Knowledge re-inspection method, equipment, storage medium and device

Abstract

The invention discloses a knowledge recheck method, equipment, a storage medium and a device. Task execution parameters are configured based on an inspection mode corresponding to a knowledge reinspection task; calling a target inspection tool to execute inspection based on slice information and task execution parameters in a preset knowledge base, and obtaining an execution record; and performing problem positioning and slice tracing on the execution records to obtain knowledge optimization suggestions. Compared with the situation that real-time synchronization of knowledge and a valid system cannot be guaranteed through manual auditing, and the auditing efficiency is low, the method calls an inspection tool to execute inspection through slicing information and task execution parameters configured in a user-defined mode to achieve modular extraction and slicing of self-adaptive structured knowledge content, the data integrity is greatly improved, and the efficiency of the self-adaptive structured knowledge is improved. The inspection time is greatly shortened, and the inspection efficiency is improved.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a knowledge retrieval method, device, storage medium and apparatus. Background Technology

[0002] With the rapid expansion of business and the increase in the user base, the amount of information in the knowledge base has experienced explosive growth. Traditional knowledge base management, maintenance, content review, and editing primarily rely on manual review and maintenance. Traditional knowledge base management solutions, entirely dependent on manual review, are inefficient, have long review cycles, and cannot guarantee real-time synchronization of knowledge with existing policies. Summary of the Invention

[0003] The main purpose of this application is to provide a knowledge review method, device, storage medium and apparatus, which aims to solve the technical problem that manual review cannot guarantee the real-time synchronization of knowledge with the effective system, resulting in low review efficiency.

[0004] To achieve the above objectives, this application provides a knowledge back-checking method, which includes the following steps: Configure task execution parameters based on the inspection mode corresponding to the knowledge review task; Based on the slice information in the preset knowledge base and the task execution parameters, the target inspection tool is invoked to perform the inspection and obtain the execution record; Problems are located and their sources are traced in the execution records to obtain knowledge optimization suggestions.

[0005] Optionally, before the step of configuring task execution parameters based on the inspection mode corresponding to the knowledge re-examination task, the method further includes: A domain knowledge dataset constructed based on historical financial data is used to train a pre-set large language model to obtain a domain expert large model. Based on the aforementioned domain expert big data model, historical documents are segmented and parsed, and segment information is determined based on the segmentation results. The slice information is stored in a preset knowledge base.

[0006] Optionally, before the step of configuring task execution parameters based on the inspection mode corresponding to the knowledge re-examination task, the method further includes: Target configuration rules are filtered from preset configuration rules based on the tool type selected by the target manager; The inspection tool is configured based on the target configuration rules to obtain the configured inspection tool; Bind the configured inspection tool to the preset knowledge base to obtain the bound inspection tool.

[0007] Optionally, the tool type includes expert large model type, custom rule type, and external interface type. The step of filtering target configuration rules from preset configuration rules based on the tool type selected by the target administrator includes: If the target manager selects the domain expert large model as the tool type, then the target configuration rule is determined from the preset configuration rules as custom configuration prompts and input / output parameters; If the tool type selected by the target manager is the defined rule type, then the target configuration rule determined from the preset configuration rules is a custom configuration rule expression; If the tool type selected by the target administrator is an external interface inspection tool, then the target configuration rule is determined from the preset configuration rules to be the rules for configuring the call parameters and response parsing of the external interface.

[0008] Optionally, the step of configuring task execution parameters based on the inspection mode corresponding to the knowledge re-examination task includes: If the inspection mode corresponding to the knowledge review task is the quick inspection mode, then configure the task execution parameters to the preset policy document range; If the inspection mode corresponding to the knowledge review task is batch inspection mode, then the task execution parameters are configured as the preset knowledge inspection range.

[0009] Optionally, the execution record includes a first execution record and a second execution record. The step of calling the target inspection tool to perform inspection based on the slice information in the preset knowledge base and the task execution parameters, and obtaining the execution record, includes: When performing a quick check, the target check tool is invoked based on the slice information in the preset knowledge base and the preset scope of policy documents to check the currently edited content and obtain the first execution record; When performing batch checks, the tasks are split based on the slice information in the preset knowledge base and the preset knowledge check scope to obtain a task set; The target inspection tool is invoked based on the task set to perform the inspection and a second execution record is obtained.

