Intelligent guide generation and application method, system and equipment based on error analysis

Abstract

The invention discloses an intelligent guide generation and application method, system and equipment based on error analysis, belongs to the technical field of intelligent query system optimization, and is used for solving the problems that an existing intelligent query system optimization method is difficult to cover all possible error conditions, is difficult to quickly adapt to newly occurring error types, and is poor in practicability. And a complex error processing mechanism increases the complexity of the system. The method comprises the following steps: analyzing errors occurring in a workflow execution process of the intelligent query system based on a workflow error analysis large model to obtain an error analysis result; generating a corresponding system optimization guide based on the error analysis result; classifying and storing the system optimization guidelines according to a plurality of dimensions; and in a subsequent workflow execution process, dynamically calling and applying the stored system optimization guide so as to optimize the workflow.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent query system optimization, and in particular to an intelligent guide generation and application method, system and device based on error analysis. Background Art

[0002] With the rapid development and widespread application of artificial intelligence technology, workflow-based intelligent query systems often face various challenges when handling complex tasks. These intelligent query systems usually involve the collaborative work of multiple modules, such as natural language understanding, task planning, information retrieval, answer generation, etc. However, due to the complexity and diversity of tasks, these systems often encounter various errors in actual operation, such as incomplete information retrieval, insufficient task execution, logical judgment errors, etc., resulting in the output of answers that do not meet user needs, seriously affecting the system performance and user experience.

[0003] At present, the main methods to solve these problems include: 1) manually writing more rules to cover various possible error conditions; 2) collecting a large number of error samples for machine learning to improve the robustness of the system; 3) using complex error handling mechanisms to capture and handle abnormal situations. However, these methods all have obvious limitations: the method of manually writing rules is difficult to cover all possible error conditions and has high maintenance costs; the method based on machine learning requires a large amount of labeled data and is difficult to quickly adapt to new error types; and complex error handling mechanisms may increase the complexity of the system and affect the overall performance. Summary of the invention

[0004] The embodiments of the present invention provide an intelligent guide generation and application method, system and device based on error analysis, which are used to solve the following technical problems: the existing intelligent query system optimization methods are difficult to cover all possible error situations and difficult to quickly adapt to new error types, and the complex error handling mechanism increases the complexity of the system.

[0005] The embodiment of the present invention adopts the following technical solutions:

[0006] On the one hand, an embodiment of the present invention provides a method for generating and applying an intelligent guide based on error analysis, the method comprising: based on a workflow error analysis macro model, analyzing errors occurring during the workflow execution of an intelligent query system to obtain error analysis results;

[0007] Based on the error analysis results, generate a corresponding system optimization guide;

[0008] Classify and store the system optimization guide according to multiple dimensions;

[0009] During subsequent workflow execution, the stored system optimization guide is dynamically retrieved and applied to optimize the workflow.

[0010] In a feasible implementation, based on the workflow error analysis model, the errors occurring during the workflow execution of the intelligent query system are analyzed to obtain error analysis results, which specifically include:

[0011] Build and train large models for workflow error analysis;

[0012] Inputting the complete workflow information of the intelligent query system into the workflow error analysis model to analyze the workflow and obtain error analysis results;

[0013] The workflow information includes at least candidate skills and descriptions, the user's original question, and information about the called module;

[0014] The error analysis results include at least: the specific location of the error, the root cause of the error, the scope of the error impact, and potential systemic problems.

[0015] In a feasible implementation, building and training a workflow error analysis model specifically includes:

[0016] Collecting workflow data containing various error types from actual business scenarios; the workflow data comes from at least system log files, user feedback, and test cases;

[0017] Performing data cleaning, data labeling, and data formatting on the workflow data;

[0018] Based on the large language model, construct the workflow error analysis large model;

[0019] The workflow error analysis big model is trained by the workflow data to train the workflow error analysis big model's ability to understand and analyze the workflow data.

[0020] In a feasible implementation manner, based on the error analysis result, a corresponding system optimization guide is generated, which specifically includes:

[0021] Extracting the factors affecting guideline generation from the error analysis results; wherein the factors affecting guideline generation include at least: the severity of the error, the frequency of occurrence of each error, and the feasibility of correcting each error;

[0022] Input the influencing factors of the guideline generation into the large language model for analysis and processing, and automatically generate the corresponding system optimization guideline; wherein the system optimization guideline includes a macro guideline and a micro guideline;

[0023] The macro guide is used to record the optimization strategy for the system level, and the micro guide is used to record the optimization details for a specific scenario.

