An agent-based method for automatic checking of mine file content compliance

Abstract

The application provides a kind of mine file content compliance automatic checking method based on agent, comprising: determining the current point to be checked and the log information associated with the point to be checked, wherein the log information includes historical task information and historical execution results associated with the point to be checked;Based on the first prompt template corresponding to the scheduling agent, the point to be checked and the log information are input into the scheduling agent to obtain the first task information and the target agent to be called;Based on the first task information, the target agent is called to obtain the first execution result returned by the target agent;Based on the first task information and the first execution result, update the log information;Based on the updated log information, return the operation of determining the first task information and the target agent to be called, until the checking result of the point to be checked is obtained.The method realizes the automatic checking of mine file, saves the labor cost, improves the checking efficiency and reliability.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of mine industry data processing, and in particular to a mine file content compliance automatic checking method based on an agent. BACKGROUND

[0002] Mine production safety is undoubtedly the top priority in the entire mining industry, as it is directly related to the safety of employees, the stable operation of enterprises, and the sustainable development of the environment. Therefore, for mine safety supervision agencies and mine enterprises themselves, regular compliance checks of key files such as various daily management records, production account books, and equipment operation logs are an indispensable part of ensuring safety in production.

[0003] File compliance checks usually require inspectors to compare each item of information (such as actual operations) recorded in the file with a large number of compliance files such as laws and regulations, national and industry standards, and regulations. This not only requires inspectors to have a solid professional background, but also increases the difficulty and time-consuming nature of the inspection work due to the complexity and constant updating of regulations and standards. SUMMARY

[0004] The present application aims to at least partially solve one of the technical problems in the related art.

[0005] To this end, the first object of the present application is to propose a mine file content compliance automatic checking method based on an agent.

[0006] The second object of the present application is to propose a mine file content compliance automatic checking device based on an agent.

[0007] The third object of the present application is to propose an electronic device.

[0008] The fourth object of the present application is to propose a computer-readable storage medium.

[0009] The fifth object of the present application is to propose a computer program product.

[0010] To achieve the above objects, the first aspect of the present application proposes a mine file content compliance automatic checking method based on an agent, comprising:

[0011] determining the current inspection points and log information associated with the inspection points, wherein the log information includes historical task information and historical execution results associated with the inspection points;

[0012] based on the first prompt template corresponding to the scheduling agent, inputting the inspection points and the log information into the scheduling agent to obtain first task information and a target agent to be called;

[0013] Based on the first task information, a target agent is called to obtain a first execution result returned by the target agent;

[0014] Based on the first task information and the first execution result, the log information is updated;

[0015] Based on the updated log information, an operation of executing and determining the first task information and the target agent to be called is returned until a check result of the check point to be checked is obtained.

[0016] To achieve the above purpose, a second aspect embodiment of the present application provides a mine file content compliance automatic checking device based on an agent, comprising:

[0017] A determination module is configured to determine a current check point to be checked and log information associated with the check point to be checked, wherein the log information includes historical task information and historical execution results associated with the check point to be checked;

[0018] A processing module is configured to input the check point to be checked and the log information into a scheduling agent based on a first prompt template corresponding to the scheduling agent to obtain first task information and a target agent to be called;

[0019] A scheduling module is configured to call the target agent based on the first task information to obtain a first execution result returned by the target agent;

[0020] An updating module is configured to update the log information based on the first task information and the first execution result;

[0021] The processing module is further configured to return an operation of executing and determining the first task information and the target agent to be called based on the updated log information until a check result of the check point to be checked is obtained.

[0022] To achieve the above purpose, a third aspect embodiment of the present application provides an electronic device, comprising a processor and a memory in communication connection with the processor;

[0023] The memory stores computer execution instructions;

[0024] The processor executes the computer execution instructions stored in the memory to implement the method according to the first aspect embodiment.

[0025] To achieve the above purpose, a fourth aspect embodiment of the present application provides a computer readable storage medium, wherein the computer readable storage medium stores computer execution instructions, and the computer execution instructions are executed by a processor to implement the method according to the first aspect embodiment.

[0026] To achieve the above object, the fifth aspect of the present application provides a computer program product comprising a computer program which, when executed by a processor, implements the method of the first aspect.