[0010] Optionally, the step of locating and tracing the problem in the execution record to obtain knowledge optimization suggestions includes: The failure records in the execution log are used to locate the problem and obtain a failure list. The slice information corresponding to the failure list is traced to obtain slice traceability information; Based on the slice tracing information, the slices are marked to obtain the marked tracing information; Based on the traceability information after the labeling, knowledge optimization suggestions are determined.

[0011] Furthermore, to achieve the above objectives, this application also provides a knowledge back-checking device, which includes: The inspection tool management module 10 is used to configure task execution parameters based on the inspection mode corresponding to the knowledge review task; The sub-task module 20 is used to call the target inspection tool to perform inspection based on the slice information in the preset knowledge base and the task execution parameters, and to obtain the execution record; The result review module 30 is used to locate problems and trace the source of the execution records to obtain knowledge optimization suggestions.

[0012] In addition, to achieve the above objectives, this application also proposes a knowledge retrieval device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the knowledge retrieval method described above.

[0013] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the knowledge back-checking method described above.

[0014] One or more technical solutions proposed in this application have at least the following technical effects: This application configures task execution parameters based on the inspection mode corresponding to the knowledge review task; it calls the target inspection tool to perform the inspection based on the slice information in the preset knowledge base and the task execution parameters, and obtains the execution record; it locates problems and traces the source of slices in the execution record to obtain knowledge optimization suggestions. Compared with manual review, which cannot guarantee the real-time synchronization of knowledge and effective systems, resulting in low review efficiency, this application uses slice information and custom-configured task execution parameters to call the inspection tool to perform the inspection, realizing the modular extraction and slicing of adaptive structured knowledge content, which greatly improves data integrity, significantly shortens inspection time, and improves inspection efficiency. Attached Figure Description

[0015] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating the first embodiment of the knowledge back-checking method of this application. Figure 2 This is a schematic diagram of document slicing and parsing provided in the first embodiment of the knowledge retrieval method of this application; Figure 3 This is a schematic diagram of the knowledge update process provided in the first embodiment of the knowledge back-checking method of this application; Figure 4 This is a schematic diagram of the execution record generation process provided in the first embodiment of the knowledge back-checking method of this application; Figure 5 This is a schematic diagram of a knowledge retrieval system provided in the second embodiment of the knowledge retrieval method of this application; Figure 6 This is a schematic diagram of the module structure of the knowledge back-checking device according to an embodiment of this application; Figure 7 This is a schematic diagram of the device structure of the hardware operating environment involved in the knowledge back-checking method in the embodiments of this application.

[0018] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of this application.

[0020] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0021] The main solution of this application embodiment is as follows: This application configures task execution parameters based on the inspection mode corresponding to the knowledge back-checking task; calls the target inspection tool to perform inspection based on the slice information in the preset knowledge base and the task execution parameters, and obtains the execution record; performs problem location and slice tracing on the execution record to obtain knowledge optimization suggestions.

[0022] In this embodiment, for ease of description, the following description uses a computing service device as the execution subject.

[0023] Because manual review cannot guarantee real-time synchronization of knowledge with existing systems, the review process is inefficient.

[0024] This application provides a solution that uses slice information and custom-configured task execution parameters to call an inspection tool to perform inspections, thereby achieving modular extraction and slicing of adaptive structured knowledge content, significantly improving data integrity, greatly shortening inspection time, and increasing inspection efficiency.

[0025] As can be seen from the above embodiments, this application configures task execution parameters based on the inspection mode corresponding to the knowledge review task; it calls the target inspection tool to perform the inspection based on the slice information in the preset knowledge base and the task execution parameters, and obtains the execution record; it locates problems and traces the source of slices in the execution record to obtain knowledge optimization suggestions. Compared with manual review, which cannot guarantee the real-time synchronization of knowledge and effective systems, resulting in low review efficiency, this application uses slice information and custom-configured task execution parameters to call the inspection tool to perform the inspection, thereby achieving modular extraction and slicing of adaptive structured knowledge content, greatly improving data integrity, significantly shortening inspection time, and improving inspection efficiency.