[0024] In a feasible implementation manner, the system optimization guide is classified and stored according to multiple dimensions, specifically including:

[0025] According to the content of each system optimization guide and the error type targeted, the system optimization guide is divided into multiple dimensions; wherein the multiple dimensions at least include: application scenario, error type and task characteristics;

[0026] Based on the multiple dimensions, the system optimization guides are classified and stored, and a retrieval index is generated for each system optimization guide; the retrieval index at least includes the type of the system optimization guide, so that the system can quickly locate the position of the system optimization guide when it is needed.

[0027] In a feasible implementation, during the subsequent workflow execution process, the stored system optimization guide is dynamically retrieved and applied to optimize the workflow, specifically including:

[0028] Acquire task information currently executed by the intelligent query system; wherein the task information includes at least: task characteristics, task context information and task history data;

[0029] Based on the task information, a most suitable micro guideline is selected, and a corresponding first search index is obtained;

[0030] Based on the first search index, the most suitable micro-guideline is quickly retrieved in the guideline storage module and inserted into the corresponding workflow of the currently executed task.

[0031] In a feasible implementation manner, the method further includes:

[0032] Based on the task information, selecting the most suitable macro guideline and obtaining a corresponding second search index;

[0033] Based on the second search index, the most suitable macro guideline is quickly retrieved in the guideline storage module and integrated into the skill orchestration module of the intelligent query system to influence the overall workflow execution strategy.

[0034] In a feasible implementation manner, after dynamically retrieving and applying the stored system optimization guide to optimize the workflow, the method further includes:

[0035] Create an adaptive optimization mechanism to dynamically adjust the guideline application strategy based on the workflow execution effect after the system optimization guideline is applied.

[0036] On the other hand, an embodiment of the present invention further provides an intelligent guide generation and application system based on error analysis, the system comprising:

[0037] The error analysis module is used to analyze the errors that occur during the workflow execution of the intelligent query system based on the workflow error analysis model to obtain error analysis results;

[0038] A guideline generation module, used for generating corresponding system optimization guidelines based on the error analysis results;

[0039] A guide storage module, used for classifying and storing the system optimization guide according to multiple dimensions;

[0040] The guideline application module is used to dynamically retrieve and apply the stored system optimization guideline during the subsequent workflow execution process to optimize the workflow.

[0041] Finally, an embodiment of the present invention further provides an intelligent guide generation and application device based on error analysis, the device comprising:

[0042] at least one processor; and,

[0043] a memory communicatively connected to the at least one processor; wherein,

[0044] The memory stores instructions that can be executed by the at least one processor, so that the at least one processor can execute the intelligent guide generation and application method based on error analysis.

[0045] Compared with the prior art, the method, system and device for generating and applying intelligent guidance based on error analysis provided by the embodiments of the present invention have the following beneficial effects:

[0046] The advantages of the present invention lie in its adaptability and continuous optimization capabilities. By automatically creating and applying guidelines, the system is able to continuously learn from its mistakes and improve its behavior. This approach is not only applicable to specific scenarios, but can also be extended to various workflow-based intelligent query systems, such as information retrieval, multi-round dialogue, complex task planning, etc. In addition, the guide system of the present invention is also highly interpretable, and each optimization decision is supported by clear guidelines, which enhances the transparency and credibility of the system behavior. At the same time, this method also supports human-computer collaboration, allowing experts to manually adjust or supplement the guidelines as needed.

[0047] In summary, the present invention provides a flexible, efficient and sustainable optimization solution for AI intelligent query systems through the automatic creation and application of intelligent guidelines, which is expected to significantly improve the performance, reliability and adaptability of various intelligent query systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0049] Figure 1 A flow chart of a method for generating and applying an intelligent guide based on error analysis provided by an embodiment of the present invention;

[0050] Figure 2 A schematic diagram of the structure of an intelligent guide generation and application system based on error analysis provided by an embodiment of the present invention;

[0051] Figure 3 A schematic structural diagram of an intelligent guide generation and application device based on error analysis provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0052] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0053] The embodiment of the present invention provides a method for generating and applying an intelligent guide based on error analysis, such as Figure 1 As shown, the method for generating and applying intelligent guidance based on error analysis specifically includes steps S101-S104:

[0054] S101. Based on the workflow error analysis model, errors occurring during the workflow execution of the intelligent query system are analyzed to obtain error analysis results.