[0027] The agent-based mine file content compliance automatic checking method provided by the present application, in the embodiment of the present application, first, the scheduling agent determines the current first task information of the checking point and the target agent to be called according to the checking point to be checked and the associated log information, and then obtains the current first execution result by calling the target agent, and then the scheduling agent determines the next task based on the updated log information, and so on, until the checking result of the checking point to be checked is obtained. Thus, the automatic checking of the checking point is realized, the manual cost is saved, and the file checking efficiency and accuracy are improved. And the scheduling agent makes autonomous decision on the next execution plan according to the actual situation, and calls the related agent to execute the corresponding task, so that the adaptability and robustness of the system to the input file content are strong.

[0028] Additional aspects and advantages of the present application will be in part apparent and in part pointed out hereinafter. BRIEF DESCRIPTION OF DRAWINGS

[0029] The above and / or additional aspects and advantages of the present application will become apparent and be readily appreciated from the following description, including the appended drawings, wherein:

[0030] Figure 1 A flowchart of an agent-based mine file content compliance automatic checking method provided by an embodiment of the present application;

[0031] Figure 2 A scheduling flowchart of a scheduling agent provided by an embodiment of the present application;

[0032] Figure 3 A flowchart of another agent-based mine file content compliance automatic checking method provided by an embodiment of the present application;

[0033] Figure 4 A specification regulation retrieval agent retrieval flowchart provided by an embodiment of the present application;

[0034] Figure 5 A content analysis checking agent checking flowchart provided by an embodiment of the present application;

[0035] Figure 6 A flowchart of generating a checking report provided by an embodiment of the present application;

[0036] Figure 7A kind of based on agent's mine file content compliance automatic inspection method overall flow schematic diagram provided for the embodiment of the application. DETAILED DESCRIPTION

[0037] The embodiments of the present application are described in detail below, examples of which are shown in the drawings, wherein the same or similar reference signs represent the same or similar elements or elements with the same or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and are intended to explain the present application, and cannot be understood as a limitation of the present application.

[0038] A kind of based on agent's mine file content compliance automatic inspection method of the embodiment of the present application is described below with reference to the drawings.

[0039] The compliance inspection of mine industry file, on the one hand, requires the inspector to conduct comprehensive and meticulous review on the vast files, to ensure that every detail meets the specified requirements. This requires them not only to have solid professional knowledge, but also to have enough patience and carefulness to cope with the heavy reading and analysis task.

[0040] On the other hand, compliance inspection is not a simple word comparison, but needs to compare the actual operation of the mine with the massive legal regulations, national and industry standards and regulations, etc. Compliance files are compared item by item. These regulations and standards are often complex and constantly updated, which undoubtedly increases the difficulty and time-consuming of the inspection work.

[0041] More troublesome is that even if the inspector makes great efforts, it is still difficult to completely avoid the risk of wrong judgment and omission caused by human error and subjective understanding deviation. This risk is likely to lead to the omission of safety hazards.

[0042] In order to improve the inspection efficiency and accuracy, and at the same time, liberate human resources to the greatest extent, the present application provides a kind of based on agent's mine file content compliance automatic inspection method, which can realize the automation and intelligentization of inspection full link, greatly improve the efficiency and accuracy of mine file content compliance inspection.

[0043] Figure 1 A kind of based on agent's mine file content compliance automatic inspection method of the embodiment of the present application is provided, and the flow chart of the method is shown.

[0044] As Figure 1 shown, the based on agent's mine file content compliance automatic inspection method includes:

[0045] Step 101, determine the current inspection point and the log information associated with the inspection point.

[0046] The log information includes historical task information and historical execution results associated with the to-be-checked point.

[0047] In some embodiments, the to-be-checked point is used to describe the content that needs to be checked by the automatic checking system. For example, whether the operation log or the operation log of a certain device conforms to laws and regulations, national and industry standard regulations, and the like.

[0048] In some embodiments, the to-be-checked point can be input by the user. That is, the checker can directly input the to-be-checked point into the automatic checking system through an automatic checking request. The checker can input the to-be-checked point into the automatic checking system through a file upload mode or a dialog box input mode. Then the automatic checking system can determine the current to-be-checked point from the received checking request to perform automatic checking on the to-be-checked point.