[0026] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device capable of implementing the above functions, including a knowledge back-checking system. The following uses a computer as an example to describe this embodiment and the subsequent embodiments. The knowledge back-checking system may include a check tool management module, a check task module, and a result back-checking module. The check tool management module includes a tool management unit, a check field management unit, and a tool binding management unit. The check task module includes a check task configurator unit, a check scope configuration unit, a check tool configuration unit, and a check task executor unit. The result back-checking includes a result management unit and a result tagging unit.

[0027] Based on this, embodiments of this application provide a knowledge retrieval method, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the knowledge back-checking method of this application.

[0028] In this embodiment, the knowledge retrieval method includes steps S10 to S30: Step S10: Configure task execution parameters based on the inspection mode corresponding to the knowledge review task.

[0029] It should be noted that the knowledge review task can be a task that is manually triggered by the user, a task that is automatically triggered based on a scheduled task, or a task that is triggered based on the update of the policy document. The knowledge review instruction is generated through the above-mentioned triggering instructions, and the knowledge review task is created.

[0030] Understandably, the knowledge review task has two check modes: quick check and batch check. The task execution parameters can be configured based on the check mode.

[0031] Furthermore, step S10 also includes: if the inspection mode corresponding to the knowledge review task is the quick inspection mode, then the task execution parameters are configured to a preset policy document range; if the inspection mode corresponding to the knowledge review task is the batch inspection mode, then the task execution parameters are configured to a preset knowledge inspection range.

[0032] It should be noted that if the inspection mode corresponding to the knowledge review task is the quick inspection mode, then the task execution parameters should be configured to the preset policy document range; if the inspection mode corresponding to the knowledge review task is the batch inspection mode, then the task execution parameters should be configured to the preset knowledge inspection range.

[0033] Understandably, the preset scope of institutional documents can be a pre-set scope of institutional documents used by users to verify whether manually edited and maintained knowledge is accurate or standardized within the specified scope of institutional documents during the knowledge writing process. The preset scope of knowledge inspection can be a scope of knowledge determined by filtering based on pre-set query conditions, such as: the last inspection time range, the knowledge creation time range, the knowledge associated with a specific institutional scope, etc.

[0034] Furthermore, before step S10, the method further includes: training a preset large language model with a domain knowledge dataset constructed based on historical financial data to obtain a domain expert large model; performing segmented parsing processing on historical documents based on the domain expert large model; determining segmented information based on the segmented parsing results; and storing the segmented information in a preset knowledge base.

[0035] It should be noted that in financial operations and maintenance scenarios, business data is highly sensitive and cannot be directly used for online inference using large models. To balance data privacy and security with the need for automated processing of domain-specific knowledge, this application proposes an offline expert large model domain adaptation scheme, aiming to improve the efficiency of knowledge extraction, information comparison and inspection, and reduce manual costs. This application selects Qwen-7B as the base model, comprehensively considering its good balance between Chinese understanding, inference ability and inference efficiency, and supporting localized offline deployment to meet the security requirement of data not leaving the domain. To address the specialized characteristics of the financial operations and maintenance field, a hybrid optimization strategy combining Supervised Fine-Tuning (SFT) and Low-Rank Adaptation (LORA) is employed to train the pre-defined Qwen-7B model. The training process includes: Data preparation: A domain knowledge dataset is constructed using historical financial data. This dataset is generated by converting anonymized domain business data into a structured instruction fine-tuning format (instruction-input-output), preserving the semantic structure, terminology, and rhetoric of the original data, and constructing a training set. SFT: The model learns the domain knowledge representation paradigm through SFT, mastering the logical reasoning and response norms in operations and maintenance scenarios. LORA optimization: During fine-tuning, the backbone parameters of the base model are frozen, and only the low-rank increment matrix is ​​trained to reduce memory usage and computational overhead. Simultaneously, efficient parameter updates are achieved using a small number of high-quality samples. Considering that expert models in actual business scenarios will handle multiple different sub-tasks (knowledge slice template classification, knowledge information comparison, and knowledge rhetoric optimization), this application introduces a lightweight hybrid expert (MoE) structure into the high-level Transformer module of the Qwen-7B model. The last five layers of the Transformer architecture in the network model are replaced with lightweight expert subnetworks (MoE units). A learnable gating mechanism dynamically routes the input to the Top-k experts, dynamically activating the corresponding experts to obtain a large domain expert model for handling the corresponding task. This improves the model's specialization and generalization ability in specific scenarios.