[0055] Specifically, we first need to build and train a large workflow error analysis model. The specific implementation method is as follows:

[0056] From the intelligent query system in actual business scenarios, workflow data containing various error types are collected. These workflow data come from at least system log files, user feedback, and test cases.

[0057] Then the workflow data is cleaned, labeled and formatted.

[0058] Furthermore, based on the big language model, a workflow error analysis big model is constructed. That is, the big language model is trained specifically through the collected workflow error data, so as to improve the understanding and error analysis capabilities of the big language model for workflow data.

[0059] Furthermore, the complete workflow information collected in real time in the intelligent query system is input into the trained workflow error analysis model to analyze the workflow and obtain error analysis results. The collected workflow information at least includes candidate skills and descriptions, the user's original question, and the called module information. The error analysis results at least include: the specific location of the error, the root cause of the error, the scope of the error impact, and potential systemic problems.

[0060] As a feasible implementation method, the error analysis module uses a general large language model to build a workflow error analysis large model. Through the collected workflow data containing various error types, the existing large language model is adjusted for parameters and feature learning, such as feature learning of workflow error data based on large language models such as ChatGPT, so that the learned large language model can analyze errors that occur during workflow execution. For example, in the scenario of querying the weather in multiple places, the workflow information generated during the query is input into the workflow error analysis large model, and the model can recognize that the error occurred in the skill orchestration stage because multiple instances of the same skill were not called correctly.

[0061] S102: Generate a corresponding system optimization guide based on the error analysis result.

[0062] Specifically, the factors influencing the generation of guidelines are extracted from the error analysis results; wherein the factors influencing the generation of guidelines at least include: the severity of the error, the frequency of occurrence of each error, and the feasibility of correcting each error.

[0063] As a feasible implementation method, according to the error result data analyzed by the model, it is matched with the error type in the error type library. When the same error type is matched, the frequency of each type of error is counted, and the error severity corresponding to each type of error and the feasibility of correcting the error are obtained in the error type library. Among them, information such as error type, error severity, and feasibility of error correction are stored in the error type library in an associated form. The error type library can be constructed based on expert experience and error situations that occur in daily applications of the intelligent query system.

[0064] Furthermore, the factors influencing the generation of guidelines are input into the large language model for analysis and processing, and the corresponding system optimization guidelines are automatically generated; wherein, the system optimization guidelines include macro guidelines and micro guidelines. The macro guidelines are used to record the optimization strategies at the system level, and the micro guidelines are used to record the optimization details for specific scenarios.

[0065] As a feasible implementation method, the guideline generation module is the core of the present invention. The guideline generation module generates influencing factor data and corresponding macro guidelines and micro guidelines through different guidelines, adjusts and learns parameters of the large language model, so that it can generate influencing factors according to the input guidelines, and automatically create and output corresponding macro guidelines and micro guidelines.

[0066] Among them, the macro guide is aimed at system-level optimization strategies, which can be applied to the entire intelligent query system. For example, "In the scenario of multiple instances of the same skill, it is necessary to call this interface repeatedly and enter the parameters of each instance separately."

[0067] Micro guidelines focus on the execution details of specific scenarios and can be applied to specific workflows, such as "When querying the weather in multiple locations, you need to call the weather query interface multiple times, using different location parameters for each query."

[0069] The guideline generation process takes into account multiple factors such as the severity of the error, the frequency of occurrence, and the feasibility of correction, ensuring that the generated guidelines are both targeted and universal.

[0070] It should be noted that the error analysis module and the guide generation module are both encapsulated in the back-end execution process and do not have a corresponding operation interface. Users can only see the query question input interface of the intelligent query system. After the user enters the question, the back-end executes the corresponding workflow based on the input question, analyzes the errors in the workflow, and creates the corresponding guide. The next time the system receives a similar question, it will call the corresponding guide to optimize the workflow so that the same error as last time does not occur. The intermediate implementation process is implemented through encapsulated programs, and users do not need to operate in the interface. After the user enters the question, after a short query time, he can receive the query results and view the query results in the interface, but will not see the intermediate execution process and the corresponding guide.

[0071] S103: Classify and store the system optimization guide according to multiple dimensions.

[0072] Specifically, according to the content of each system optimization guide and the error type it targets, the system optimization guide is divided into multiple dimensions; wherein the multiple dimensions include at least: application scenarios, error types, and task characteristics.

[0073] Furthermore, based on multiple dimensions, the system optimization guides are classified and stored, and a retrieval index is generated for each system optimization guide; the retrieval index at least includes the type of the system optimization guide, so that the system can quickly locate the location of the system optimization guide when it is needed.

