A diagnostic analysis thought chain management method and system based on human-computer interaction

Abstract

This invention provides a diagnostic analysis thought chain management method and system based on human-computer interaction. The method responds to the user's diagnostic scenario selection command, calls a preset thought chain template to construct an initial diagnostic analysis thought chain, and displays it through a visual interactive interface. Based on the user's interactive dynamic adjustment of the initial diagnostic analysis thought chain through the visual interactive interface, an adjusted diagnostic analysis thought chain is obtained. This dynamic adjustment includes dynamic editing of thought chain nodes and associating the required data items of the nodes. Professional content is embedded into the adjusted diagnostic analysis thought chain and a pre-built professional knowledge base to obtain an embedded diagnostic analysis thought chain. Based on the embedded diagnostic analysis thought chain and a prompt word rule base, a prompt word sequence is generated through a pre-trained large model, solving the problems of uncontrollable reasoning processes and strong randomness of results in existing technologies for large model applications in professional fields.

Description

Technical Field

[0001] This invention relates to the field of diagnostic analysis, and more specifically to a diagnostic analysis thought chain management method and system based on human-computer interaction. Background Technology

[0002] Power grid diagnostic analysis is a complex task involving large amounts of data, a heavy workload, and high technical requirements. Comprehensive and scientific power grid diagnostic analysis is a crucial foundation for promoting the scientific planning and efficient development of the power grid. Power grid diagnostic analysis involves numerous internal and external factors, requiring both an overall diagnostic result and detailed quantitative conclusions at key points, which is very challenging. Due to limitations in technology, manual processing speed, and scale, traditional large-scale model applications rely on single, static prompts, making it difficult to handle the complex, multi-step, multi-branch reasoning tasks in power grid diagnosis. Experts in specific fields cannot effectively "transmit" their analytical logic and processes to large-scale models, resulting in insufficient professionalism and low reliability in the model outputs. Current technologies lack mechanisms for human-computer interaction during the reasoning process, failing to introduce human expert judgment and guidance at key decision points, leading to insufficient flexibility and accuracy in automated diagnosis.

[0003] As large-scale models are increasingly applied across various professional fields, thought chain technology is widely used to enhance model reasoning capabilities. Existing technologies mostly employ predefined prompt templates or rule engines to generate thought chains, lacking the dynamic adjustment capabilities of human-machine collaboration. While knowledge base systems can provide terminology explanations, they are not deeply integrated with the thought chain construction process, resulting in limitations in the model's ability to understand and generate thought chains within specialized contexts.

[0004] Currently, traditional methods for the diagnostic and analysis of power distribution networks mainly suffer from the following problems: First, existing solutions use static, linear prompt word templates to guide large models, which cannot adapt to the various reasoning paths and uncertainties that may occur during the diagnostic process, resulting in a rigid analysis process when facing complex and atypical cases.

[0005] Second, the reasoning process lacks a mechanism to incorporate human expert judgment into the reasoning chain. At key decision points, the model may go in the wrong direction of reasoning due to bias in training data or misunderstanding, and this cannot be corrected in time, resulting in inaccurate or unreliable final conclusions.

[0006] Third, the analytical methodologies and experiential knowledge of experts cannot be systematically learned and reused in a structured manner; they can only be reflected in the initial template design.

[0007] Fourth, the entire inference chain is not transparent to users, making it difficult for them to understand how the model arrives at the final conclusion.

[0008] Therefore, finding a solution to the above problems is of great significance. Summary of the Invention

[0009] To address the problems of uncontrollable reasoning processes, highly random results, and difficulty in incorporating human expert experience and knowledge in existing technologies for large-scale model applications in professional fields, as well as the difficulty of traditional static prompt word methods in handling complex multi-step reasoning tasks, this invention proposes a diagnostic analysis thinking chain management method and system based on human-computer interaction.

[0010] Firstly, a diagnostic analysis thinking chain management method based on human-computer interaction is provided, including: In response to the user's diagnostic scenario selection command, the system calls the preset thinking chain template to construct the initial diagnostic analysis thinking chain and displays it through a visual interactive interface. Based on the user's interactive dynamic adjustment of the initial diagnostic analysis thinking chain through the visual interactive interface, the adjusted diagnostic analysis thinking chain is obtained. The interactive dynamic adjustment includes dynamic editing of the thinking chain nodes and associating the data items required by the nodes. Based on the adjusted diagnostic analysis thinking chain and the pre-built professional knowledge base, professional content is embedded to obtain the embedded diagnostic analysis thinking chain. Based on the embedded diagnostic analysis thought chain and the prompt word rule base, a prompt word sequence is generated through a pre-trained large model.