[0036] Understandably, to further illustrate the document parsing process in this solution, please refer to... Figure 2The document slicing and parsing diagram shown illustrates how a domain expert large model performs slicing and parsing of historical documents. Based on the parsing results, slicing information is determined and stored in a pre-set knowledge base. This pre-set knowledge base can be a pre-configured parsing template library for storing policy documents and corresponding slicing information within the domain's policy knowledge framework. This application employs an adaptive structured knowledge parsing and slicing scheme and a customized domain expert large model technology. Through a fully automated standardized process, it achieves fully automated processing of knowledge discovery, knowledge information comparison, knowledge expression refinement, knowledge information review, and updates. First, to automate the processing of professional domain knowledge, reduce labor costs, and improve the efficiency of knowledge checking and information comparison, this application utilizes domain-accumulated knowledge by fine-tuning and retraining the neural network of the large model. This allows the large model to accumulate richer knowledge and experience in the professional domain, enabling it to better handle incremental domain knowledge. Second, this application proposes an adaptive structured knowledge parsing and slicing scheme. Considering that in a fixed professional domain, document knowledge sources are not diverse but typically have a fixed format or a relatively specific structure. Therefore, this application combines a finely tuned domain expert model with the extraction and analysis of the layout structure of various institutional documents to construct a parsing template library for domain institutional knowledge, i.e., a pre-defined knowledge base. When parsing institutional documents, the domain expert model is used to specify the corresponding parsing model for document slicing. Simultaneously, to better identify the document's format and content, the YOLO model is used to analyze the paragraph structure during the slicing stage. Text within the same structural range is treated as a separate slice. Through this processing, using a combination of slice templates recommended by the expert domain model and structural division recommendations from a visual model, each individual slice can retain the integrity of its content to a great extent, ultimately achieving adaptive structured knowledge slicing of documents. Furthermore, each slice paragraph can retain its corresponding original text's hierarchical heading. Thus, in subsequent knowledge retrieval schemes, the domain expert model can better understand and update knowledge based on the slice range, improving knowledge representation and accuracy.

[0037] It should be understood that, to further illustrate the process for updating policy documents in this plan, please refer to... Figure 3 The diagram illustrates the knowledge update process. When a policy document is modified in the database, a knowledge backcheck task is automatically triggered to check whether the knowledge information associated with the original policy document has changed and whether adjustments are needed. Simultaneously, based on the adaptive structured slicing process, the original information of the new policy document corresponding to the specified slice content is traversed sequentially according to knowledge granularity to check for changes. If changes occur, the slice range corresponding to the latest policy document is obtained, and the slice information is identified by an expert domain model. The original knowledge information is then adjusted and optimized in conjunction with existing knowledge information to ensure real-time synchronous updates of knowledge and policy documents.

[0038] Step S20: Based on the slice information in the preset knowledge base and the task execution parameters, call the target inspection tool to perform the inspection and obtain the execution record.

[0039] It should be noted that the knowledge check is performed by calling the filtered target check tool based on the slice information stored in the preset knowledge base and the task execution parameters, and the knowledge check results and execution status are determined; an execution record is generated based on the knowledge check results and execution status.

[0040] Furthermore, step S20 also includes: when performing a quick check, calling a target check tool to check the currently edited content based on the slice information in the preset knowledge base and the preset policy document scope, and obtaining a first execution record; when performing a batch check, splitting the task based on the slice information in the preset knowledge base and the preset knowledge check scope, and obtaining a task set; calling a target check tool to perform a check based on the task set, and obtaining a second execution record.

[0041] It should be noted that knowledge checking tasks can be divided into batch checking and quick checking. Batch checking task configuration first determines the tools used and the scope of knowledge to be checked. The scope of the selected checking tools is the tools bound in the previous step. The knowledge scope can be filtered using system-defined query conditions, such as the last checking time range, the knowledge creation time range, or the scope of specific systems associated with the knowledge. The configured task is pushed to the task executor, which splits the task into execution records according to the selected tools and scope. Each execution record corresponds to one knowledge and one tool. During task execution, the status of the execution records changes to pending execution, executing, successful execution, and failed execution. When all execution records of the task are in the successful or failed execution state, the entire task is completed and enters the result review stage. Quick checking is used during the user's knowledge writing process. It allows for quick selection of tools, and the checking results are displayed on the knowledge editing page, improving the efficiency of user knowledge writing.