[0074] As a feasible implementation method, the guide storage module adopts a multi-level index structure to organize and index the system optimization guides generated by the guide generation module according to multiple dimensions such as application scenarios, error types, and task characteristics, and then store them. This storage method supports fast retrieval and real-time updates, ensuring that the system can quickly find the most relevant guides when needed.

[0075] S104. In the subsequent workflow execution process, dynamically retrieve and apply the stored system optimization guide to optimize the workflow.

[0076] Specifically, the task information currently executed by the intelligent query system is obtained; wherein the task information at least includes: task application scenarios and task features.

[0077] Furthermore, based on the two dimensions of task application scenario and task characteristics, the type of micro-guideline to be retrieved is determined, and the corresponding first search index is determined according to the type of micro-guideline. Based on the first search index, the micro-guideline is quickly retrieved in the guideline storage module and inserted into the corresponding workflow of the currently executed task, so that the content of the micro-guideline is executed while the workflow is executed.

[0078] At the same time, the two dimensions of task application scenario and task characteristics determine the type of macro guideline that needs to be retrieved, and determine the corresponding second search index based on the macro guideline type. Based on the second search index, the macro guideline is quickly retrieved in the guide storage module and integrated into the skill arrangement module of the intelligent query system to affect the overall workflow execution strategy.

[0079] Furthermore, an adaptive optimization mechanism is created to dynamically adjust the guideline application strategy according to the workflow execution effect after the system optimization guideline is applied.

[0080] As a feasible implementation method, the guide application module is responsible for dynamically inserting corresponding optimization guides during the workflow execution process. It will intelligently select the most suitable guide based on the application scenario, task characteristics, contextual information and historical data of the current task. For example, when it is detected that the current task is a weather information query, the system will automatically insert micro-guidelines and macro-guidelines related to the weather information query to affect the execution process of the task, so that the intelligent query system will pay special attention to those places that are prone to errors when executing the task, and modify the workflow or check the execution process according to the content of the guide.

[0081] The workflow of the present invention is as follows: First, the system executes the initial workflow to process the user request. When an error is detected, the error analysis module performs an in-depth analysis. Based on the error analysis results, the guide generation module creates the corresponding optimization guide, which is classified and stored by the guide storage module. When the system receives a user request with the same characteristics again, the guide application module will dynamically insert the relevant optimization guide into the appropriate position in the workflow. For example, in a scenario where weather is queried in multiple locations, the system will apply the relevant guide in the skill arrangement stage to ensure that the weather query interface is called separately for each location. The system will also continue to monitor the effect of the guide application and trigger a new round of optimization cycle as needed.

[0082] The following is explained in detail through two specific embodiments:

[0083] Embodiment 1: This embodiment takes weather query in multiple locations as an example to illustrate the specific implementation method of the present invention.

[0084] 1. Problem presentation:

[0085] The user queries the weather in places A and B through the intelligent query system, and the execution result returned by the intelligent query system only contains the summary result of the weather in place A.

[0086] 2. Error analysis:

[0087] After extracting the workflow data of the query task, we input it into the workflow error analysis model and obtained the error analysis result: the user queried the weather of places A and B through the intelligent query system. When the system executed the workflow, it only called the weather query interface once and only returned the weather information of place A. And through the analysis of the big model, we concluded that the error occurred in the skill orchestration module because the corresponding interface was not called repeatedly when processing multiple instances of the same skill.

[0088] 3. Guideline Generation:

[0089] Based on the error analysis results, the guideline generation module creates the following guidelines:

[0090] Macro Guide: When dealing with multiple instances of the same skill, you need to call the skill repeatedly and enter the parameters for each instance separately.

[0091] Micro Guide: When querying the weather in multiple locations, the weather query interface should be called multiple times, using a different location as a parameter each time.

[0092] 4. Guide storage:

[0093] The guideline storage module stores the generated guidelines in categories:

[0094] The macro guide is classified as "skill orchestration strategy" (classified based on task feature dimensions), and the tags include "multi-instance processing", "interface call", etc.

[0095] The micro-guidelines are classified as "weather query scenarios" (classified based on the application scenario dimension), and the labels include "multi-location query", "parameter processing", etc.

[0096] 5. Guidelines Application:

[0097] When the system receives a similar multi-location weather query request again, the guide application module calls the corresponding macro guide and micro guide, and integrates the macro guide into the prompt information of the skill arrangement module. Then, using the regular matching algorithm, when the "weather" keyword exists in the matching user question, the micro guide is inserted into the workflow to ensure that the weather query interface is called separately for each location.

