to the user.

Description

Technical Field

[0001] This specification relates to the field of artificial intelligence technology, and in particular to a data processing method, system and storage medium. Background Technology

[0002] In many fields, there are numerous and complex calculation tasks. Taking investment decision-making as an example, when conducting calculations, the first step is usually to configure expressions in an Excel spreadsheet. Then, the project's investment decisions are entered into the Excel spreadsheet, and the investment decisions are calculated using the expressions configured in the Excel spreadsheet to obtain the calculation results.

[0003] Currently, configuring expressions in Excel spreadsheets requires manually configuring corresponding expressions for different measurement indicators. However, using manually configured expressions in Excel spreadsheets is inefficient for calculations.

[0004] Therefore, how to improve the efficiency of calculation has become an urgent problem to be solved.

[0005] The background information is merely information known only to the inventor and does not imply that such information had entered the public domain before the date of this application, nor does it imply that it could be considered prior art in this disclosure. Summary of the Invention

[0006] This specification provides a data processing method, system, and storage medium that, based on user-inputted demand information, utilizes an intelligent agent system to obtain the calculation process corresponding to the requested calculation task. Then, through a target intelligent agent within the intelligent agent system, decentralized data interaction is performed according to the calculation process to execute the calculation task and obtain the calculation result.

[0007] To achieve the above objectives, the embodiments in this specification adopt the following technical solutions: Firstly, this specification provides a data processing method, the method comprising: receiving user-inputted demand information, the demand information being used to request the execution of a measurement task in a target domain; invoking an intelligent agent system based on the demand information to obtain a measurement result corresponding to the measurement task output by the intelligent agent system, wherein the intelligent agent system comprises: a planning intelligent agent and multiple execution intelligent agents, the planning intelligent agent being used to generate a measurement process corresponding to the measurement task based on the demand information, and invoking a target intelligent agent among the multiple execution intelligent agents to execute the measurement process based on the measurement process to obtain a measurement result, the target intelligent agents being able to perform decentralized data interaction based on the measurement process; and outputting the measurement result to the user.

[0008] In some embodiments, the planning agent is trained to focus on the measurement planning task of the target domain; generating the measurement process corresponding to the measurement task based on the requirement information includes: determining the measurement task corresponding to the requirement information based on the requirement information, and generating the measurement process required to execute the measurement task based on the measurement task.

[0009] In some embodiments, generating the computational process required to execute the computational task based on the computational task includes: searching a preset process library based on the computational task; when a candidate computational process corresponding to the computational task is found in the preset process library, determining the candidate computational process matching the computational task as the computational process required to execute the computational task; the process library stores candidate computational processes corresponding to different computational tasks; and when no candidate computational process corresponding to the computational task is found in the preset process library, inputting the computational task into a pre-trained planning large language model, guiding the planning large language model to perform inference based on the computational task, and obtaining the computational process corresponding to the computational task.

[0010] In some embodiments, the calculation process includes at least one calculation subtask and execution logic between the at least one calculation subtask. The agent system stores registration information of the plurality of execution agents, which represents the task type and interaction rules corresponding to the plurality of execution agents respectively. The step of inputting the calculation task into a pre-trained planning language model and guiding the planning language model to perform reasoning based on the calculation task to obtain the calculation process corresponding to the calculation task includes: generating a first prompt instruction based on the calculation task and the registration information of the plurality of execution agents. The first prompt instruction is used to guide the planning language model to determine at least one calculation subtask for implementing the calculation task based on the task type corresponding to the calculation task and the plurality of execution agents respectively, and to determine the execution logic of the at least one calculation subtask based on the interaction rules corresponding to the plurality of execution agents respectively. The first prompt instruction is then input into the planning language model to obtain the calculation process generated by the planning language model.

[0011] In some embodiments, the demand information is described in natural language. Determining the measurement task corresponding to the demand information based on the demand information includes: performing semantic analysis on the demand information; when there is text in the semantics of the demand information that expresses the measurement intent, determining that the measurement intent expressed by the text is the measurement task corresponding to the demand information; or, when there is no text in the semantics of the demand information that expresses the measurement intent, inputting the demand information into a pre-trained planning large language model, guiding the planning large language model to reason based on the demand information to obtain the measurement task corresponding to the demand information.

[0012] In some embodiments, the measurement process includes at least one measurement subtask and execution logic between the at least one measurement subtask. The plurality of execution agents includes measurement agents configured to execute measurement subtasks, with different measurement agents executing measurement subtasks of different task types. The step of calling a target agent among the plurality of execution agents to execute the measurement process based on the measurement process to obtain measurement results includes: determining a target agent corresponding to each measurement subtask from the measurement agents based on the task type corresponding to each measurement subtask in the measurement process; using the target agent to execute the measurement subtask based on the execution logic between the at least one measurement subtask to obtain measurement data corresponding to each measurement subtask; and generating a measurement result corresponding to the measurement task based on the measurement data corresponding to each measurement subtask.

[0013] In some embodiments, the step of using the target agent to execute the measurement sub-tasks based on the execution logic between the at least one measurement sub-task to obtain measurement data corresponding to each measurement sub-task includes: initiating a call request to the target agent; using the target agent to infer the measurement network corresponding to the measurement sub-task, wherein the measurement network includes multiple measurement nodes and the topological relationships between the multiple measurement nodes; obtaining the measurement expressions corresponding to the multiple measurement nodes respectively, and obtaining the input data of the measurement network based on the measurement expressions corresponding to the multiple measurement nodes respectively; and performing calculations based on the measurement network, the input data, and the measurement expressions corresponding to the multiple measurement nodes respectively to obtain the measurement data corresponding to the measurement sub-task.

[0014] In some embodiments, each of the computation nodes is used to represent a computation dimension; obtaining the computation expressions corresponding to the plurality of computation nodes respectively, and obtaining the input data of the computation network based on the computation expressions corresponding to the plurality of computation nodes respectively, includes: obtaining the computation expression corresponding to each of the computation nodes from a pre-configured expression library based on the computation dimensions corresponding to the plurality of computation nodes respectively, wherein the expression library stores the computation expressions corresponding to the plurality of computation dimensions respectively; determining the parameters that need to be input in the computation expressions corresponding to the plurality of computation nodes respectively as the input parameters of the computation network; and obtaining the parameter values ​​corresponding to the input parameters, and obtaining the input data based on the parameter values ​​of the input parameters.

[0015] In some embodiments, the plurality of execution agents includes a parameter acquisition agent configured to acquire input parameters; acquiring the parameter value corresponding to the input parameter includes: using the parameter acquisition agent to acquire the parameter value corresponding to the input parameter from a target location; when the parameter value of the target parameter cannot be acquired at the target location, displaying a parameter input page; and using the parameter input page to receive the parameter value of the target parameter input by the user.

[0016] In some embodiments, the target location includes at least one of the following: a knowledge base of the target domain; or at least one web page related to the target domain.

[0017] In some embodiments, the plurality of execution agents include an expression derivation agent configured to derive a measurement expression, and the method further includes: when a measurement expression corresponding to a target computational dimension cannot be obtained from the expression library, using the expression derivation agent to infer and generate a measurement expression corresponding to the target computational dimension based on domain knowledge of the target domain; and storing the measurement expression corresponding to the target computational dimension into the expression library.

[0018] In some embodiments, the step of calculating the measurement data corresponding to the measurement subtask based on the measurement network, the input data, and the measurement expressions corresponding to the plurality of measurement nodes includes: inputting the measurement network, the input data, and the measurement expressions corresponding to the plurality of measurement nodes into a calculation engine, wherein the calculation engine is configured to have the ability to identify and calculate measurement expressions of the target domain; and using the calculation engine to traverse and calculate the plurality of measurement nodes in the measurement network to obtain the measurement data corresponding to the measurement subtask.