[0042] Understandably, during a quick check, the target check tool is invoked based on the slice information in the preset knowledge base and the preset scope of policy documents to check the currently edited content and obtain a first execution record; during a batch check, the task is split based on the slice information in the preset knowledge base and the preset knowledge check scope, and the target check tool is invoked based on the task set to perform the check and obtain a second execution record. In practice, the explosive growth of knowledge in the knowledge base makes manual verification of each item extremely difficult and time-consuming, hindering comprehensive coverage. Therefore, the inspection task module supports user-configured custom periodic tasks to enable periodic inspection and verification of knowledge. This can be divided into batch inspection and quick inspection. 1) Batch inspection task configuration: Configure the inspection tools and task execution parameters to be used: preset the knowledge inspection scope. The knowledge scope can be filtered using system-defined query conditions, such as the last inspection time range, the knowledge creation time range, and the scope of specific regulations associated with the knowledge. The configured task will be pushed to the task executor. For further explanation of the executor's process of generating execution records, please refer to... Figure 4 The flowchart shown illustrates the execution record generation process. The executor breaks down tasks into execution records based on selected tools and scopes. Each execution record corresponds to one piece of knowledge and one tool. During task execution, the status of the execution record changes to pending execution, executing, successful execution, and failed execution. When a task is successfully or unsuccessfully executed, the entire task is completed and enters the result review stage. 2) The execution parameter configuration for quick checks includes preset policy document scopes. This is mainly used by users to verify whether manually edited and maintained knowledge is accurate within the specified policy document scope, or whether the knowledge content is standardized, during the knowledge writing process. This improves the efficiency of user knowledge maintenance.

[0043] Step S30: Locate the problem and trace its source in the execution record to obtain knowledge optimization suggestions.

[0044] It should be noted that this application provides result management and labeling functions through the result review module. Users can quickly locate the source of a specific policy document segment through the knowledge review results, facilitating user verification. Users can use the review and labeling function to mark execution records that the tool has identified as problematic, thereby helping to improve the tool's capabilities.

[0045] Furthermore, step S30 also includes: locating the problem in the execution records that failed to execute, and obtaining a failure list; tracing the source of the slice information corresponding to the failure list, and obtaining slice source information; marking the slice source information to obtain marked source information; and determining knowledge optimization suggestions based on the marked source information.

[0046] It should be noted that this application obtains a failure list by locating problems in execution records whose execution status is failed. The failure list is constructed from knowledge items, inspection tools, problem descriptions, and knowledge content. The application traces the source information of the slice information corresponding to the knowledge content in the failure list to determine the slice source information and marks the essential problems of the slice source information to obtain the marked source information. Based on the marked source information, knowledge optimization suggestions are determined.

[0047] This embodiment configures task execution parameters based on the inspection mode corresponding to the knowledge review task; it calls the target inspection tool to perform the inspection based on the slice information in the preset knowledge base and the task execution parameters, and obtains the execution record; it locates problems and traces the source of slices in the execution record to obtain knowledge optimization suggestions. Compared with manual review, which cannot guarantee the real-time synchronization of knowledge and effective systems, resulting in low review efficiency, this embodiment uses slice information and custom-configured task execution parameters to call the inspection tool to perform the inspection, realizing the modular extraction and slicing of adaptive structured knowledge content, greatly improving data integrity, significantly shortening inspection time, and improving inspection efficiency.

[0048] Based on the above Figure 1 The first embodiment shown presents a second embodiment of the knowledge back-checking method of this application; based on the first embodiment of this application, the same or similar content as the first embodiment described above can be referred to the above description, and will not be repeated hereafter.

[0049] In this embodiment, before step S10, the method further includes: filtering target configuration rules from preset configuration rules based on the tool type selected by the target manager; configuring the inspection tool based on the target configuration rules to obtain the configured inspection tool; and binding the configured inspection tool with a preset knowledge base to obtain the bound inspection tool.