[0049] In some embodiments, the to-be-checked point can also be determined by the automatic checking system by parsing and splitting the to-be-checked document. The automatic checking system can automatically match the format of the to-be-checked document, and use the corresponding technology to parse the document into a structured text format (for example, a PDF format document can be parsed through an OCR technology). Then, if there is a table in the file, a large language model (LLM) can be used to convert the table content into text content described in natural language. Then, the processed to-be-checked content can be split into a plurality of to-be-checked points according to semantic units (usually split according to paragraphs) to form a to-be-checked point list.

[0050] In some embodiments, since there are various types of documents in the mining industry that need to be checked, the checking points corresponding to different types of documents can be different. In the present disclosure, different candidate checking point lists can be configured for different types of documents. When parsing the to-be-checked document, the candidate checking point list can be determined according to the type of the to-be-checked document, and then the to-be-checked document can be parsed based on the candidate checking point list to determine the checking point list corresponding to the to-be-checked document. Thus, the accuracy and reliability of the checking point list determined by the checking system are improved, and the integrity and reliability of the file checking are ensured.

[0051] In some embodiments, the automatic checking system can include a plurality of agents, and different agents can complete different types of tasks. For example, the automatic checking system can include a scheduling agent for planning the next task for the to-be-checked point, a specification regulation retrieval agent for retrieving the current required specification regulation according to the instruction of the scheduling agent, a content analysis and checking agent for checking and analyzing the to-be-checked point according to the specification regulation to determine the checking result, and a checking result recording agent for formatting and recording the checking result.

[0052] In some embodiments, since when checking the inspection points, the relevant legal regulations and other materials need to be referred to, different inspection points may rely on different legal regulations and other documents, and the operation sequence or number of times may be different. Therefore, the automatic inspection system can create log information of each inspection point when checking the inspection point, to provide operation basis for the scheduling agent and other agents.

[0053] Step 102, based on the first prompt template corresponding to the scheduling agent, input the to-be-inspected point and the log information into the scheduling agent, to obtain the first task information and the target agent to be called.

[0054] In some embodiments, the automatic inspection system can use the to-be-inspected point and the log information to fill in the to-be-filled positions in the first prompt template to obtain the corresponding prompt information, and then input the prompt information into the scheduling agent to obtain the first task information and the target agent output by the scheduling agent.

[0055] In some embodiments, the first prompt template can include the following to-be-filled items: LLM task item (such as generating the next execution plan according to the overall goal and the action and result of the last step in the log, and the execution plan needs to be specific and for a single specific task), decision basis (the decision depends on the content of the execution log, and the log can automatically accumulate the execution result and task of each round), callable agent (including the specific functions and input parameter descriptions of the three agents of the specification regulation retrieval agent (or called retrieval agent), the content analysis inspection agent, and the inspection result recording agent), and output content and format (the output format is consistent with the elements in the execution log, which is convenient for overall data flow management).

[0056] In some embodiments, the first task information can include the next execution task determined by the scheduling agent, the decision basis of the next task, the format and content of the result of the next task, and the like.

[0057] In some embodiments, the target agent can be any one of the specification regulation retrieval agent, the content analysis inspection agent, and the inspection result recording agent.

[0058] Step 103, based on the first task information, call the target agent to obtain the first execution result returned by the target agent.

[0059] In some embodiments, after determining the target agent, the automatic checking system can first determine the second prompt template associated with the target agent, and then generate the second prompt information based on the second prompt template corresponding to the target agent and the first task information; and then input the second prompt information into the target agent to obtain the first execution result output by the target agent.

[0060] In some embodiments, the target agent can be any one of a regulation retrieval agent, a content analysis checking agent, and a checking result recording agent.

[0061] Correspondingly, the first execution result can be a retrieval result, or a checking result, or a recording result, etc.

[0062] In step 104, the log information is updated based on the first task information and the first execution result.

[0063] In some embodiments, the automatic checking system can set a log storage area for the to-be-checked point to store information of the to-be-checked point in the checking process, such as the executed task (or operation), the execution result, etc., to provide an information source for each agent.