[0019] In some embodiments, initiating a call request to the computational agent and using the computational agent to infer the computational network corresponding to the computational subtask includes: generating a second prompt instruction based on the requirement information and the computational subtask, wherein the second prompt instruction is used to guide a pre-trained computational large language model to refer to a knowledge base of the target domain and to perform inference based on the requirement information and the computational subtask to generate the computational network corresponding to the computational subtask; and inputting the second prompt instruction into the computational large language model to generate the computational network corresponding to the computational subtask through the computational large language model.

[0020] In some embodiments, after obtaining the measurement data corresponding to the measurement subtask, the method further includes: responding to receiving incremental demand information for the measurement subtask, the incremental demand information being used to instruct incremental measurement to be performed based on the measurement subtask; using the measurement agent, inferring the incremental subnetwork based on the incremental demand information, and connecting the incremental subnetwork to the measurement network to obtain an incremental measurement network, wherein the incremental subnetwork and the measurement network have the same data interaction specification; obtaining an incremental measurement expression and incremental input data, the incremental measurement expression being the measurement expression corresponding to the measurement node in the incremental subnetwork, and the incremental input data being the input data corresponding to the incremental subnetwork; and performing calculations based on the incremental measurement network, the input data of the measurement network, the measurement expression of the measurement network, the incremental input data, and the incremental measurement expression to obtain measurement data for incremental measurement performed based on the measurement subtask.

[0021] In some embodiments, the plurality of execution agents includes a summarizing agent configured to summarize the measurement results. The step of generating the measurement result corresponding to each measurement sub-task based on the measurement data includes: obtaining the summarizing requirement of the measurement task; generating a third prompt instruction based on the summarizing requirement and the measurement data corresponding to at least one measurement sub-task; the third prompt instruction being used to guide a pre-trained summarizing language model to perform reasoning based on the summarizing requirement to generate the measurement result; and inputting the third prompt instruction into the summarizing language model to generate the measurement result.

[0022] In some embodiments, the method further includes: when an exception occurs during the execution of the calculation process by the target agent, instructing the planning agent to update the calculation process corresponding to the calculation task based on the requirement information and the exception; and invoking the target agent among the plurality of execution agents to execute the calculation process based on the updated calculation process.

[0023] Secondly, this specification also provides a data processing system, comprising: at least one storage medium storing at least one instruction set; and at least one processor communicatively connected to the at least one storage medium, wherein, when the data processing system is running, the at least one processor reads the at least one instruction set and implements the method provided in any of the first aspects according to the instructions of the at least one instruction set.

[0024] Thirdly, this specification also provides a computer-readable non-volatile storage medium, wherein the computer-readable non-volatile storage medium stores at least one instruction set, which, when executed by at least one processor, implements the method provided in any of the first aspects.

[0025] Other functions of the data processing methods, systems, and storage media provided in this specification will be partially listed in the following description. The inventive aspects of the data processing methods, systems, and storage media provided in this specification can be fully understood through practice or use of the methods, apparatus, and combinations described in the detailed examples below. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of this specification, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 A schematic diagram illustrating an application scenario of a data processing method provided according to an embodiment of this specification is shown. Figure 2 A hardware structure diagram of a computing system provided according to an embodiment of this specification is shown; Figure 3 A flowchart of a data processing method provided according to an embodiment of this specification is shown; Figure 4 A schematic diagram of an intelligent agent system provided according to an embodiment of this specification is shown; Figure 5 A schematic diagram is shown illustrating an embodiment of this specification of obtaining measurement results using an intelligent agent system; Figure 6 A schematic diagram illustrating an embodiment of this specification for obtaining measurement results based on a measurement process is shown. Figure 7 A schematic diagram of a measurement network provided according to an embodiment of this specification is shown; Figure 8A schematic diagram of a measurement agent generation measurement network provided according to an embodiment of this specification is shown; Figure 9 A schematic diagram illustrating the acquisition of the calculation expression according to an embodiment of this specification is shown; Figure 10 A schematic diagram illustrating the acquisition of parameter values ​​for measurement input parameters according to embodiments of this specification is shown; and Figure 11 A schematic diagram of an incremental measurement network provided according to an embodiment of this specification is shown. Detailed Implementation

[0028] The following description provides specific application scenarios and requirements for this specification, intended to enable those skilled in the art to make and use the contents of this specification. Various partial modifications to the disclosed embodiments will be apparent to those skilled in the art, and the general principles defined herein can be applied to other embodiments and applications without departing from the spirit and scope of this specification. Therefore, this specification is not limited to the embodiments shown, but rather to the widest scope consistent with the claims.

[0029] The terminology used herein is for the purpose of describing particular exemplary embodiments only and is not restrictive. For example, unless the context clearly indicates otherwise, the singular forms “a,” “an,” and “the” used herein may also include the plural forms. When used in this specification, the terms “comprising,” “including,” and / or “containing” mean that the associated integers, steps, operations, elements, and / or components are present, but do not exclude the presence of one or more other features, integers, steps, operations, elements, components, and / or groups, or that other features, integers, steps, operations, elements, components, and / or groups may be added to the system / method.

[0030] Considering the following description, these and other features of this specification, as well as the operation and function of the related components of the structure, and the economy of assembly and manufacture of the parts, can be significantly improved. All of these form part of this specification with reference to the accompanying drawings. However, it should be clearly understood that the drawings are for illustrative and descriptive purposes only and are not intended to limit the scope of this specification. It should also be understood that the drawings are not drawn to scale.

[0031] The flowcharts used in this specification illustrate operations implemented according to some embodiments of this specification. It should be clearly understood that the operations in the flowcharts may not be implemented in a sequential order. Instead, the operations may be implemented in reverse order or simultaneously. Furthermore, one or more additional operations may be added to the flowcharts. One or more operations may be removed from the flowcharts.

[0032] For ease of description, the terms that will appear later in this manual will be explained first.

[0033] Term 1: Investment Assessment. Also known as investment decision assessment or financial evaluation, it is a process of systematically analyzing and predicting the economic feasibility of an investment project. During investment assessment, based on the input project requirements information, pre-defined expressions are used to perform calculations and output multi-dimensional indicators (i.e., assessment results).

[0034] Term 2: End-to-End. End-to-end refers to using a unified model to directly generate the final measurement results from the original input.

[0035] Term 3: Intelligent Agent. An intelligent agent is a computational entity that perceives the environment and takes actions to achieve a specific goal. The computational entity corresponding to an intelligent agent can be a computing device that runs computer programs or deploys large models (or can call large models), or a chip, intelligent module, etc. integrated into the computing device.

[0036] Term 4: Tools. Tools for an intelligent agent refer to the external resources, functional modules, or auxiliary capabilities that the agent invokes to achieve its goals. For example, in a virtual environment, tools may include Application Programming Interfaces (APIs), services, databases, storage, simulators, knowledge graphs, and algorithm libraries.

[0037] In this specification, the Large Language Model (LLM) may also be referred to simply as the Large Model. A Large Language Model is a natural language processing model based on deep learning techniques, typically with billions to hundreds of billions or even more parameters, possessing powerful language understanding and generation capabilities. Large Language Models can employ the Transformer architecture or its variants (such as GPT, BERT, etc.), which utilizes an attention mechanism to globally model sequential data, efficiently handling long-distance dependencies and thus performing excellently in natural language tasks. Large Language Models learn the statistical features and semantic relationships of language through pre-training on large-scale corpora, giving them excellent generalization capabilities. The core capabilities of Large Language Models include, but are not limited to: understanding contextual semantics, generating coherent and grammatically correct text, performing logical reasoning, and handling multi-task scenarios. Its usage typically includes two modes: direct inference and fine-tuning. In direct inference mode, the user guides the Large Language Model to generate specific outputs by designing prompts. Cue words can be task descriptions or instructions in text form, used to stimulate the semantic understanding and generation capabilities of large language models. In fine-tuning mode, large language models are further trained on small-scale datasets in specific domains to optimize their performance on specific tasks. The powerful generalization ability and flexibility of large language models make them an important tool in the field of artificial intelligence, providing efficient and accurate solutions for automated text generation and understanding.