[0050] It should be noted that, to further illustrate the knowledge retrieval system in this solution, please refer to... Figure 5 The diagram illustrates a knowledge review system implemented in three modules: a tool management module, a task module, and a manual review module. The tool management module includes tool management, field management, and tool binding management. The task module includes a task configurator, scope configuration, tool configuration, and task executor. Result review includes result management and labeling. The tool module manages and configures component tools. This module primarily combines user commands to fine-tune the expert model, driving it to execute customized tasks based on specified policy domains. The component module provides preset driver component tools to complete general knowledge review tasks. The system provides preset review tools, and users can also define their own. After binding a review tool to the knowledge base, it can be selected to create a review task in the task module. Users can share their defined review tools, which can then be bound to their own knowledge bases for use by other users. The system provides three types of custom review tools: general large-model review tools, custom rule review tools, and external interface review tools. It is suitable for various business scenarios and has strong scalability.

[0051] Understandably, target managers can select the type of inspection tool and configure rules for the selected tools, then bind the configured tools to a preset knowledge base to obtain the bound inspection tool. This application executes knowledge inspection tasks through the bound inspection tool.

[0052] Furthermore, the tool types include expert large model types, custom rule types, and external interface types. The step of filtering target configuration rules from preset configuration rules based on the tool type selected by the target administrator includes: if the tool type selected by the target administrator is a domain expert large model, then the target configuration rule is determined from the preset configuration rules as a custom configuration prompt word and input / output parameters; if the tool type selected by the target administrator is a custom rule type, then the target configuration rule is determined from the preset configuration rules as a custom configuration rule expression; if the tool type selected by the target administrator is an external interface inspection tool, then the target configuration rule is determined from the preset configuration rules as the rules for configuring external interface call parameters and response parsing.

[0053] It should be noted that the inspection tools include: 1) Domain expert large model inspection tools, where users can customize execution instructions and agree on task execution processes and processing results to achieve customized inspection of knowledge base knowledge. 2) Custom rule type inspection tools, which provide a visual configuration interface, allowing users to customize rule expressions on the page to set the relationship between "inspected content" and "target value". Users can choose comparison symbols (such as greater than, less than, equal to, and contain) to define these rules, thereby achieving automatic inspection of specific content. 3) To meet the diverse customization needs of users, this module also supports directly extending external tool capabilities, performing knowledge back-inspection by directly connecting to external interfaces.

[0054] Understandably, if the target administrator selects the domain expert large model as the tool type, the target configuration rule is determined from the preset configuration rules to be the custom configuration prompt words and input / output parameters; if the target administrator selects the definition rule type as the tool type, the target configuration rule is determined from the preset configuration rules to be the custom configuration rule expression; if the target administrator selects the external interface inspection tool as the tool type, the target configuration rule is determined from the preset configuration rules to be the configuration of external interface call parameters and response parsing rules.

[0055] This embodiment configures task execution parameters based on the inspection mode corresponding to the knowledge review task; it calls the target inspection tool to perform the inspection based on the slice information in the preset knowledge base and the task execution parameters, and obtains the execution record; it locates problems and traces the source of slices in the execution record to obtain knowledge optimization suggestions. Compared with manual review, which cannot guarantee the real-time synchronization of knowledge and effective systems, resulting in low review efficiency, this embodiment uses slice information and custom-configured task execution parameters to call the inspection tool to perform the inspection, realizing the modular extraction and slicing of adaptive structured knowledge content, greatly improving data integrity, significantly shortening inspection time, and improving inspection efficiency.

[0056] This application also provides a knowledge retrieval device; please refer to [reference needed]. Figure 6 The knowledge review device includes: The inspection task configuration module 10 is used to configure task execution parameters based on the inspection mode corresponding to the knowledge back-examination task. The task execution module 20 is used to call the target inspection tool to perform inspection based on the slice information in the preset knowledge base and the task execution parameters, and to obtain the execution record; The result review module 30 is used to locate problems and trace the source of the execution records to obtain knowledge optimization suggestions.

[0057] Furthermore, the knowledge retrieval device also includes a model building module, which is used to train a preset large language model on a domain knowledge dataset built based on historical financial data to obtain a domain expert large model; to perform segmented parsing processing on historical documents based on the domain expert large model; and to determine the segmented information according to the segmented parsing results and store the segmented information in a preset knowledge base.