[0011] Secondly, a diagnostic analysis thought chain management system based on human-computer interaction is provided, including: The thought chain building module is used to respond to the user's diagnostic scenario selection command, call the preset thought chain template to build the initial diagnostic analysis thought chain, and display it through a visual interactive interface; The data item mounting module is used to obtain the adjusted diagnostic analysis thinking chain based on the user's human-computer interactive dynamic adjustment operation on the initial diagnostic analysis thinking chain through the visual interactive interface. The human-computer interactive dynamic adjustment includes dynamic editing of the thinking chain nodes and associating and mounting the data items required by the nodes. The knowledge base interaction module is used to embed professional content based on the adjusted diagnostic analysis thinking chain and the pre-built professional knowledge base to obtain the embedded diagnostic analysis thinking chain. The intelligent prompt word generation module is used to generate prompt word sequences based on the embedded diagnostic analysis thought chain and prompt word rule library through a pre-trained large model.

[0012] In another aspect, this application also provides an electronic device, comprising: at least one processor and a memory; the memory and the processor are connected via a bus; The memory is used to store one or more programs; When the one or more programs are executed by the at least one processor, a human-computer interaction-based diagnostic analysis thinking chain management method as described above is implemented.

[0013] Furthermore, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed, implements the human-computer interaction-based diagnostic analysis thought chain management method as described above.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention provides a diagnostic analysis thought chain management method and system based on human-computer interaction. The method responds to the user's diagnostic scenario selection command, calls a preset thought chain template to construct an initial diagnostic analysis thought chain, and displays it through a visual interactive interface. Based on the user's interactive dynamic adjustment of the initial diagnostic analysis thought chain through the visual interactive interface, an adjusted diagnostic analysis thought chain is obtained. This interactive dynamic adjustment includes dynamic editing of thought chain nodes and associating the required data items of the nodes. Professional content is embedded into the adjusted diagnostic analysis thought chain and a pre-built professional knowledge base to obtain an embedded diagnostic analysis thought chain. Based on the embedded diagnostic analysis thought chain and the prompt word rule base, a prompt word sequence is generated through a pre-trained large model. This solves the problems of uncontrollable reasoning processes, strong randomness of results, difficulty in incorporating human expert experience and knowledge, and the inability of traditional static prompt word methods to handle complex multi-step reasoning tasks in the application of large models in professional fields.

[0015] Specifically, this invention, based on a large model, introduces real-time expert judgment through human-computer interaction, making the construction of thought chains more flexible and accurate, while simultaneously realizing the digitization and process-oriented accumulation of domain expert knowledge. By leveraging a data item mounting mechanism, data retrieval time is significantly reduced; relying on a professional knowledge base, terminology is automatically supplemented, enhancing the large model's understanding of professional expressions and context; and intelligent generation of prompt words through the large model reduces manual writing costs and improves overall efficiency. Attached Figure Description

[0016] Figure 1 The flowchart shows the diagnostic analysis thinking chain management method based on human-computer interaction of the present invention. Figure 2 This is a schematic diagram of the human-computer interaction-based diagnostic analysis thinking chain management system of the present invention; Figure 3 This is a schematic diagram illustrating the specific process of the diagnostic analysis thinking chain management method based on human-computer interaction of the present invention. Figure 4 This is a schematic diagram of an electronic device structure according to the present invention. Detailed Implementation

[0017] The purpose of this invention is to provide a human-computer interaction-based diagnostic analysis thought chain management system to improve the reasoning accuracy and practicality of large models in professional scenarios. First, it enables dynamic management and advancement of the thought chain, upgrading the diagnostic process from a static template to dynamically configurable process instances. Second, it introduces an effective human-computer collaboration mechanism, integrating expert experience into the automated analysis process in real time, and solidifying excellent diagnostic processes and methodologies into configurable "thought chain templates," enabling the accumulation and reuse of expert knowledge and improving the overall analytical capabilities of the system. Third, it enhances process transparency and result credibility by visually displaying the entire thought chain advancement process and involving users. Fourth, it automatically retrieves and supplements terminology explanations and contextual information from the knowledge base.