[0098] 6. Effect monitoring and optimization:

[0099] The intelligent query system executes the optimized workflow and successfully returns the weather information of locations A and B. The effect evaluation module records this successful case for effect evaluation and iterative optimization of subsequent guidelines.

[0100] Embodiment 2: Guide generation and application in real-time news search scenario.

[0101] 1. Problem presentation:

[0102] When users query real-time news information, the system searches for information online, but the information summarized by the intelligent query system in the end is a large amount of irrelevant information.

[0103] 2. Error analysis:

[0104] After extracting the workflow data of the query task, we input it into the workflow error analysis model and obtained the error analysis result: the system searches for information through the network module, but the answers returned are all irrelevant information and do not contain the search keywords. The model identified that the problem lies in the processing stage of network search results, and the analysis shows that the search crawler returns the top-ranked ads as valid results, causing the answer model to only summarize the irrelevant ads when summarizing.

[0105] 3. Guideline Generation:

[0106] The above error analysis results are input into the large language model of the guideline generation module, and the following system optimization guideline is output:

[0107] Macro Guide: When processing search results, it is necessary to effectively distinguish between advertisements and valid information, filter the advertisement content, and retain only valid information.

[0108] Micro-Guide: When searching for news scenarios, the number of advertisements and irrelevant content will increase, and more careful identification is required.

[0109] 4. Guide storage:

[0110] The guideline storage module categorizes the generated guidelines:

[0111] The macro guidelines are classified as "search result processing strategies" (classified based on task feature dimensions), and the labels include "ad identification", "information screening", etc.

[0112] The micro-guidelines are classified as "news search scenarios" (classified based on the application scenario dimension), and the labels include "ad filtering", "content verification", etc.

[0113] 5. Guidelines Application:

[0114] In the subsequent real-time news query task, the guide application module:

[0115] Integrate the macro guidelines into the overall strategy of the search result processing module, that is, add them to the large model prompt words of the result summary module.

[0116] Using the regular matching algorithm, when keywords such as recent and latest exist in the search question, micro-guidelines are applied to add them to the macro-model prompt words of the result summary module.

[0117] 6. Effect monitoring and optimization:

[0118] The system executes the optimized workflow and successfully returns relevant news information without advertising interference. The effect evaluation module records the effect of this optimization and is used to continuously improve the generation and application strategy of the guide.

[0119] Through the above examples, the present invention demonstrates the ability to automatically identify errors, generate and apply optimization guidelines in different scenarios, effectively improving the performance and reliability of the intelligent query system. The method has strong versatility and scalability and can be widely used in various workflow-based intelligent query systems.

[0120] In addition, the embodiment of the present invention further provides an intelligent guide generation and application system 200 based on error analysis, the system comprising:

[0121] The error analysis module 210 is used to analyze errors that occur during the execution of the workflow of the intelligent query system based on the workflow error analysis model to obtain error analysis results;

[0122] A guideline generation module 220, configured to generate a corresponding system optimization guideline based on the error analysis result;

[0123] A guideline storage module 230, used for classifying and storing the system optimization guideline according to multiple dimensions;

[0124] The guideline application module 240 is used to dynamically retrieve and apply the stored system optimization guideline during the subsequent workflow execution process to optimize the workflow.

[0125] Finally, the embodiment of the present invention also provides an intelligent guide generation and application device based on error analysis, such as Figure 3 As shown, the intelligent guide generation and application device based on error analysis specifically includes:

[0126] at least one processor; and a memory in communication with the at least one processor; wherein,

[0127] The memory stores instructions executable by at least one processor to enable the at least one processor to perform:

[0128] Based on the workflow error analysis model, the errors that occur during the workflow execution of the intelligent query system are analyzed to obtain error analysis results;

[0129] Based on the error analysis results, generate a corresponding system optimization guide;

[0130] Classify and store the system optimization guide according to multiple dimensions;

[0131] During subsequent workflow execution, the stored system optimization guide is dynamically retrieved and applied to optimize the workflow.

[0132] Each embodiment of the present invention is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0133] The above describes specific embodiments of the present invention. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0134] The above description is only an embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the embodiments of the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the embodiments of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for generating and applying intelligent guidelines based on error analysis, characterized in that: The method comprises: Based on the workflow error analysis model, the errors that occur during the workflow execution of the intelligent query system are analyzed to obtain error analysis results; Based on the error analysis results, generate a corresponding system optimization guide; Classify and store the system optimization guide according to multiple dimensions; During subsequent workflow execution, the stored system optimization guide is dynamically retrieved and applied to optimize the workflow.