[0038] In some embodiments, large language models can also understand and generate data from other modalities (such as visual and audio data). In this case, large language models can also be called multimodal large language models (MLLMs). MLLMs provide a richer and more natural interactive experience by integrating multiple types of input and output, such as text, images, and sound. The core advantage of MLLMs lies in their ability to process and understand information from different modalities and fuse this information to complete complex tasks. For example, MLLMs can analyze an image and generate descriptive text, or generate a corresponding image based on a text description. This cross-modal understanding and generation capability makes MLLMs widely applicable across multiple fields.

[0039] It should be noted that the key technologies of large language models can be found in the detailed description in the paper "A Survey of Large Language Models" (paper number: arXiv:2303.18223v16, published on March 11, 2025, public link: https: / / doi.org / 10.48550 / arXiv.2303.18223), and will not be repeated here.

[0040] The following section introduces the application scenarios of this manual.

[0041] This specification provides a data processing method that can be applied to scenarios involving investment calculations for projects to be measured. The project to be measured can be an investment project within a target sector; for example, when the target sector is the energy sector, the project to be measured could be a photovoltaic power plant project, an energy storage system project, or other projects requiring construction within the energy sector.

[0042] In related technologies, when conducting investment calculations for a project, the process typically begins by configuring expressions related to the required indicators in an Excel spreadsheet. Then, project data and demand information are collected, and the calculations are performed using the expressions within the Excel spreadsheet to obtain the results. However, configuring these indicator expressions manually is overly reliant on human experience. When dealing with large projects, this manual configuration consumes significant time, leading to inefficiency and a high risk of errors. Furthermore, Excel's limited calculation capabilities result in prolonged calculation times for each operation, further contributing to the overall low efficiency.

[0043] This specification provides a data processing method applicable to a data processing system. The method includes: the data processing system first receives user-inputted requirement information, which requests the execution of a measurement task in a target domain. Then, the data processing system invokes an intelligent agent system based on the requirement information to obtain the measurement result corresponding to the measurement task output by the intelligent agent system. The intelligent agent system includes a planning intelligent agent and multiple execution intelligent agents. The planning intelligent agent generates a measurement process corresponding to the measurement task based on the requirement information and invokes a target intelligent agent among the multiple execution intelligent agents to execute the measurement process, obtaining the measurement result. The target intelligent agents can perform decentralized data interaction based on the measurement process. Finally, the data processing system outputs the measurement result generated by the intelligent agent system to the user.

[0044] In the data processing method provided in this specification, the data processing system receives user-input demand information and invokes an intelligent agent system. Then, the data processing system utilizes the planning agent within the intelligent agent system to generate a calculation process based on the calculation tasks corresponding to the demand information. It then invokes multiple execution agents, including the target agent, to perform decentralized data interaction based on the calculation process to execute it. When executing calculation tasks in this way, after the user provides demand information, the intelligent agent system can automatically analyze the demand information and execute the corresponding calculation tasks, achieving end-to-end automated calculation, effectively improving calculation efficiency and reducing the probability of calculation errors.

[0045] Figure 1 A schematic diagram illustrating an application scenario of a data processing method provided according to an embodiment of this specification is shown. For example... Figure 1 As shown, the application scenario 100 may include a data processing system 11, a terminal device 12, and a user 13.

[0046] The data processing system 11 can be deployed on a device or cluster of devices with data processing capabilities. For example, the data processing system 11 can be deployed on physical devices such as servers, server clusters, and cloud servers. In this case, the physical device corresponding to the data processing system 11 can store data or instructions for executing the data processing methods described in this specification, and can execute or be used to execute the data or instructions. In some embodiments, the physical device corresponding to the data processing system 11 may include hardware devices with data information processing functions and the necessary programs required to drive the hardware devices to work.

[0047] refer to Figure 1 The data processing system 11 may include / deploy an intelligent agent system. This intelligent agent system includes multiple intelligent agents. During the operation of the intelligent agent system, the multiple intelligent agents can achieve decentralized data interaction. It should be noted that the embodiments in this specification do not limit the number of intelligent agents included in the intelligent agent system.

[0048] Terminal device 12 can respond to the input operation of user 13, obtain the user's input request information, and send it to the data processing system 11.

[0049] In some embodiments, terminal device 12 may include mobile devices, tablets, laptops, built-in devices in motor vehicles, or similar content, or any combination thereof. In some embodiments, the mobile device may include smart home devices, smart mobile devices, virtual reality devices, augmented reality devices, or similar devices, or any combination thereof. In some embodiments, the smart home device may include smart TVs, desktop computers, etc., or any combination thereof. In some embodiments, the smart mobile device may include smartphones, personal digital assistants, gaming devices, navigation devices, etc., or any combination thereof. In some embodiments, the virtual reality device or augmented reality device may include virtual reality headsets, virtual reality glasses, virtual reality patches, augmented reality headsets, augmented reality glasses, augmented reality patches, or similar content, or any combination thereof. For example, the virtual reality device or the augmented reality device may include Google Glass, head-mounted displays, VR, etc. In some embodiments, the built-in device in the motor vehicle may include an in-vehicle computer, an in-vehicle television, etc.

[0050] In some embodiments, the terminal device 12 may have one or more applications (APPs) installed. The APPs provide human-computer interaction capabilities and interfaces. These APPs include, but are not limited to: web browser APPs, search APPs, chat APPs, shopping APPs, video APPs, financial management APPs, instant messaging tools, email clients, social media platforms, etc. In some embodiments, the terminal device 12 may have a calculation APP installed. This calculation APP can respond to the input operations of the user 13 and obtain the user's input request information.

[0051] It should be understood that Figure 1 The number of data processing systems 11 and terminal devices 12 shown is merely illustrative. Depending on implementation needs, any number of data processing systems 11 and terminal devices 12 can be used.

[0052] Figure 2 A hardware structure diagram of a computing system provided according to an embodiment of this specification is shown. The computing system 200 can serve as... Figure 1 The data processing system 11 in the specification executes the data processing method described herein.

[0053] like Figure 2 As shown, the computing system 200 may include at least one storage medium 230 and at least one processor 220. In some embodiments, the computing system 200 may also include a communication port 250 and an internal communication bus 210. The computing system 200 may also include I / O components 260.

[0054] The internal communication bus 210 can connect to different system components. For example, the internal communication bus 210 can connect to storage medium 230, processor 220, communication port 250, and I / O component 260, etc.

[0055] I / O component 260 supports input / output between computing system 200 and other components.

[0056] Communication port 250 is used for data communication between computing system 200 and the outside world. For example, communication port 250 can be used for data communication between computing system 200 and a network. Communication port 250 can be a wired communication port or a wireless communication port.

[0057] Storage medium 230 may include a data storage device. The data storage device may be a non-transitory storage medium or a temporary storage medium. For example, the data storage device may include one or more of a disk 232, a read-only storage medium (ROM) 235, or a random access storage medium (RAM) 236. Storage medium 230 also includes at least one instruction set stored in the data storage device. The instruction set may include computer program code, which may include programs, routines, objects, components, data structures, procedures, modules, etc.

[0058] At least one processor 220 may be communicatively connected to at least one storage medium 230. When the computing system 200 is running, at least one processor 220 reads the at least one instruction set and, according to the instructions of the at least one instruction set, executes the test method for the target system provided in this specification. The processor 220 may perform the steps included in the test method for the target system. The processor 220 may be in the form of one or more processors. In some embodiments, the processor 220 may include one or more hardware processors, such as a microcontroller, microprocessor, reduced instruction set computer (RISC), application-specific integrated circuit (ASIC), application-specific instruction set processor (ASIP), central processing unit (CPU), graphics processing unit (GPU), physical processing unit (PPU), microcontroller unit, digital signal processor (DSP), field-programmable gate array (FPGA), advanced RISC machine (ARM), programmable logic device (PLD), any circuit or processor capable of performing one or more functions, or any combination thereof.