[0018] To better understand the present invention, the following description, in conjunction with the accompanying drawings and embodiments, will further illustrate the content of the present invention.

[0019] Example 1: A diagnostic analysis thinking chain management method based on human-computer interaction, such as Figure 1 As shown, it includes: Step 1: In response to the user's diagnostic scenario selection command, call the preset thinking chain template to construct the initial diagnostic analysis thinking chain, and display it through a visual interactive interface; Step 2: Based on the user's interactive dynamic adjustment of the initial diagnostic analysis thought chain through the visual interface, the adjusted diagnostic analysis thought chain is obtained. Step 3: Based on the adjusted diagnostic analysis thinking chain and the pre-built professional knowledge base, professional content is embedded to obtain the embedded diagnostic analysis thinking chain; Step 4: Based on the embedded diagnostic analysis thought chain and the prompt word rule base, generate prompt word sequences through a pre-trained large model.

[0020] Among them, the human-computer interactive dynamic adjustment includes the dynamic editing of thought chain nodes and the association and mounting of data items required by the nodes.

[0021] In this embodiment, step 2 provides a visual interface to enable users to edit the thought chain in real time. Based on the user's editing of nodes in the visual interface, the reasoning process is automatically paused and awaits manual review and confirmation, increasing the user's editing freedom. Specifically, this includes: In response to user editing operations on nodes in the visual interface, perform add, delete, or modify operations on the nodes; In response to the user's node insertion operation in the thought chain, a manual review node is inserted at the target position in the thought chain. When the large model inference reaches the manual review node, the inference process is automatically paused and the review is awaited. Based on the initial inference feedback of the large model, logical breakpoints or deviation nodes are automatically marked, and node optimization suggestions are pushed to the user. The node information is then corrected based on the user's correction operations based on the node optimization suggestions.

[0022] In this embodiment, to achieve accurate association between the thought chain and multi-source data in step 2, data items can be mounted in multiple ways, either through automatic matching by the system or manual selection by the user. Specifically, this includes: Based on the reasoning requirement tags of the thought chain nodes, the appropriate data items are retrieved from the business database and the appropriate data items are fed back to the user. In response to user confirmation or manual selection, the system maps and mounts the thought chain nodes and data items, and performs quality verification on the mounted data items. When a critical data item is detected to be missing, an automatic data backtracking process is triggered. If the backtracking fails, the user is prompted to manually complete the data, and the missing information is recorded in the business database.

[0023] In this embodiment, in step 3, to achieve automatic retrieval and explanation of professional terms in the thought chain and improve the coherence of the diagnostic process, the system can identify professional terms in the thought chain and automatically extract and embed corresponding explanatory information, etc., specifically including: Identify the technical terms in the adjusted diagnostic analysis thinking chain, and extract the corresponding explanatory information from the pre-built professional knowledge base. The explanatory information includes definitions, explanations, applicable scenarios, or standard clauses. The extracted explanatory information is embedded as an explanatory node into the adjusted diagnostic analysis thought chain to obtain the embedded diagnostic analysis thought chain.

[0024] In this embodiment, to improve the intelligence and accuracy of prompt word generation in step 4, the structured thought chain can be transformed into a dialogue sequence that the large model can understand, guiding the large model to optimize the thought chain reasoning logic. Specifically, this includes: Based on the diagnostic scenario, the pre-set prompt word generation rule base is invoked to transform the embedded diagnostic analysis thought chain into a dialogue sequence that the large model can understand. Then, based on the pre-trained large model, semantic optimization is performed to generate initial prompt words. The system generates a sequence of prompt words in response to the user's editing and adjustment of the initial prompt words, and records the user's modification preferences to optimize subsequent prompt word generation rules.

[0025] In this embodiment, to ensure the professionalism and reliability of the reasoning process, a thought chain review process can be introduced, which is triggered after the thought chain is edited or modified. Specifically, this includes: The revised diagnostic analysis thought chain, reasoning path, analysis results and related data are packaged and sent to the designated expert for review, and a new version of the thought chain is generated after the review is approved. By comparing the differences between the new version of the thinking chain and the adjusted diagnostic analysis thinking chain, the added checkpoints and judgment logic are extracted and incorporated into the thinking chain template library after verification.