[0064] In some embodiments, the log information can include four elements, i.e., “feedback result of the current step” (recording the execution result information fed back by the target agent in the current step), “description of the next step” (recording the next task information (action description) generated by the dispatcher agent), “next execution node and execution parameter” (recording the target agent identifier and the corresponding input parameter information needed for the next action), and “relevant regulation set recalled in the current checking process” (accumulatively recording the relevant regulations recalled in the current checking process).

[0065] In some embodiments, each content in the log information can be stored in the log information in a step-by-step manner. For example, after inputting a to-be-checked point and an execution log (initially, each element is empty) into the dispatcher agent, the next task information, the corresponding execution tool (target agent), and the execution parameter are generated by the dispatcher agent. Then, the content of the “description of the next step” in the log information can be updated according to the content generated by the dispatcher agent. Then, the target agent (for example, any one of the regulation retrieval agent, the content analysis checking agent, and the checking result recording agent) is called to execute a specific task according to the next execution tool and the execution parameter in the execution log, and the execution result is obtained, so that the content of the “feedback result of the current step” in the log information can be updated.

[0066] In some embodiments, in the case that the target agent is not the check result recording agent, it can be determined that the current first execution result is not the final check result corresponding to the to-be-checked point, that is, other tasks need to be performed for the current to-be-checked point to obtain the final check result corresponding thereto. The automatic checking system can then update the log information based on the first task information and the first execution result.

[0067] In some embodiments, in the case that the target agent is the check result recording agent, that is, the first execution result is the final check result corresponding to the to-be-checked point, the automatic checking system can store the first execution result and the to-be-checked point in association.

[0068] Step 105: Based on the updated log information, the operation of determining the first task information and the to-be-called target agent is returned until the check result of the to-be-checked point is obtained.

[0069] In some embodiments, after obtaining the updated log information, the automatic checking system can continue to input the updated log information and the to-be-checked point into the scheduling agent, so that the scheduling agent can make a scheduling decision to determine the next task and the to-be-called target agent, etc.

[0070] The scheduling process of the scheduling agent will be further described below with reference to the scheduling flowchart of the scheduling agent shown in FIG. 8, taking a multi-round scheduling example as an example. Figure 2

[0071] First round: After determining the to-be-checked point and the log information (which is empty at this time), the scheduling agent makes a scheduling decision to determine that the next step is to call the “regulation and rule retrieval agent” to perform relevant basis retrieval for the to-be-checked point, and the retrieval result is returned by the agent.

[0072] Second round: Based on the updated log information, the scheduling agent makes a decision to call the “content analysis and checking agent” in the next step, which performs automatic checking for the to-be-checked point according to the retrieval result in the first round. The agent then feeds back the result, which is “missing xx aspect of research and judgment basis”.

[0073] Third round: According to the feedback result in the second round, the scheduling agent calls the “regulation and rule retrieval agent” again to perform relevant basis retrieval for the missing xx aspect, and the retrieval result is returned by the agent.

[0074] ​Fourth round: the scheduling agent decides the next step to call the "content analysis checking agent" to automatically check the points to be checked according to the search results in the first and third rounds, and feedback the checking results. The content analysis checking agent feeds back the checking results, such as "xx aspect violates the rules, the judgment basis is YY, and no violation is found in the remaining aspects".

[0075] Fifth round: the scheduling agent decides the next step to call the "checking result recording agent" to record the checking results in a unified checking result collection variable.

[0076] Sixth round: the scheduling agent decides to end the task and performs the final execution.

[0077] In the embodiments of the present application, first, the scheduling agent determines the current first task information of the points to be checked and the target agent to be called according to the points to be checked and the associated log information, and then obtains the current first execution result by calling the target agent. Then, the scheduling agent determines the next task based on the updated log information, and so on, until the checking result of the points to be checked is obtained. Thus, the automatic checking of the checking points is realized, the manual cost is saved, and the file checking efficiency and accuracy are improved. Moreover, the scheduling agent autonomously decides the next execution plan according to the actual situation, and calls the relevant agent to execute the corresponding task, so that the system has strong adaptability and robustness to the input file content.