[0058] Furthermore, the inspection task configuration module 10 is also used to filter target configuration rules from preset configuration rules based on the tool type selected by the target manager; configure the inspection tool based on the target configuration rules to obtain the configured inspection tool; and bind the configured inspection tool to a preset knowledge base to obtain the bound inspection tool.

[0059] Furthermore, the tool types include expert large model type, custom rule type, and external interface type. The inspection task configuration module 10 is also used to determine the target configuration rule as a custom configuration prompt word and input / output parameters from the preset configuration rules if the target manager selects the domain expert large model tool type; if the target manager selects the defined rule type tool type, then the target configuration rule is determined as a custom configuration rule expression from the preset configuration rules; if the target manager selects the external interface inspection tool tool, then the target configuration rule is determined as the configuration of external interface call parameters and response parsing rules from the preset configuration rules.

[0060] Furthermore, the inspection task configuration module 10 is also used to configure the task execution parameters to a preset policy document range if the inspection mode corresponding to the knowledge review task is a quick inspection mode; and to configure the task execution parameters to a preset knowledge inspection range if the inspection mode corresponding to the knowledge review task is a batch inspection mode.

[0061] Furthermore, the inspection task execution module 20 is also used to, when performing a quick inspection, call the target inspection tool to inspect the currently edited content based on the slice information in the preset knowledge base and the preset policy document scope, and obtain a first execution record; when performing a batch inspection, split the task based on the slice information in the preset knowledge base and the preset knowledge inspection scope to obtain a task set; and call the target inspection tool to perform the inspection based on the task set, and obtain a second execution record.

[0062] Furthermore, the result review module 30 is also used to locate the failure records in the execution record and obtain a failure list; to trace the source of the slice information corresponding to the failure list and obtain slice source information; to mark the slice source information and obtain marked source information; and to determine knowledge optimization suggestions based on the marked source information.

[0063] The knowledge back-checking device provided in this application, employing the knowledge back-checking method described in the above embodiments, can solve the technical problem of low review efficiency caused by the inability to guarantee real-time synchronization of knowledge with the effective system compared to manual review. Compared with the prior art, the beneficial effects of the knowledge back-checking device provided in this application are the same as those of the knowledge back-checking method described in the above embodiments, and other technical features in the knowledge back-checking device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0064] This application provides a knowledge back-checking device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the knowledge back-checking method described in the first embodiment above.

[0065] The following is for reference. Figure 7The diagram illustrates a structural schematic suitable for implementing the knowledge back-checking device in the embodiments of this application. The knowledge back-checking device in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 7 The knowledge back-checking device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0066] like Figure 7 As shown, the knowledge retrieval device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the knowledge retrieval device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. Communication device 1009 allows the knowledge retrieval device to communicate wirelessly or wiredly with other devices to exchange data. While the figures show knowledge retrieval devices with various systems, it should be understood that implementing or having all of the systems shown is not required. More or fewer systems may be implemented alternatively.

[0067] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0068] The knowledge back-checking device provided in this application, employing the knowledge back-checking method described in the above embodiments, can solve the technical problem of low review efficiency caused by the inability to guarantee real-time synchronization of knowledge with effective systems through manual review. Compared with the prior art, the beneficial effects of the knowledge back-checking device provided in this application are the same as those of the knowledge back-checking method described in the above embodiments, and other technical features of this knowledge back-checking device are the same as those disclosed in the method of the previous embodiment, and will not be repeated here.

[0069] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0070] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0071] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, which are used to execute the knowledge back-checking method described in the above embodiments.

[0072] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0073] The aforementioned computer-readable storage medium may be included in the knowledge re-examination device; or it may exist independently and not be assembled into the knowledge re-examination device.

[0074] The aforementioned computer-readable storage medium carries one or more programs. When the aforementioned one or more programs are executed by the knowledge review device, the knowledge review device performs the following actions: configures task execution parameters based on the inspection mode corresponding to the knowledge review task; calls the target inspection tool to perform inspection based on the slice information in the preset knowledge base and the task execution parameters, and obtains the execution record; performs problem localization and slice tracing on the execution record, and obtains knowledge optimization suggestions.

[0075] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0076] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0077] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0078] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described knowledge back-checking method, thereby solving the technical problem of knowledge back-checking. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the knowledge back-checking method provided in the above embodiments, and will not be repeated here.