[0026] This invention, based on a large model, introduces real-time expert judgment through human-computer interaction, making the construction of thought chains more flexible and accurate, while simultaneously realizing the digitization and process-oriented accumulation of domain expert knowledge. Utilizing a data item mounting mechanism, it significantly shortens data retrieval time; relying on a professional knowledge base, it automatically supplements nouns and terms, enhancing the large model's ability to understand professional expressions and context; and through intelligent generation of prompt words by the large model, it reduces manual writing costs and improves overall efficiency.

[0027] Example 2: This invention proposes a diagnostic analysis thought chain management system based on human-computer interaction, such as... Figure 2 As shown, it includes a mind chain construction module, a data item mounting module, a mind chain review process module, a knowledge base interaction module, and a prompt word intelligent generation module, which specifically include: The thought chain building module is used to respond to the user's diagnostic scenario selection command, call the preset thought chain template to build the initial diagnostic analysis thought chain, and display it through a visual interactive interface; The data item mounting module is used to obtain the adjusted diagnostic analysis thinking chain based on the user's human-computer interactive dynamic adjustment operation on the initial diagnostic analysis thinking chain through the visual interactive interface. The human-computer interactive dynamic adjustment includes dynamic editing of the thinking chain nodes and associating and mounting the data items required by the nodes. The knowledge base interaction module is used to embed professional content based on the adjusted diagnostic analysis thinking chain and the pre-built professional knowledge base to obtain the embedded diagnostic analysis thinking chain. The intelligent prompt word generation module is used to generate prompt word sequences based on the embedded diagnostic analysis thought chain and prompt word rule library through a pre-trained large model.

[0028] In this embodiment, the MindChain review process module is specifically used for: The revised diagnostic analysis thought chain, reasoning path, analysis results and related data are packaged and sent to the designated expert for review, and a new version of the thought chain is generated after the review is approved. By comparing the differences between the new version of the thinking chain and the adjusted diagnostic analysis thinking chain, the added checkpoints and judgment logic are extracted and incorporated into the thinking chain template library after verification.

[0029] The diagnostic analysis thinking chain management system of the present invention is described below with a specific implementation process, such as Figure 3 As shown, it specifically includes: 2.1 Mind Chain Construction Module The thought chain building module provides users with a visual interface for editing and optimizing thought chains, supporting the creation of thought chains, node editing, logic adjustment, and real-time feedback, enabling human-computer collaborative optimization of the thought chain reasoning process. The specific process is as follows: (1) Initialization of thinking chain construction: Users can preset thinking chain templates based on the diagnostic scenario. For example, when the user selects the task type "diagnosis of medium voltage grid structure in power distribution network", the system automatically loads the preset thinking chain template for medium voltage grid structure diagnosis, which includes five basic nodes: "data acquisition → structural standardization judgment → key parameter analysis → problem attribution → improvement suggestions".

[0030] (2) Dynamic Node Editing and Branch Expansion: The system supports users to add, delete, modify, and query nodes in the thought chain. For example, when analyzing a 10kV line, the system initially determines that it is a "non-standard structure". Users can use the thought chain editing function to select "Add Thought Chain Node", insert the "Historical Modification Status Verification" sub-node, and associate it with related projects. The system supports adding "Manual Review Points" at any node. When the reasoning process reaches this point, it will automatically pause and prompt experts to intervene.

[0031] (3) Real-time feedback optimization: Based on the reasoning results of the large model, the logical rationality of the thinking chain is initially verified, possible logical breakpoints or deviation nodes are marked, and optimization suggestions are pushed to the user (such as supplementing a certain type of data node or adjusting the reasoning order). The user can directly correct the node information or logical relationship on the interface according to the feedback, forming a closed-loop interaction mechanism of "model reasoning - user correction - model re-optimization".

[0032] 2.2 Mounting Data Item Module The data item mounting module enables precise association between thought chain nodes and multi-source data, supports rapid data retrieval and real-time updates, and ensures that thought chain reasoning has reliable data support.