[0078] Figure 3 Another flowchart of the method for automatically checking the compliance of the mine file content based on the agent provided in the embodiments of the present application.

[0079] As shown in Figure 3 , the method for automatically checking the compliance of the mine file content based on the agent comprises:

[0080] Step 301: parsing the document to be checked to determine the list of checking points corresponding to the document to be checked.

[0081] Step 302: traversing the list of checking points to determine the current points to be checked.

[0082] In some embodiments, after the user determines the content to be checked (such as the points to be checked or the document to be checked), the user can first upload the content to be checked to the automatic checking system. After obtaining the document to be checked, the automatic checking system can parse the document, convert the table content into natural language, and perform other processing to complete the splitting of the checking points and obtain the list of checking points. Then, the list of checking points is traversed to complete the checking of each checking point.

[0083] Step 303, based on the first prompt template corresponding to the scheduling agent, input the to-be-checked points into the scheduling agent to obtain the first task information and the to-be-called target agent.

[0084] The specific implementation of steps 301-303 can refer to the detailed description of steps 101-102 in the above embodiments, which will not be repeated here.

[0085] Step 304, based on the first task information, call the target agent to obtain the first execution result returned by the target agent.

[0086] In some embodiments, if the target agent is a regulation retrieval agent, the process of obtaining the first execution result by the target agent can be as shown in 4. Figure 4 The regulation retrieval agent retrieval process provided by the embodiments of the present application.

[0087] As shown in Figure 4 , after receiving the first task information, the regulation retrieval agent can first parse the first task information to determine the query statement (Query) and the target knowledge base (such as a mine law and regulation database, or a mine standard specification database, etc.); then, according to the query statement, the target knowledge base is queried through dense retrieval and sparse retrieval respectively to obtain the first query result; then the first query result is rearranged (reranker) (such as according to the semantic relevance of the first query result and the query statement) to obtain the first rearranged result and the confidence of the first rearranged result (such as the semantic relevance between the query statement can be determined as the confidence); if the confidence of the first rearranged result is greater than a threshold, then the first rearranged result can be determined as the first execution result.

[0088] In some embodiments, if the confidence of the first rearranged result is less than or equal to the threshold, the retrieval agent can first determine the keywords corresponding to the query statement (such as extracting the keywords from the query statement using the LLM model); then based on the keywords, query the target knowledge base to obtain the second query result; then rearrange the second query result (such as according to the semantic relevance of the second query result and the query statement) to obtain the first execution result.

[0089] In some embodiments, in order to improve the retrieval efficiency of the retrieval agent, as shown in Figure 4As shown, the automatic inspection system can first collect various mining laws, regulations and national and industry standards and other normative documents; then parse the documents using professional techniques (such as Optical Character Recognition (OCR) and the like) and perform text block splitting according to the document structure; then use an Embedding model to vectorize the semantics of each text block, and use a sparse retrieval model (such as BM25, BM42 and the like) to construct a sparse index; and finally store the text blocks, structured information (such as hierarchical structure) and attachments (such as tables, formulas) into a vector database to form a knowledge base.

[0090] In some embodiments, if the target agent is a content analysis inspection agent, the process of obtaining the first execution result can be as shown in Figure 5 Figure 5 An inspection flow diagram of the content analysis inspection agent provided by the embodiments of the present application.

[0091] As shown in Figure 5 The main function of the content analysis inspection agent is to analyze and inspect the points to be inspected according to the reference regulation content recalled by the search, and to make a judgment. The core is an LLM analyzer, which is essentially an LLM matched after prompting engineering. It can use the powerful reasoning ability of the large language model to compare and analyze the points to be inspected and the reference regulation content, and generate a judgment conclusion.

[0092] The LLM analyzer analyzes and inspects the points to be inspected and the relevant regulations recalled by the search to determine whether some key judgment reference information is still missing. If so, a description of the missing information is generated, otherwise, an inspection result of the point to be inspected is generated.

[0093] Step 305, updating the log information based on the first task information and the first execution result.

[0094] Step 306, based on the updated log information, returning the operation of determining the first task information and the target agent to be called, until the inspection result of the point to be inspected is obtained.