[0059] For illustrative purposes only, the accompanying drawings show only one processor 220 for the computing system 200. However, it should be noted that the computing system 200 may also include multiple processors; therefore, the operations and / or method steps disclosed herein may be executed by one processor or by multiple processors in combination. For example, if the processor 220 of the computing system 200 described in this specification executes steps A and B, it should be understood that steps A and B may also be executed jointly or separately by two different processors 220 (e.g., a first processor executes step A, a second processor executes step B, or the first and second processors jointly execute steps A and B).

[0060] Figure 3 A flowchart of a data processing method provided according to an embodiment of this specification is shown. As previously described, a data processing system can execute this data processing method.

[0061] like Figure 3 As shown, data processing methods may include: S310: Receives user input request information, which is used to request the execution of a measurement task in the target domain.

[0062] In some embodiments, reference Figure 1 The terminal device can receive user input request information and send it to the data processing system. Specifically, the terminal device can provide an input page and respond to user interactions with that page to obtain the user's input request information. For example, the input page could be a workbench page or a chat window page.

[0063] As an example, the workbench page can display multiple preset project requirements, along with an interactive window for each requirement. Input interactions can be performed on the corresponding interactive window for each project requirement. For instance, a user can define the project requirements involved in the calculation and interact with the corresponding interactive window to input the data. Assuming the target field is energy, project requirements could include: project location, project configuration, project scale, and calculation tasks. Correspondingly, the requirement information can be project requirements recorded in a structured data format, along with the specific data corresponding to those requirements.

[0064] As an example, a chat window page can display an input interface, and the input interaction can be an operation performed on this interface. The input interface can respond to the user's input and receive the user's request information described in natural language. This request information can include details such as the project location, project configuration, project scale, and calculation tasks, all described in natural language.

[0065] S320: Based on the demand information, the intelligent agent system is invoked to obtain the calculation results corresponding to the calculation task output by the intelligent agent system. The intelligent agent system includes: a planning intelligent agent and multiple execution intelligent agents. The planning intelligent agent is used to generate the calculation process corresponding to the calculation task based on the demand information, and to invoke the target intelligent agent among the multiple execution intelligent agents to execute the calculation process based on the calculation process to obtain the calculation results. The target intelligent agents can perform decentralized data interaction based on the calculation process.

[0066] In some embodiments, an intelligent agent system is a multi-agent system (MAS), which includes multiple intelligent agents that can cooperate and serve each other to jointly complete the measurement task.

[0067] In this specification, decentralization means that data flow between multiple agents does not depend on a single control node, but rather is achieved through loosely coupled communication via a distributed communication mechanism. For example, in an agent system, the planning agent and multiple execution agents can publish messages or subscribe to data through a data bus.

[0068] In this architecture, when agents within the system interact with each other (e.g., make calls), they do not send data directly to the target agent. Instead, they publish messages (e.g., call requests) through a data bus as an intermediary. Correspondingly, the called party can receive and respond to these call requests by subscribing to relevant messages on the data bus.

[0069] For example, when a planning agent calls upon a target agent to execute a calculation process, it can publish a call request for the target agent on the data bus. The target agent can then receive and execute the call request by subscribing to relevant messages on the data bus.

[0070] Figure 4 A schematic diagram of an intelligent agent system provided according to an embodiment of this specification is shown.

[0071] In some embodiments, reference Figure 4 The intelligent agent system includes a planning intelligent agent and multiple execution intelligent agents. The data processing system can input demand information into the planning intelligent agent in the intelligent agent system to enable the system to be invoked.

[0072] In an intelligent agent system, each agent includes a large language model corresponding to the agent and the tools that the agent can invoke. The large language model can be deployed within the entity to which the agent resides. Alternatively, the large language model can be deployed within other entities, and the agent can invoke the large language model via an API.

[0073] It should be noted that in the agent system provided in this specification, the data interaction between the planning agent and multiple executing agents is decentralized. Decentralized data interaction means that in the agent system, there is no hierarchy of parent agents (i.e., the center) and child agents; that is, there is no hierarchical restriction that only the parent agent can call the child agents. In the agent system provided in this specification, all agents can call each other, thereby realizing multi-agent collaborative decision-making and task execution, effectively improving the flexibility and efficiency of the agent system when handling complex tasks.

[0074] For example, both the planning agent and the multiple execution agents can include tools for invoking other agents, and both can read the registration information recorded by the agent system. During operation, both the planning agent and the multiple execution agents can use the registration information as a reference, allowing any agent to autonomously invoke the capabilities of other agents when needed, and utilize those capabilities to perform tasks.

[0075] In some embodiments, the planning agent is trained to focus on computational planning tasks within the target domain. The data processing system can utilize the planning agent to determine the computational tasks corresponding to the demand information, and generate the computational flow required to execute the computational tasks. Training the planning agent refers to training the large language model within the planning agent. For example, when training the large language model, supervised fine-tuning, reinforcement learning, or other methods can be used to fine-tune the base large language model based on datasets related to the computational planning tasks in the target domain, resulting in the planning large language model used by the planning agent.

[0076] Figure 5 A schematic diagram is shown illustrating an embodiment of this specification of obtaining measurement results using an intelligent agent system.

[0077] In some embodiments, referring to the example in S310, the requirement information can be represented by structured data or by natural language. Specifically, for requirement information represented by structured data, project requirements may include calculation tasks. Therefore, the planning agent can directly determine the calculation tasks corresponding to the requirement information from the structured data.

[0078] However, the measurement task may not be directly determined from the demand information expressed in natural language. Therefore, refer to Figure 5 When demand information is described in natural language, the data processing system can use a planning agent to perform semantic analysis on the demand information. When there is text in the semantics of the demand information that expresses the measurement intention, the measurement intention expressed in the text is determined to be the measurement task corresponding to the demand information.

[0079] For example, suppose the requirement information is: "I want to evaluate the economics of building a 100MW photovoltaic power station in location A and find the optimal component configuration scheme, taking into account local conditions." After the data processing system inputs the requirement information into the planning agent, the planning agent can first perform semantic analysis on the requirement information, detect text containing words such as "economics" and "optimal" that express the intention of the calculation, and determine that the calculation task at least requires economic calculation and finding the optimal configuration.

[0080] Alternatively, when the semantics of the demand information do not contain text to express the measurement intent, the data processing system can input the demand information into a pre-trained planning language model, guiding the planning language model to reason based on the demand information and obtain the measurement task corresponding to the demand information.

[0081] For example, suppose the demand information is: "What is the electricity price in location A? The project scale is 50MW." After the data processing system inputs the demand information into the planning agent, the planning agent can first perform semantic analysis on the demand information. Since no text expressing the calculation intent is detected in the demand information, the planning agent can use a planning big language model to infer based on the demand information. For example, the result of the planning big language model's inference based on the demand information could be: "Based on the user's query about the electricity price and project scale in location A, it is inferred that the user's intent is to conduct an economic calculation of the photovoltaic power station to assess the project's feasibility." In this case, the planning agent determines that the calculation task requires at least an economic calculation.

[0082] In this embodiment, the data processing system utilizes the reasoning capabilities of the planning language model within the planning agent to analyze demand information described in natural language, thereby determining the corresponding computational tasks. Since using natural language to describe demand information aligns more closely with users' normal behavior, it effectively reduces the user's learning curve and improves the user experience.

[0083] In some embodiments, reference Figure 5 After the data processing system obtains the calculation task using the planning agent, it can obtain the calculation process required to execute the calculation task based on the calculation task.

[0084] As an example, refer to Figure 5 The data processing system can utilize planning agents to search within a pre-defined process library based on the calculation task.

[0085] When a candidate calculation process corresponding to a calculation task is found in the preset process library, the planning agent can determine that the candidate calculation process matching the calculation task is the calculation process required to execute the calculation task.