[0033] (1) Data item classification management: The system has a built-in data item classification system, covering diagnostic indicator data (such as grid voltage and equipment load rate), equipment file data (such as equipment model and operating years), etc., and supports users to customize and add new data item types; (2) Mounting Mechanism: Users can mount data items through two methods: manual association or automatic matching. When manually associating, users select the target data item in the mind chain node editing interface; when automatically matching, the system retrieves relevant data items (such as transformer load rate data and historical overload records) from the database based on the node reasoning requirements (such as the "Analyze the causes of transformer overload" node) and recommends suitable data items for users to confirm mounting.

[0034] (3) Data quality verification and supplementation: For example, when the key data item [segment capacity] of a certain node is detected to be empty, the system automatically triggers the data completion process: First, it tries to obtain real-time load data from the distribution automation system to back-calculate. If it is still not possible, it prompts the user to manually input the data and records the missing data to the knowledge base.

[0035] 2.3 Mind Chain Review Process Module The MindChain review process module establishes a human-machine collaborative review mechanism to ensure the professionalism and reliability of the reasoning process.

[0036] (1) Mind Chain Review Trigger: When the mind chain is edited or modified, the mind chain review process is triggered. The system packages the current reasoning path, the analyzed results and related data and sends them to the designated expert. After the review is approved, the diagnostic analysis calls the modified and improved mind chain.

[0037] (2) Version management and knowledge accumulation: Each important modification to the thinking chain will generate a new version. The system will automatically compare the version differences, extract the checkpoints and judgment logic added by experts, and integrate them into the standard thinking chain template library after verification, so as to realize the continuous accumulation of expert experience.

[0038] 2.4 Knowledge Base Interaction Module The knowledge base interaction module is linked with the diagnostic analysis professional knowledge base to realize the automatic retrieval and explanation of professional terms in the thought process, thereby improving the coherence of the diagnostic process.

[0039] (1) Automatic embedding of terminology explanations: When the system recognizes professional terms such as “multi-segment moderate communication”, it automatically extracts explanatory information such as definitions and applicable scenarios from the knowledge base. Users can choose to permanently embed them into the mind chain as explanation nodes.

[0040] (2) Real-time reference to standards and specifications: When the “Structural Standardization Judgment” node is executed, the system automatically associates the relevant clauses in the “Q / GDW10738-2020 Technical Guidelines for Distribution Network Planning and Design” and provides quick access in the form of hyperlinks to ensure the authority and transparency of the judgment basis.

[0041] (3) Intelligent push of typical cases: When the problem of "insufficient number of contacts" is diagnosed, the system automatically retrieves historical cases with similar problems from the knowledge base, displays their solutions and implementation effects, and provides reference for the current diagnosis.

[0042] 2.5 Intelligent Prompt Word Generation Module The intelligent prompt word generation module transforms structured thought chains into dialogue sequences that can be understood by large models based on diagnostic scenarios, guiding the large models to optimize the reasoning logic of thought chains and improving the intelligence and accuracy of prompt word generation.

[0043] (1) Prompt word generation rule base: Construct a prompt word generation rule base, which includes prompt word templates for different scenarios (such as the diagnostic scenario template: "Based on {data item 1} and {data item 2}, analyze the possible reasons for {node reasoning target} and sort out the reasoning logic chain") and logic optimization rules (such as prioritizing the association of data items and highlighting key reasoning nodes).

[0044] (2) Intelligent generation and optimization: Based on the identified scene, needs and thought chain information, the system calls the rule base to generate initial prompt words, and then performs semantic optimization through a large model to ensure that the prompt words are accurate and logically clear; users can edit and adjust the generated prompt words, and the system records the user's modification preferences to optimize the subsequent prompt word generation effect; at the same time, the prompt words can be dynamically adjusted according to the update of the thought chain to achieve synchronous optimization of prompt words and thought chain.

[0045] The present invention has the following beneficial effects: (1) Human-computer interaction-based diagnostic analysis thinking chain optimization mechanism: Using a visual interactive interface as a carrier, the system supports users in efficiently creating, modularizing, and dynamically adjusting the thinking chain. During this process, the system can introduce expert experience in real time, integrating expert knowledge into the thinking chain construction process, ultimately forming a closed-loop optimization system of "model initial reasoning to generate basic thinking chain → user correction and optimization based on actual scenarios and expert experience → model secondary verification and iterative improvement based on correction results". This mechanism not only improves the convenience and flexibility of thinking chain construction, but also ensures the scientific nature, adaptability, and practical value of the diagnostic analysis thinking chain through human-computer collaboration and dynamic verification, providing reliable logical support for accurate diagnosis.