[0095] In some embodiments, when the target agent determined by the scheduling agent decision is an inspection result recording agent, the inspection result recording agent can structure each inspection result (such as dividing into inspection conclusion, inspection details, inspection basis, judgment reason, etc.) and output to a unified inspection result collection variable for saving.

[0096] The specific implementation of steps 305-306 described above can refer to the detailed description of steps 104-105 in the above embodiments, which will not be repeated here. ​

[0097] Step 307, in the case that it is checked that all the check points in the check point list have been checked, according to the check results corresponding to all the check points in the check point list, a check report corresponding to the to-be-checked document is generated.

[0098] Since one to-be-checked document can correspond to multiple check points, in order to improve the readability of the check results, the automatic check system can summarize and organize all the check results to generate a check report with high readability and rich content. The execution process can be as shown in Figure 6 . Figure 6 The flowchart of generating a check report provided by the embodiments of the present application. First, a preset report template is selected according to the check document (which needs to be prepared in advance), and then numerical statistics of key information (such as the number of compliance, the number of non-compliance, the proportion of compliance, etc.) are performed using statistical analysis techniques according to the check results. Further, statistical charts can also be generated (for example, the Text2Code technology can be used to achieve this), and finally a check report is generated according to the template.

[0099] Through the above analysis, the overall flow of the agent-based mine file content compliance automatic checking method provided by the present disclosure can be as shown in Figure 7 . Figure 7 The overall flowchart of the agent-based mine file content compliance automatic checking method provided by the embodiments of the present application.

[0100] The agent-based mine file content compliance automatic checking method of the embodiments of the present application realizes automatic checking of the compliance of the content of the mine file by utilizing the coordinated cooperation of multiple agents. The artificial cost is reduced, and the checking efficiency and accuracy are improved. And using comprehensive dense + sparse + full-text retrieval and Reranker reordering filtering mechanism, the high accuracy and integrity of the retrieval recall are guaranteed, and false recall and omission are avoided, which is of great significance to compliance checking. In addition, by means of the powerful reasoning ability of the large language model, the automatic analysis, comparison and judgment of the to-be-checked points are completed, the automation and intelligentization of file content checking are realized, and at the same time, the misjudgment caused by the quality of the artificial is avoided. Finally, the output of the rich and high readability check report is convenient for the user to clearly master the key information.

[0101] In order to realize the above-mentioned embodiments, the present application further provides an agent-based mine file content compliance automatic checking device.

[0102] The agent-based mine file content compliance automatic checking device comprises:

[0103] A determination module is configured to determine a current to-be-checked point and log information associated with the to-be-checked point, wherein the log information comprises historical task information and historical execution results associated with the to-be-checked point.

[0104] a processing module, configured to input the points to be checked and the log information into the scheduling agent based on a first prompt template corresponding to the scheduling agent, to obtain first task information and a target agent to be invoked;

[0105] a scheduling module, configured to invoke the target agent based on the first task information, to obtain a first execution result returned by the target agent;

[0106] an updating module, configured to update the log information based on the first task information and the first execution result;

[0107] The processing module is further configured to return an operation of executing the first task information and the target agent to be invoked based on the updated log information, until an inspection result of the points to be checked is obtained.

[0108] Further, in a possible implementation of the embodiment of the application, the updating module is configured to:

[0109] update the log information based on the first task information and the first execution result in a case where the target agent is not the inspection result recording agent.

[0110] Further, in a possible implementation of the embodiment of the application, the system further includes a storage module configured to, in a case where the target agent is the inspection result recording agent, store the first execution result and the points to be checked in association.

[0111] Further, in a possible implementation of the embodiment of the application, the scheduling module is further configured to:

[0112] generate second prompt information based on a second prompt template corresponding to the target agent and the first task information;

[0113] input the second prompt information into the target agent, to obtain the first execution result output by the target agent.