[0079] The above are only some embodiments of this application and do not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A knowledge recheck method, characterized by, The knowledge review method comprises: Based on the task execution parameter configured by the inspection mode corresponding to the knowledge review task; Based on the slice information in the preset knowledge base and the task execution parameter, the target inspection tool is called to perform inspection, and the execution record is obtained; The execution record is subjected to problem positioning and slice tracing to obtain knowledge optimization suggestions.

2. The knowledge back-checking method of claim 1, wherein, Before the step of configuring the task execution parameter based on the inspection mode corresponding to the knowledge review task, it further comprises: Training a preset large language model based on a domain knowledge dataset constructed based on historical financial data to obtain a domain expert large model; Based on the domain expert large model, the historical documents are subjected to slice analysis processing, and the slice information is determined according to the slice analysis result The slice information is stored in the preset knowledge base.

3. The knowledge back-checking method of claim 1, wherein, Before the step of configuring the task execution parameter based on the inspection mode corresponding to the knowledge review task, it further comprises: Selecting a tool type from the target management personnel based on the target configuration rule; Based on the target configuration rule, the inspection tool is configured to obtain the configured inspection tool; The configured inspection tool is bound to the preset knowledge base to obtain the bound inspection tool.

4. The knowledge back-checking method of claim 3, wherein, The tool type comprises an expert large model type, a custom rule type and an external interface type, and the step of selecting a tool type from the target management personnel based on the target configuration rule comprises: If the tool type selected by the target management personnel is a domain expert large model, the target configuration rule is determined as a custom configuration prompt word and an input / output parameter from the preset configuration rule; If the tool type selected by the target management personnel is a custom rule type, the target configuration rule is determined as a custom rule expression from the preset configuration rule; If the tool type selected by the target management personnel is an external interface inspection tool, the target configuration rule is determined as a call parameter and a response analysis rule for configuring an external interface from the preset configuration rule.

5. The knowledge back-checking method of claim 1, wherein, The step of configuring the task execution parameter based on the inspection mode corresponding to the knowledge review task comprises: If the inspection mode corresponding to the knowledge review task is a quick inspection mode, the task execution parameter is configured as a preset system document range; If the inspection mode corresponding to the knowledge review task is a batch inspection mode, the task execution parameter is configured as a preset knowledge inspection range.

6. The knowledge back-checking method of claim 5, wherein, The execution record comprises a first execution record and a second execution record, and the step of calling the target inspection tool to perform inspection based on the slice information in the preset knowledge base and the task execution parameter to obtain the execution record comprises: When performing quick inspection, the target inspection tool is called based on the slice information in the preset knowledge base and the preset system document range to perform inspection on the current editing content, and a first execution record is obtained; When performing batch inspection, the task is split based on the slice information in the preset knowledge base and the preset knowledge inspection range to obtain a task set; Based on the task set, the target inspection tool is called to perform inspection, and a second execution record is obtained.

7. The knowledge back-checking method of any one of claims 1-6, wherein, The step of positioning the problem and tracing the slice based on the execution record to obtain the knowledge optimization suggestion comprises: locating problems in records of execution failures in the execution records to obtain a failure list tracing slice information corresponding to the failure list to obtain slice tracing information marking based on the slice tracing information to obtain marked tracing information determining knowledge optimization suggestions based on the marked tracing information.

8. A knowledge back-checking device, characterized by The device comprises a memory, a processor, and a computer program stored on the memory and executable on the processor, and the computer program is configured to implement the steps of the knowledge review method according to any one of claims 1 to 7.

9. A storage medium, characterized by The storage medium is a computer readable storage medium, and the storage medium stores a computer program, and the computer program is executed by the processor to implement the steps of the knowledge review method according to any one of claims 1 to 7.

10. A knowledge back-checking apparatus, characterized by The device comprises: An inspection tool management module configured to configure task execution parameters based on an inspection mode corresponding to a knowledge review task; An inspection sub-task module configured to invoke a target inspection tool to perform inspection based on slice information in a preset knowledge base and the task execution parameters, and obtain execution records; A result review module configured to locate problems in the execution records and trace slices to obtain knowledge optimization suggestions.

Images