[0086] In some embodiments, a pre-defined process library can serve as a tool for planning the intelligent agent. The process library stores candidate calculation processes corresponding to different calculation tasks. The candidate calculation processes corresponding to the calculation tasks can be pre-configured or calculation processes corresponding to historical calculation tasks.

[0087] As an example, the calculation process may include at least one calculation subtask and the execution logic between at least one calculation subtask. For example, the calculation subtask may include: economic calculation, sensitivity analysis, reverse optimization, and outputting calculation results. Economic calculation refers to calculating economic indicators based on the project configuration in the demand information. Sensitivity analysis refers to analyzing the impact of a change in a parameter in the project configuration on the economic indicators. Reverse optimization refers to specifying a value for a certain economic indicator and adjusting the project configuration with that value as the target to obtain a project configuration that achieves the economic indicator. Outputting calculation results refers to outputting a calculation report or calculation table in a preset format based on the calculation results.

[0088] In some embodiments, referring to the example of demand information above, when the calculation task is determined to be "performing economic calculations and finding the optimal configuration" based on the demand information, matching can be performed in the process library based on keywords such as "economic calculation" and "finding the optimal configuration". When a candidate calculation process corresponding to the calculation task exists, the candidate calculation process matching the calculation task is determined as the calculation process required to execute the calculation task. For example, the calculation process matched based on "economic calculation" and "finding the optimal configuration" may include three sub-tasks: "economic calculation", "sensitivity analysis", and "outputting calculation results". The execution logic between the three calculation sub-tasks is as follows: first, perform economic calculations; then, based on the results of the economic calculations, perform sensitivity analysis; and finally, output the calculation results based on the results of the economic calculations and the sensitivity analysis.

[0089] As an example, when there is no candidate calculation process corresponding to the calculation task in the preset process library, the planning agent can input the calculation task into the pre-trained planning language model, guide the planning language model to reason based on the calculation task, and obtain the calculation process corresponding to the calculation task.

[0090] In some embodiments, reference Figure 5 The intelligent agent system stores the registration information of multiple executing intelligent agents. The registration information is used to represent the task type and interaction rules (i.e., the capabilities and invocation methods of each executing intelligent agent) corresponding to the multiple executing intelligent agents.

[0091] The planning agent can generate a first prompt instruction based on the measurement task and the registration information of multiple execution agents. The first prompt instruction is used to guide the planning big language model to determine at least one measurement sub-task to implement the measurement task based on the task type corresponding to the measurement task and the multiple execution agents respectively, and to determine the execution logic of at least one measurement sub-task based on the interaction rules corresponding to the multiple execution agents respectively. The first prompt instruction is then input into the planning big language model to obtain the measurement process generated by the planning big language model.

[0092] The first prompt instruction can be generated based on a template. This template can be pre-configured with text to guide the large language model's reasoning process. For example, in the energy sector, the following is an example template for a first prompt instruction: "You are a seasoned energy project planning expert. Based on the given calculation tasks and the available expert agent team, please decompose the tasks and plan the process."

[0093] Task Objective: [Calculation Task] Available expert teams (executing agents) and their capabilities: [Registration information for multiple executing agents] Planning requirements: 1. Task decomposition: Please decompose the above "task objectives" into at least one calculation sub-task that can be performed by the above "available expert team".

[0094] 2. Logical Arrangement: Based on the capabilities of each expert, please clarify the collaboration logic and data flow among them. Please clarify the dependencies between the measurement subtasks.

[0095] 3. Output Format: Please output the computation flow in the form of a Directed Acyclic Graph (DAG). Nodes in the DAG represent computation subtasks, edges represent the execution logic between these subtasks, and node information identifies the corresponding agent executing the subtask. The planning agent can input the calculation task and the registration information of multiple executing agents into the template of the first prompt instruction to generate the first prompt instruction. Then, the planning agent can input the first prompt instruction into the planning language model to guide the planning language model to generate the calculation process.

[0096] In this embodiment, the data processing system utilizes a planning agent to determine the corresponding candidate calculation process from a process library based on the calculation task. Alternatively, if no corresponding candidate calculation process exists in the process library, the planning agent inputs the calculation task into a pre-trained planning language model to deduce the calculation process. This method of obtaining the calculation process can break down complex tasks into multiple simple sub-tasks, enabling the data processing system to adapt to the needs of different calculation tasks and obtain calculation processes with a higher degree of matching to the task. The data processing system then uses the obtained calculation process to execute the calculation task, effectively improving the execution efficiency and accuracy of the task.

[0097] Figure 6 A schematic diagram illustrating a calculation result obtained based on a calculation process according to an embodiment of this specification is shown.

[0098] In some implementations, reference Figure 6 Among the multiple execution agents are computation agents configured to perform computation subtasks, with different computation agents performing computation subtasks of different task types.

[0099] The data processing system can utilize an intelligent agent system to determine the target intelligent agent for each measurement sub-task based on its task type within the measurement process. Then, the data processing system can leverage the execution logic between at least one measurement sub-task to execute the measurement sub-task using the target intelligent agent, obtaining the measurement data for each sub-task. Finally, the data processing system can use the intelligent agent system to generate the measurement results for each measurement sub-task based on the measurement data.

[0100] For example, refer to Figure 6 The measurement process includes three sub-tasks, of which sub-tasks 1 and 2 can be executed in parallel. Sub-task 3 requires execution based on the measurement data corresponding to sub-tasks 1 and 2 respectively to obtain the measurement result. Correspondingly, when the data processing system utilizes the intelligent agent system to execute the measurement task based on the measurement process, it can first determine the target intelligent agents corresponding to sub-tasks 1, 2, and 3. For example, refer to... Figure 6 The target agent corresponding to the calculation subtask 1 is the calculation agent 1, the target agent corresponding to the calculation subtask 2 is the calculation agent 2, and the target agent corresponding to the calculation subtask 3 is the calculation agent 3.

[0101] As an example, the data processing system can use an intelligent agent system to call computational intelligent agent 1 and computational intelligent agent 2 to execute computational subtask 1 and computational subtask 2 respectively, obtaining the computational data corresponding to computational subtask 1 and computational subtask 2. Then, the data processing system can use the intelligent agent system to input the computational data corresponding to computational subtask 1 and computational subtask 2 into computational intelligent agent 3, and computational intelligent agent 3 will generate the computational results corresponding to the computational tasks.

[0102] In some implementations, the data processing system can utilize an intelligent agent system to initiate a call request to a target intelligent agent, and use the target intelligent agent to infer the computational network corresponding to the computational subtask. The computational network includes multiple computational nodes and the topological relationships between these nodes. The system then obtains the computational expressions corresponding to each of the multiple computational nodes and acquires the input data for the computational network based on these expressions. Furthermore, the data processing system can also utilize the intelligent agent system to perform calculations based on the computational network, the input data, and the computational expressions corresponding to the multiple computational nodes to obtain the computational data corresponding to the computational subtask.

[0103] Figure 7 A schematic diagram of a measurement network provided according to an embodiment of this specification is shown.

[0104] In some implementations, reference Figure 7 The measurement network can include measurement nodes 1 to N, and the topological relationships between these nodes (represented by dashed arrows). Each measurement node can represent a computational dimension, which can represent an economic indicator or an intermediate quantity used to calculate that indicator. Each measurement node includes at least one measurement expression used to calculate the computational dimension corresponding to that node.

[0105] Figure 8 A schematic diagram of a measurement agent generation measurement network provided according to an embodiment of this specification is shown.

[0106] In some implementations, reference Figure 8 The data processing system can utilize an intelligent agent system to generate a second prompt instruction based on demand information and computational sub-tasks. This second prompt instruction guides a pre-trained computational language model to reference a knowledge base in the target domain and perform reasoning based on the demand information and computational sub-tasks, generating a computational network corresponding to the computational sub-tasks. Furthermore, the data processing system can use the intelligent agent system to input the second prompt instruction into the computational language model, thereby generating the computational network corresponding to the computational sub-tasks.