[0046] (2) Method for mounting thought chain nodes and data items: A precise mounting scheme is provided. On the one hand, it can intelligently associate data item attributes with the reasoning logic tags of thought chain nodes, reducing the cost of manual operation. On the other hand, it retains the flexible manual association function, allowing users to manually adjust the mapping relationship between data items and nodes according to the actual diagnostic scenario requirements. Through the collaboration of "automatic + manual" dual modes, the deep integration of data and reasoning process is achieved, ensuring the accuracy and flexibility of data support during the reasoning process.

[0047] (3) Automatic embedding technology for terminology recognition and knowledge base interpretation: It can link the knowledge base for intelligent retrieval and matching, quickly retrieve the standard definition and other interpretation content of the corresponding terminology, and automatically embed it into the diagnostic thinking chain, effectively improving the coherence and accuracy of the diagnostic decision-making logic and reducing process blockage caused by terminology misunderstanding or information loss.

[0048] (4) Intelligent generation of prompt words: It can track the progress of the thought chain in real time and intelligently adjust the content of prompt words according to the changes in the thought chain during the diagnosis process to ensure accurate matching of the current analysis scenario, avoid information redundancy, eliminate the lack of key information, and achieve high efficiency and accuracy in the diagnosis process.

[0049] Example 3 like Figure 4 As shown, the present invention also provides an electronic device, which may be a computer device, a microcontroller device, a smart mobile device, etc. The electronic device in this embodiment may include a processor, a memory, a transceiver component, etc. The memory, processor, and transceiver component are connected via a bus; the memory can be used to store executable programs, and an exemplary executable program may include instructions; the processor is used to execute the instructions stored in the memory. The memory can also be used to store data, which can be accessed and / or modified when instructions are executed.

[0050] The processor may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, and it is suitable for implementing one or more instructions. Specifically, it is suitable for loading and executing one or more instructions in the storage medium to realize the corresponding method flow or corresponding function, so as to realize the steps of the human-computer interaction-based diagnostic analysis thinking chain management method in the above embodiments.

[0051] Example 4 Based on the same inventive concept, this invention also provides a readable storage medium, specifically an electronic device readable storage medium (Memory). This readable storage medium is a memory device within an electronic device used to store programs and data. It is understood that the storage medium here can include both built-in storage media within the electronic device and extended storage media supported by the electronic device. The storage medium provides storage space, which stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more executable programs (including program code). It should be noted that the storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. Loading and executing one or more instructions stored in the storage medium by the processor can implement the steps of the human-computer interaction-based diagnostic analysis thought chain management method described in the above embodiments.

[0052] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0053] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0054] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1The function specified in one or more boxes.

[0055] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0056] The above are merely embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of the claims of the present invention pending approval.

Claims

1. A diagnostic analysis thought chain management method based on human-computer interaction, characterized in that, The method comprises the following steps: In response to the user's diagnostic scene selection instruction, a preset thinking chain template is called to construct an initial diagnostic analysis thinking chain, and the initial diagnostic analysis thinking chain is displayed through a visual interactive interface; Based on the user's human-computer interactive dynamic adjustment operation on the initial diagnostic analysis thinking chain in the visual interactive interface, an adjusted diagnostic analysis thinking chain is obtained, and the human-computer interactive dynamic adjustment includes dynamic editing of a thinking chain node and association and mounting of a required data item of the node; Professional content embedding is performed on the adjusted diagnostic analysis thinking chain and a pre-constructed professional knowledge base to obtain an embedded diagnostic analysis thinking chain; Based on the embedded diagnostic analysis thinking chain and a prompt word rule library, an initial prompt word is generated through a pre-trained large model.

2. The method of claim 1, wherein, The dynamic editing of the thinking chain node in the human-computer interactive dynamic adjustment comprises: In response to the user's editing operation on the node in the visual interface, an add, delete or modify operation is performed on the node; In response to the user's node insertion operation on the thinking chain, an artificial audit point node is inserted at a target position of the thinking chain, and when the large model reasoning is executed to the artificial audit point node, the reasoning process is automatically paused and waits for audit confirmation; Based on the preliminary reasoning feedback of the large model, a logical breakpoint or deviation node is automatically marked, a node optimization suggestion is pushed to the user, and the node information is corrected based on the user's correction operation on the node optimization suggestion.