2. The method for generating and applying intelligent guidance based on error analysis according to claim 1, characterized in that: Based on the workflow error analysis model, the errors that occur during the workflow execution of the intelligent query system are analyzed to obtain error analysis results, including: Build and train large models for workflow error analysis; Inputting the complete workflow information of the intelligent query system into the workflow error analysis model to analyze the workflow and obtain error analysis results; The workflow information includes at least candidate skills and descriptions, the user's original question, and information about the called module; The error analysis results include at least: the specific location of the error, the root cause of the error, the scope of the error impact, and potential systemic problems.

3. The method for generating and applying intelligent guidance based on error analysis according to claim 2, characterized in that: Build and train a large model for workflow error analysis, including: Collecting workflow data containing various error types from actual business scenarios; the workflow data comes from at least system log files, user feedback, and test cases; Performing data cleaning, data labeling, and data formatting on the workflow data; Based on the large language model, construct the workflow error analysis large model; The workflow error analysis big model is trained by the workflow data to train the workflow error analysis big model's ability to understand and analyze the workflow data.

4. The method for generating and applying intelligent guidance based on error analysis according to claim 1, characterized in that: Based on the error analysis results, a corresponding system optimization guide is generated, specifically including: Extracting the factors affecting guideline generation from the error analysis results; wherein the factors affecting guideline generation include at least: the severity of the error, the frequency of occurrence of each error, and the feasibility of correcting each error; Input the influencing factors of the guideline generation into the large language model for analysis and processing, and automatically generate the corresponding system optimization guideline; wherein the system optimization guideline includes a macro guideline and a micro guideline; The macro guide is used to record the optimization strategy for the system level, and the micro guide is used to record the optimization details for a specific scenario.

5. The method for generating and applying intelligent guidance based on error analysis according to claim 1, characterized in that: The system optimization guide is classified and stored according to multiple dimensions, including: According to the content of each system optimization guide and the error type targeted, the system optimization guide is divided into multiple dimensions; wherein the multiple dimensions at least include: application scenario, error type and task characteristics; Based on the multiple dimensions, the system optimization guides are classified and stored, and a retrieval index is generated for each system optimization guide; the retrieval index at least includes the type of the system optimization guide, so that the system can quickly locate the position of the system optimization guide when it is needed.

6. The method for generating and applying intelligent guidance based on error analysis according to claim 1, characterized in that: In the subsequent workflow execution process, the stored system optimization guide is dynamically retrieved and applied to optimize the workflow, specifically including: Acquire task information currently executed by the intelligent query system; wherein the task information includes at least: task characteristics, task context information and task history data; Based on the task information, a most suitable micro guideline is selected, and a corresponding first search index is obtained; Based on the first search index, the most suitable micro-guideline is quickly retrieved in the guideline storage module and inserted into the corresponding workflow of the currently executed task.

7. The method for generating and applying intelligent guidance based on error analysis according to claim 6, characterized in that: The method further comprises: Based on the task information, selecting the most suitable macro guideline and obtaining a corresponding second search index; Based on the second search index, the most suitable macro guideline is quickly retrieved in the guideline storage module and integrated into the skill orchestration module of the intelligent query system to influence the overall workflow execution strategy.

8. The method for generating and applying intelligent guidance based on error analysis according to claim 6, characterized in that: After dynamically retrieving and applying the stored system optimization guide to optimize the workflow, the method further includes: Create an adaptive optimization mechanism to dynamically adjust the guideline application strategy based on the workflow execution effect after the system optimization guideline is applied.

9. An intelligent guide generation and application system based on error analysis, characterized in that: The system comprises: The error analysis module is used to analyze the errors that occur during the workflow execution of the intelligent query system based on the workflow error analysis model to obtain error analysis results; A guideline generation module, used for generating corresponding system optimization guidelines based on the error analysis results; A guide storage module, used for classifying and storing the system optimization guide according to multiple dimensions; The guideline application module is used to dynamically retrieve and apply the stored system optimization guideline during the subsequent workflow execution process to optimize the workflow.

10. An intelligent guide generation and application device based on error analysis, characterized in that: The device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, so that the at least one processor can execute the intelligent guide generation and application method based on error analysis according to any one of claims 1-8.

Images