[0107] The second prompt instruction can be generated based on a template. This template can be pre-configured with text to guide the large language model in reasoning. For example, in the energy sector, the following is an example of a second prompt instruction template: "You are an energy measurement expert. Based on the provided requirements and the measurement sub-tasks to be performed, please build an executable, dependent computational network."

[0108] Project Requirements: [Requirements Information] Test operator subtask: [Test operator subtask] Reference domain knowledge: [Knowledge base of the target domain]. Please be sure to refer to and utilize the reference domain knowledge for reasoning.

[0109] Task: Construct a measurement network.

[0110] Based on the above information, please reason and decompose the "measurement subtask" into a measurement network consisting of multiple measurement nodes. Each node represents an independent computational dimension, and the nodes form topological dependencies through data input and output. The computational agent can input the demand information and computational subtasks into the template of the second prompt instruction to generate the second prompt instruction. Then, the computational agent can input the second prompt instruction into the computational large language model to guide the computational large language model to generate the computational network.

[0111] In this embodiment, the data processing system utilizes an intelligent agent system to generate a second prompt instruction based on demand information and the calculation sub-task. Then, the data processing system uses the intelligent agent system to input the second prompt instruction into the calculation large language model, guiding the calculation large language model to perform reasoning with reference to the target domain's knowledge base, generating the calculation network corresponding to the calculation sub-task. The calculation network obtained in this way can dynamically construct a more highly matched calculation network for different demand information and calculation sub-tasks, effectively improving the execution efficiency of the calculation sub-tasks.

[0112] Figure 9 A schematic diagram is shown illustrating the acquisition of the calculation expression according to an embodiment of this specification.

[0113] In some implementations, reference Figure 9 The data processing system can utilize the corresponding computational agents within the intelligent agent system to retrieve the computational expressions corresponding to each computational node from a pre-configured expression library, based on the computational dimensions corresponding to multiple computational nodes. The expression library can be deployed within the intelligent agent system or, alternatively, within the data processing system.

[0114] The expression library stores measurement expressions for multiple computational dimensions related to the target domain. The data processing system can utilize a measurement agent to search the expression library based on the computational dimension. When a computational dimension is found in the expression library, the measurement agent can retrieve the corresponding measurement expression for that dimension and use it as the measurement expression for the measurement node.

[0115] In some implementations, reference Figure 8 The system includes multiple execution agents, including an expression derivation agent configured to derive computational expressions. When the computational expression corresponding to the target computational dimension cannot be found in the expression library, the data processing system can use the computational agent to call the expression derivation agent. The expression derivation agent then uses its domain knowledge of the target domain to infer and generate the computational expression corresponding to the target computational dimension. Furthermore, the data processing system can use the expression derivation agent to store the computational expression corresponding to the target computational dimension into the expression library.

[0116] As an example, the expression derivation agent can generate expression derivation reference instructions based on the target computational dimension and domain knowledge of the target domain. These reference instructions can then guide a pre-trained expression derivation model, enabling it to derive the computational expression corresponding to the target computational dimension based on the domain knowledge. The expression derivation agent can then input the reference instructions into the model to obtain the computational expression corresponding to the target computational dimension. Finally, the agent can store the computational expression for the target computational dimension in an expression library for later use.

[0117] In this embodiment, the data processing system utilizes an intelligent agent system to obtain the calculation expression corresponding to each calculation node from a pre-configured expression library, based on the calculation dimensions corresponding to multiple calculation nodes in the calculation network. When the expression corresponding to the target calculation dimension cannot be obtained from the expression library, the intelligent agent uses expression derivation to derive the expression corresponding to the target calculation dimension by referring to the knowledge base of the target domain. This method of obtaining calculation expressions can automatically generate the required expression when the expression corresponding to the target calculation dimension is missing, and use the generated expression to complete the calculation task. This effectively improves the adaptive capability and reliability of the data processing system in complex calculation scenarios.

[0118] In some implementations, the data processing system can utilize a computational agent to determine the parameters required as input to the computational expressions corresponding to multiple computational nodes, which are then used as input parameters for the computational network. Furthermore, the data processing system can also utilize this computational agent to obtain the parameter values ​​corresponding to the input parameters and derive input data based on these parameter values.

[0119] The calculation expressions corresponding to the multiple calculation nodes each include original parameters and intermediate parameters. Original parameters are the parameters that need to be input, while intermediate parameters are the parameters calculated based on the original parameters. Therefore, the input parameters of the calculation network are the original parameters in the calculation expressions corresponding to the multiple calculation nodes.

[0120] Figure 10 A schematic diagram illustrating the acquisition of parameter values ​​for measurement input parameters according to an embodiment of this specification is shown.

[0121] In some implementations, reference Figure 10 Among the multiple execution agents is a parameter acquisition agent configured to acquire input parameters.

[0122] The data processing system can utilize the measurement agent to invoke parameters to obtain the agent, and then obtain the parameter values ​​corresponding to the input parameters through the parameter-obtaining agent. For example, see [reference needed]. Figure 10 The data processing system can utilize a parameter acquisition agent to obtain the parameter values ​​corresponding to the input parameters from a target location. The target location includes at least one of the following: a knowledge base in the target domain; or at least one web page related to the target domain. For example, some input parameters are fixed values, which can be obtained from a knowledge base in the target domain. For instance, the knowledge base may include relevant literature in the target domain, such as papers and textbooks, and the parameter acquisition agent can search for relevant literature to obtain the parameter values ​​corresponding to the input parameters recorded in the literature. Alternatively, some input parameters may change over time; the parameter values ​​of these input parameters can be obtained from web pages related to the target domain, which periodically update these parameter values.

[0123] In some implementations, when the parameter value of the target parameter cannot be obtained at the target location, the parameter acquisition agent can also display a parameter input page on the workbench page or chat window page, and use the parameter input page to receive the parameter value of the target parameter input by the user.

[0124] In this embodiment, the data processing system utilizes an intelligent agent system to determine the parameters required for input into the measurement expression as the input parameters of the measurement network. Then, the data processing system uses a parameter acquisition agent to obtain the parameter values ​​corresponding to the input parameters from the target location. Furthermore, when the parameter acquisition agent cannot obtain the parameter value of the target parameter at the target location, the data processing system displays a parameter input page to receive the parameter value input by the user. This parameter acquisition method enables automatic acquisition and dynamic updating of parameter values, resulting in the use of more accurate parameter values ​​during the measurement process, thereby improving the reliability and real-time performance of the measurement results.

[0125] In some implementations, the data processing system can utilize an intelligent agent system to input the computation network, input data, and computation expressions corresponding to multiple computation nodes into a computation engine. The computation engine is configured to recognize and compute computation expressions for the target domain. Furthermore, the data processing system can use the computation engine to traverse and compute multiple computation nodes in the computation network to obtain computation data corresponding to the computation subtasks.

[0126] The computing engine provides computing services. It is configured with computing logic relevant to the target domain. For example, taking the energy domain as an example, the data processing system can first acquire computing knowledge of the target domain, such as computing logic and expression logic input by domain experts. Then, the data processing system uses LLM to convert the computing knowledge of the target domain, described in natural language, into an Abstract Syntax Tree (AST), and registers the AST with the computing engine. The computing engine can then provide computing services based on the computing logic recorded in the AST.

[0127] As an example, refer to Figure 7 The computational network serves as a guiding framework for the computational engine, directing it to traverse each computational node from left to right. At each node, the engine performs calculations based on the output parameter values ​​and the corresponding computational expressions. Since the AST (Analog Abstraction Syntax Tree) mandates consistent units of measurement for all parameters, the engine performs unit conversions on the parameters corresponding to the computational expressions in the target domain to ensure that all parameter values ​​have the same unit of measurement.