[0114] Further, in a possible implementation of the embodiment of the application, the target agent is a retrieval agent, and the processing module is further configured to:

[0115] analyze the first task information, to determine a query statement and a target knowledge base;

[0116] query the target knowledge base through dense retrieval and sparse retrieval respectively according to the query statement, to obtain a first query result;

[0117] rearrange the first query result, to obtain a first rearranged result and a confidence degree of the first rearranged result;

[0118] In a case where the confidence of the first rearrangement result is greater than the threshold, the first rearrangement result is determined as the first execution result.

[0119] Further, in a possible implementation manner of the embodiment of the present application, the processing module is further configured to:

[0120] In a case where the confidence of the first rearrangement result is less than or equal to the threshold, the key word corresponding to the query statement is determined.

[0121] Based on the key word, the target knowledge base is queried to obtain a second query result.

[0122] The second query result is rearranged to obtain the first execution result.

[0123] Further, in a possible implementation manner of the embodiment of the present application, the determining module is further configured to: acquire the current to-be-checked point from the received checking request.

[0124] Further, in a possible implementation manner of the embodiment of the present application, the determining module is further configured to:

[0125] The to-be-checked document is parsed to determine a checking point list corresponding to the to-be-checked document.

[0126] The checking point list is traversed to determine the current to-be-checked point.

[0127] Further, in a possible implementation manner of the embodiment of the present application, the determining module is further configured to:

[0128] According to the type of the to-be-checked document, a candidate checking point list is determined.

[0129] Based on the candidate checking point list, the to-be-checked document is parsed to determine the checking point list corresponding to the to-be-checked document.

[0130] Further, in a possible implementation manner of the embodiment of the present application, the processing module is further configured to:

[0131] In a case where all checking points in the checking point list have been checked, a checking report corresponding to the to-be-checked document is generated according to checking results corresponding to all checking points in the checking point list.

[0132] It should be noted that the foregoing explanation and description of the embodiment of the method for automatically checking the compliance of the mine file content based on the intelligent agent also apply to the embodiment of the device for automatically checking the compliance of the mine file content based on the intelligent agent, which will not be described here again.

[0133] To achieve the above-mentioned embodiments, the present application further provides an electronic device, comprising: a processor, and a memory connected with the processor in communication; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the method provided by the foregoing embodiments.

[0134] To achieve the above-mentioned embodiments, the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are executed by a processor to implement the method provided by the foregoing embodiments.

[0135] To achieve the above-mentioned embodiments, the present application further provides a computer program product, comprising a computer program, and the computer program is executed by a processor to implement the method provided by the foregoing embodiments.

[0136] In the foregoing embodiment description, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present description, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any suitable manner in any one or more embodiments or examples. In addition, the person skilled in the art can combine and combine the different embodiments or examples described in the present description and the features of the different embodiments or examples without contradiction.

[0137] In addition, the terms "first", "second" are only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include at least one of the features. In the description of the present application, the meaning of "a plurality of" is at least two, for example, two, three, etc., unless otherwise specifically limited.

[0138] Any process or method descriptions in flow charts or otherwise described herein can be understood as representing code modules, segments, or portions of code that include one or more executable instructions for implementing the specified logical functions or processes, and the various embodiments of the present application contemplate that the scope of the application includes additional implementation in which the functions are performed in a different order, including substantially simultaneously, or in reverse order, or in an order that is different from the order shown or discussed, including as an additional set of functions performed substantially simultaneously with or in place of some of the functions described herein, unless otherwise specifically stated.

[0139] The logic and / or steps represented in flow diagrams or otherwise described herein, for example, can be considered as a sequence of executable instructions, and can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For purposes of this specification, a "computer-readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer-readable medium can be a product of the manufacturing and / or processing. The computer-readable medium can include, but is not limited to, the following: an electronic connection (an electronic device having one or more wires), a portable computer diskette (a magnetic device), a RAM (random access memory), a ROM (read-only memory), an EPROM (erasable programmable ROM) or Flash memory, an optical fiber device, and a portable CD ROM. Additionally, the computer-readable medium can be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, via, for example, an optical scanner, then compiled, interpreted, or otherwise processed, and stored in a computer memory in order to be executed.