3. The method of claim 1, wherein, The association and mounting of the required data item of the node in the human-computer interactive dynamic adjustment comprises: Based on the reasoning requirement label of the thinking chain node, an adaptive data item in a business database is retrieved, and the adaptive data item is fed back to the user; In response to the user's confirmation or manual selection operation, the thinking chain node and the data item are mapped and mounted, and the mounted data item is quality checked; When a key data item is detected to be missing, an automatic data backtracking process is triggered, if the backtracking fails, the user is prompted to manually complete, and the missing condition is recorded to the business database.

4. The method of claim 1, wherein, The professional content embedding on the adjusted diagnostic analysis thinking chain and the pre-constructed professional knowledge base to obtain the embedded diagnostic analysis thinking chain comprises: Professional terms in the adjusted diagnostic analysis thinking chain are identified, and explanation information corresponding to the professional terms is extracted from the pre-constructed professional knowledge base, the explanation information including definition explanation, applicable scene or standard clause; The extracted explanation information is embedded as an explanatory node into the adjusted diagnostic analysis thinking chain to obtain the embedded diagnostic analysis thinking chain.

5. The method of claim 1, wherein, The generation of the prompt word sequence based on the embedded diagnostic analysis thinking chain and the prompt word rule library through the pre-trained large model comprises: According to the preset prompt word generation rule library, the embedded diagnostic analysis thinking chain is converted into a dialogue sequence understandable by the large model, and semantic optimization is performed based on the pre-trained large model to generate an initial prompt word; In response to the user's editing adjustment operation on the initial prompt word, a prompt word sequence is generated, and the user's modification preference is recorded to optimize subsequent prompt word generation rules.

6. The method of claim 1, wherein, After obtaining the adjusted diagnostic analysis thought chain by dynamically adjusting the initial diagnostic analysis thought chain through a visual interactive interface, the process further includes: The revised diagnostic analysis thought chain, reasoning path, analysis results and related data are packaged and sent to the designated expert for review, and a new version of the thought chain is generated after the review is approved. By comparing the differences between the new version of the thinking chain and the adjusted diagnostic analysis thinking chain, the added checkpoints and judgment logic are extracted and incorporated into the thinking chain template library after verification.

7. A diagnostic analysis thought chain management system based on human-computer interaction, characterized by, include: The thought chain building module is used to respond to the user's diagnostic scenario selection command, call the preset thought chain template to build the initial diagnostic analysis thought chain, and display it through a visual interactive interface; The data item mounting module is used to obtain the adjusted diagnostic analysis thinking chain based on the user's human-computer interactive dynamic adjustment operation on the initial diagnostic analysis thinking chain through the visual interactive interface. The human-computer interactive dynamic adjustment includes dynamic editing of the thinking chain nodes and associating and mounting the data items required by the nodes. The knowledge base interaction module is used to embed professional content based on the adjusted diagnostic analysis thinking chain and the pre-built professional knowledge base to obtain the embedded diagnostic analysis thinking chain. The intelligent prompt word generation module is used to generate prompt word sequences based on the embedded diagnostic analysis thought chain and prompt word rule library through a pre-trained large model.

8. The system of claim 7, wherein, The system also includes a mind chain review process module, used for: The revised diagnostic analysis thought chain, reasoning path, analysis results and related data are packaged and sent to the designated expert for review, and a new version of the thought chain is generated after the review is approved. By comparing the differences between the new version of the thinking chain and the adjusted diagnostic analysis thinking chain, the added checkpoints and judgment logic are extracted and incorporated into the thinking chain template library after verification.

9. An electronic device, comprising: include: At least one processor and memory; The memory and processor are connected via a bus; The memory is used to store one or more programs; When the one or more programs are executed by the at least one processor, the human-computer interaction-based diagnostic analysis thinking chain management method as described in any one of claims 1 to 6 is implemented.

10. A readable storage medium, characterized by, It contains an executable program, which, when executed, implements the human-computer interaction-based diagnostic analysis thinking chain management method as described in any one of claims 1 to 6.

Images