[0128] In some implementations, after obtaining the measurement data corresponding to the measurement subtask, the data processing system may also receive incremental requirement information for the measurement subtask. This incremental requirement information is used to instruct the data processing system to perform incremental measurement based on the measurement subtask. For example, the user may interact with the chat page again, instructing the data processing system to update the measurement data of the measurement subtask based on the incremental requirement information.

[0129] In this scenario, in response to receiving incremental requirement information for the measurement subtask, the data processing system can use the measurement agent to infer an incremental subnetwork based on the incremental requirement information, and then connect the incremental subnetwork to the measurement network to obtain an incremental measurement network. The incremental subnetwork and the measurement network share the same data interaction specifications. Next, the data processing system can use the measurement agent to obtain the incremental measurement expression and incremental input data. The incremental measurement expression is the measurement expression corresponding to the measurement node in the incremental subnetwork, and the incremental input data is the input data corresponding to the incremental subnetwork. Finally, the data processing system can use the measurement agent again to perform calculations based on the incremental measurement network, the input data of the measurement network, the measurement expression of the measurement network, the incremental input data, and the incremental measurement expression to obtain the measurement data for incremental measurement based on the measurement subtask.

[0130] In some embodiments, the incremental demand information is acquired in a manner similar to that of the demand information. The method by which the measurement agent obtains the incremental measurement network based on the incremental demand information is also similar to the method of obtaining the measurement network based on the demand information, and will not be elaborated upon here. Since the incremental measurement network and the measurement network are acquired in similar ways and share the same data interaction specifications, the incremental measurement network can be directly incorporated into the measurement network to obtain the incremental measurement network.

[0131] Figure 11 A schematic diagram of an incremental measurement network provided according to an embodiment of this specification is shown.

[0132] refer to Figure 11 The incremental subnetwork includes measurement nodes 12 to 16, as well as measurement nodes N+1 and N+2. The methods for obtaining the measurement expressions corresponding to the measurement nodes in the incremental subnetwork and for obtaining the incremental input data are similar to the methods for obtaining the measurement expressions corresponding to the measurement nodes in the measurement network and for obtaining the input data of the measurement network, and will not be elaborated upon here.

[0133] In this embodiment, the data processing system, in response to received incremental demand information, uses a computational agent to infer an incremental sub-network with the same data interaction specifications as the computational network. Then, the data processing system uses the computational agent to connect the incremental sub-network to the computational network, generating an incremental computational network. Next, the data processing system uses the computational agent to obtain the incremental computational expression and incremental input data, and executes the computational sub-task based on the incremental computational network, the input data and computational expression of the original computational network, and the incremental input data and incremental computational expression. In this way, when incremental computation is required for a computational sub-task, it is not necessary to rebuild the entire computational network; only the incremental network needs to be constructed to complete the incremental computation of the computational sub-task. This avoids the repeated inference process of the computational network, effectively saving the computing resources of the data processing system and improving the execution efficiency of incremental computation of computational sub-tasks.

[0134] In some embodiments, the plurality of executing agents includes a summarizing agent configured to summarize the measurement results. (See reference) Figure 6 The measurement agent 3 can be a summarizing agent. The summarizing agent can summarize the measurement data output by other measurement agents and generate measurement results in a specified format.

[0135] As an example, the data processing system can utilize a summarizing agent to obtain the summarizing requirements of a measurement task, and generate a third prompt instruction based on the summarizing requirements and measurement data corresponding to at least one measurement sub-task. This third prompt instruction guides a pre-trained summarizing language model to reason based on the summarizing requirements and generate measurement results. Furthermore, the data processing system can also use the summarizing agent to input the third prompt instruction into the summarizing language model, which then generates the measurement results.

[0136] Among them, the summary requirements can be integrated into the requirement information. The summary requirements can be requirements for the summary format, the output calculation result format, etc.

[0137] The third prompt instruction can be generated based on a template. This template can be pre-configured with text to guide the large language model in reasoning. For example, in the energy field, the following is an example of a third prompt instruction template: "You are a professional energy investment analyst. Based on the provided requirements, the calculation data of each calculation sub-task, and the summary requirements, please generate a calculation result report that meets the requirements."

[0138] Project Background Review Requirements: [Requirement Information] Summary of requirements: Please organize and present the report strictly according to the following requirements: Summary Format (Template): [Summarize requirements, summary format] Output format: [Summary of requirements, calculation result format] Available measurement data: [Measurement data for measurement subtask 1, measurement data for measurement subtask 2] Task: Generate a formatted report Based on all the information above, please use the template for [summarizing requirements and format] to generate a complete project calculation report.

[0139] The final output format of the assessment report must meet the requirements of [summary of requirements, assessment result format]. The summary agent can input the summary requirements and the measurement data corresponding to at least one measurement subtask into the template of the third prompt instruction to generate the third prompt instruction. Then, the summary agent can input the third prompt instruction into the summary language model to guide the summary language model to output the measurement results.

[0140] In some embodiments, when an anomaly occurs during the execution of the measurement process by the target agent, the agent system can instruct the planning agent to update the measurement process corresponding to the measurement task based on the requirement information and the anomaly. Furthermore, the agent system can invoke the target agent among multiple executing agents to execute the measurement process based on the updated measurement process.

[0141] In this embodiment, when an anomaly occurs during the execution of the calculation process by the target agent, the agent system can instruct the planning agent to update the calculation process based on the demand information and the anomaly. Then, the agent system can re-invoke the target agent among multiple executing agents based on the updated calculation process to execute the calculation task. In this way, the agent system automatically replans and optimizes the task flow when an anomaly occurs, automatically correcting erroneous processes. This enables the calculation task to autonomously recover and be executed, improving the fault tolerance of the data processing system and the success rate of task execution.

[0142] S330: Outputs the calculation results to the user.

[0143] In some embodiments, reference Figure 1 The data processing system can send the calculated results to the terminal device. After receiving the calculated results, the terminal device can display them on the page where the user inputs their requirements (e.g., the workbench page or chat window page).

[0144] In summary, the data processing method and system provided in this specification allow the data processing system to receive user-inputted demand information and invoke the intelligent agent system. Then, the data processing system utilizes the planning intelligent agent within the intelligent agent system to generate a calculation process based on the calculation tasks corresponding to the demand information. It then invokes multiple execution intelligent agents to perform decentralized data interaction based on the calculation process to execute the calculation process. When executing calculation tasks in this way, after the user provides demand information, the intelligent agent system can automatically analyze the demand information and execute the corresponding calculation tasks, achieving end-to-end automated calculation, effectively improving calculation efficiency and reducing the probability of calculation errors.

[0145] This specification, in another aspect, provides a computer-readable non-transitory storage medium storing at least one instruction set for performing a calculation task in a target domain. When the at least one instruction set is executed by a processor, it instructs the processor to implement the steps of the data processing method described in this specification. In some possible embodiments, various aspects of this specification can also be implemented as a program product comprising program code. When the program product is run on a computing system 200, the program code causes the computing system 200 to perform the steps of the data processing method described in this specification. The program product for implementing the above method may employ a portable compact disc read-only memory (CD-ROM) containing program code and may run on the computing system 200. However, the program product of this specification is not limited thereto. In this specification, the readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system. The program product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. The computer-readable storage medium may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable storage medium may also be any readable medium other than a readable storage medium that can send, propagate, or transmit programs for use by or in connection with an instruction execution system, apparatus, or device. Program code contained on a readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof. Program code for performing the operations described herein can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on computing system 200, partially on computing system 200, as a standalone software package, partially on computing system 200 and partially on a remote computing device, or entirely on a remote computing device.

[0146] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0147] In summary, after reading this detailed disclosure, those skilled in the art will understand that the foregoing detailed disclosure is presented by way of example only and is not restrictive. Although not explicitly stated herein, those skilled in the art will understand that this specification requires various reasonable changes, improvements, and modifications to the embodiments. These changes, improvements, and modifications are intended to be made by this specification and are within the spirit and scope of the exemplary embodiments described herein.