[0140] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the above embodiments, various steps or methods can be implemented in software or firmware that is stored in memory and executed by a suitable instruction execution system. As such, in some embodiments, specifically configured hardware can be used to implement at least some of the functionality described herein. For example, if implemented in hardware, the hardware can include any or a combination of the following: a discrete logic circuit having logic gates for implementing logic functions upon data signals, an application specific integrated circuit having appropriate combinational logic gates, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0141] Those of skill in the art would understand that information and signals can be represented using any of a variety of technologies and techniques. For example, data, instructions, commands, information, signals, bits, symbols, and chips that can be referenced throughout the above description can be represented by voltages, currents, electromagnetic waves, magnetic fields or particles, optical fields or particles, or any combination thereof.

[0142] In addition, each of the functional units in the various embodiments of the present application can be integrated in one processing module, or each of the units can be physically present separately, or two or more units can be integrated in one module. The integrated module can be implemented in the form of hardware or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer readable storage medium.

[0143] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A method for automatically checking the compliance of mining document content based on intelligent agents, characterized in that, Includes the following steps: Determine the current key points to be inspected and the log information associated with them, wherein the log information includes historical task information and historical execution results associated with the key points to be inspected. Determining the current key points to be inspected includes: Determine a list of candidate inspection points based on the type of document to be inspected; Based on the candidate checkpoint list, the document to be checked is parsed to determine the checkpoint list corresponding to the document to be checked. The first prompt template corresponding to the scheduling agent is filled with the key points to be checked and log information to obtain prompt information. The prompt information is then input into the scheduling agent to obtain the first task information and the target agent to be called. The first task information includes the next task to be executed determined by the scheduling agent, the decision basis for the next task, and the format and content of the result of the next task. Based on the first task information, the target agent is invoked to obtain the first execution result returned by the target agent, including: The second prompt information is generated based on the second prompt template corresponding to the target intelligent agent and the first task information; The second prompt information is input into the target agent, and the first execution result output by the target agent is obtained. The target intelligent agent is any one of the following: a regulatory retrieval intelligent agent, a content analysis and inspection intelligent agent, and an inspection result recording intelligent agent; the execution result is any one of the following: retrieval result, inspection result, or recording result. Wherein, if the target agent is a normative regulation retrieval agent, the process by which the target agent obtains the first execution result includes: The first task information is parsed to determine the query statement and the target knowledge base; Based on the query statement, the target knowledge base is queried using both dense and sparse retrieval methods to obtain the first query result; The first query result is rearranged to obtain the first rearranged result and the confidence level of the first rearranged result; If the confidence level of the first rearrangement result is greater than the threshold, the first rearrangement result is determined as the first execution result; Based on the first task information and the first execution result, the log information is updated. The log information includes the feedback result of this step, the description of the next step, the next execution node and execution parameters, and a set of retrieval and recall related regulations. The feedback result of this step records the execution result information fed back by the target agent in this step. The description of the next step records the next task information generated by the scheduler agent. The next execution node and execution parameters record the target agent identifier and corresponding input parameter information used in the next action. The set of retrieval and recall related regulations accumulates and records the relevant regulations for all retrieval behaviors recalled during this inspection process. These contents are stored in the log information step by step, including: When all elements are initially empty, a check point and execution log are input into the scheduler agent. The scheduler agent generates the next task information, the corresponding execution tool and execution parameters. Based on the content generated by the scheduler agent, the content of the next step description in the log information is updated. The execution tool is the target agent. After calling the target agent to execute the specific task according to the next execution tool and execution parameters in the execution log, and obtaining the execution result, the content of the feedback result section of this step in the log information is updated; Updating the log information also includes: In the case where the target agent is not the agent recording the inspection result, it is determined that the current first execution result is not the final inspection result corresponding to the point to be inspected, and the log information is updated based on the first task information and the first execution result; When the target agent is the agent that records the inspection results, the first execution result and the key points to be inspected are stored together. Based on the updated log information, return to the operation of determining the first task information and the target intelligent agent to be called, until the content analysis and inspection intelligent agent outputs the inspection results that meet the record based on the retrieval results of the standard regulation retrieval intelligent agent, and schedules the inspection result recording intelligent agent to record the inspection results of the key points to be inspected; The method further includes generating an inspection report for the document to be inspected, provided that all inspection points in the inspection point list have been inspected.

Images