[0148] Furthermore, certain terms in this specification have been used to describe embodiments of this specification. For example, "an embodiment," "an embodiment," and / or "some embodiments" mean that a particular feature, structure, or characteristic described in connection with that embodiment may be included in at least one embodiment of this specification. Therefore, it is to be emphasized and understood that two or more references to "an embodiment" or "an embodiment" or "alternative embodiment" in various parts of this specification do not necessarily refer to the same embodiment. Moreover, specific features, structures, or characteristics may be suitably combined in one or more embodiments of this specification.

[0149] It should be understood that in the foregoing description of the embodiments in this specification, various features are combined in a single embodiment, drawing, or description for the purpose of simplifying the description and to aid in understanding a feature. However, this does not mean that the combination of these features is necessary, and those skilled in the art, upon reading this specification, may readily identify some of the devices as separate embodiments. That is, the embodiments in this specification can also be understood as an integration of multiple secondary embodiments. And the content of each secondary embodiment is valid even if it contains fewer than all the features of a single foregoing disclosed embodiment.

[0150] Every patent, patent application, publication of a patent application, and other material such as articles, books, specifications, publications, documents, articles, etc., cited herein, except for those inconsistent with or conflicting with this document, or those having a restrictive effect on the widest scope of the claims, may be incorporated herein by reference for all purposes now or hereafter associated with this document. Furthermore, in the event of any inconsistency or conflict between the description, definition, and / or use of relevant terms in any material and the description, definition, and / or use of relevant terms in this document, the terms in this document shall prevail.

[0151] Finally, it should be understood that the embodiments disclosed herein are illustrative of the principles of the embodiments described in this specification. Other modified embodiments are also within the scope of this specification. Therefore, the embodiments disclosed in this specification are merely examples and not limitations. Those skilled in the art can implement the applications described in this specification using alternative configurations based on the embodiments in this specification. Therefore, the embodiments in this specification are not limited to the embodiments precisely described in the applications.

Claims

1. A data processing method, the method comprising: Receive user input request information, which is used to request the execution of a measurement task in the target domain; Based on the aforementioned demand information, an intelligent agent system is invoked to obtain the measurement results corresponding to the measurement task output by the intelligent agent system. The intelligent agent system includes a planning intelligent agent and multiple execution intelligent agents. The planning intelligent agent generates a measurement process corresponding to the measurement task based on the demand information, and invokes a target intelligent agent among the multiple execution intelligent agents to execute the measurement process, thereby obtaining the measurement results. The target intelligent agents can perform decentralized data interaction based on the measurement process. The calculation results are output to the user.

2. The method according to claim 1, wherein, The planning agent is trained to focus on the measurement and planning tasks of the target domain; The step of generating the calculation process corresponding to the calculation task based on the requirement information includes: Based on the demand information, determine the calculation task corresponding to the demand information, and generate the calculation process required to execute the calculation task based on the calculation task.

3. The method according to claim 2, wherein, The calculation process required to execute the calculation task based on the calculation task includes: Based on the calculation task, a search is performed in a preset process library. When a candidate calculation process corresponding to the calculation task is found in the preset process library, the candidate calculation process matching the calculation task is determined as the calculation process required to execute the calculation task. The process library stores candidate calculation processes corresponding to different calculation tasks. When there is no candidate calculation process corresponding to the calculation task in the preset process library, the calculation task is input into the pre-trained planning language model, which guides the planning language model to reason based on the calculation task to obtain the calculation process corresponding to the calculation task.

4. The method according to claim 3, wherein, The calculation process includes at least one calculation subtask and the execution logic between the at least one calculation subtask. The intelligent agent system stores the registration information of the multiple execution intelligent agents. The registration information is used to represent the task type and interaction rules corresponding to the multiple execution intelligent agents respectively. The step of inputting the calculation task into a pre-trained planning large language model and guiding the planning large language model to perform inference based on the calculation task to obtain the calculation process corresponding to the calculation task includes: Based on the measurement task and the registration information of the multiple execution agents, a first prompt instruction is generated. This first prompt instruction guides the planning language model to determine at least one measurement sub-task to implement the measurement task based on the task types corresponding to the measurement task and the multiple execution agents, and to determine the execution logic of the at least one measurement sub-task based on the interaction rules corresponding to the multiple execution agents. Input the first prompt command into the planning language model to obtain the calculation process generated by the planning language model.

5. The method according to any one of claims 1-4, wherein, The calculation process includes at least one calculation subtask and execution logic between the at least one calculation subtask. The plurality of execution agents include a calculation agent configured to execute the calculation subtask. Different calculation agents are used to execute calculation subtasks of different task types. The step of calling the target agent among the multiple execution agents to execute the calculation process based on the calculation process, and obtaining the calculation result, includes: Based on the task type corresponding to each sub-task in the measurement process, the target agent corresponding to each sub-task is determined from the measurement agent; Based on the execution logic between the at least one measurement sub-task, the target agent executes the measurement sub-task to obtain measurement data corresponding to each measurement sub-task; and Based on the measurement data corresponding to each measurement subtask, the measurement result corresponding to the measurement task is generated.

6. The method according to claim 5, wherein, The execution logic based on the at least one measurement sub-task, utilizing the target agent to execute the measurement sub-task, and obtaining measurement data corresponding to each measurement sub-task, includes: A call request is initiated to the target intelligent agent, and the measurement network corresponding to the measurement subtask is obtained by reasoning using the target intelligent agent. The measurement network includes multiple measurement nodes and the topological relationship between the multiple measurement nodes. Obtain the measurement expressions corresponding to the plurality of measurement nodes respectively, and obtain the input data of the measurement network based on the measurement expressions corresponding to the plurality of measurement nodes respectively; and The calculation data corresponding to the calculation subtask is obtained by performing calculations based on the calculation network, the input data, and the calculation expressions corresponding to the multiple calculation nodes.

7. The method according to claim 6, wherein, Each of the aforementioned measurement nodes is used to represent a computational dimension; The step of obtaining the measurement expressions corresponding to the plurality of measurement nodes, and obtaining the input data of the measurement network based on the measurement expressions corresponding to the plurality of measurement nodes, includes: Based on the calculation dimensions corresponding to the multiple calculation nodes, the calculation expression corresponding to each calculation node is obtained from the pre-configured expression library, which stores the calculation expressions corresponding to multiple calculation dimensions. The parameters that need to be input into the measurement expressions corresponding to the multiple measurement nodes are determined as the input parameters of the measurement network; and Obtain the parameter value corresponding to the input parameter, and obtain the input data based on the parameter value of the input parameter.

8. The method according to claim 6, wherein, After obtaining the measurement data corresponding to the measurement subtask, the method further includes: In response to receiving incremental requirement information for the calculation subtask, the incremental requirement information is used to instruct the execution of incremental calculation based on the calculation subtask; Using the measurement agent, the incremental sub-network is obtained based on the incremental demand information, and the incremental sub-network is connected to the measurement network to obtain the incremental measurement network, wherein the incremental sub-network and the measurement network have the same data interaction specifications; Obtain the incremental calculation expression and incremental input data, wherein the incremental calculation expression is the calculation expression corresponding to the calculation node in the incremental sub-network, and the incremental input data is the input data corresponding to the incremental sub-network; and The incremental measurement network, the input data of the measurement network, the measurement expression of the measurement network, the incremental input data, and the incremental measurement expression are used to calculate the measurement data obtained by performing incremental measurement based on the measurement subtask.

9. A data processing system, comprising: At least one storage medium storing at least one instruction set for performing measurement tasks in the target domain; as well as At least one processor is communicatively connected to the at least one storage medium, wherein, when the data processing system is running, the at least one processor reads the at least one instruction set and implements the method as described in any one of claims 1-8 according to the instructions of the at least one instruction set.

10. A computer-readable non-volatile storage medium, wherein, The computer-readable non-volatile storage medium stores at least one instruction set, which, when executed by at least one processor, implements the method as described in any one of claims 1-8.

Images