Supports intelligent problem-solving methods, devices, media, and program products.

Abstract

This invention discloses a method, device, medium, and program product for supporting intelligent problem processing. The method includes: first, reading pending records of supporting problems from an external ledger and parsing the field mappings of the pending records; for each pending record, reading preset combined fields from the pending record and concatenating the preset combined fields into a combined query text; using the combined query text to perform a search within a specified knowledge base to obtain search results, and generating a solution using the search results, which include cited fragments, evidence summaries, hit counts, related document links, and / or debugging information; finally, writing the solution back to the external ledger. This invention effectively improves the processing efficiency of supporting problems, reduces the risk of low credibility of problem answers, and achieves high traceability throughout the entire process.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a method, device, medium, and program product that supports intelligent problem processing. Background Technology

[0002] In existing technologies, support issues typically rely on external collaborative forms, work order systems, or group chat logs for registration. Then, manual staff search for answers in knowledge bases, historical documents, and personal experience, manually write back the results, and notify the relevant personnel. A common workflow generally includes: 1) After an issue is registered, support staff manually review the pending list; 2) Based on the issue text and a few supplementary notes, support staff switch to the knowledge base or search system to retrieve information; 3) Support staff organize the search results, manually assess the usability of the answers, and then write the answers back to the log; 4) If the issue cannot be directly handled, it is manually forwarded to the corresponding module manager; 5) Manual feedback usually only remains in table fields, lacking structured synchronization and subsequent consolidation.

[0003] The aforementioned existing technologies have the following drawbacks: 1) The problem entry point and knowledge retrieval link are disconnected, requiring support personnel to repeatedly copy, retrieve, organize, and rewrite, resulting in low efficiency and a high risk of missing evidence; 2) Traditional question-answering robots are mostly oriented towards single-interaction, lacking the ability to batch read ledgers, process line by line, rewrite results, and advance status, making it difficult to embed them into real support processes; 3) Ordinary retrieval enhancement generation schemes lack a reliable diversion mechanism, and when knowledge hits are insufficient or confidence is low, unstable answers are easily output directly, leading to business risks; 4) Human feedback and experience accumulation do not form a feedback loop, and the results of modification, adoption, direct adoption, and rejection cannot continuously update the knowledge base; 5) The management side lacks the ability to identify knowledge gaps based on no-hit rate, human rewriting rate, adoption rate, and high-frequency question classification.

[0004] Therefore, there is an urgent need to invent a method to support intelligent problem processing in order to solve the problems of low efficiency, poor traceability of answers, and low credibility of answers in existing technologies. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a method, device, medium, and program product that supports intelligent problem handling, which at least partially solves the problems existing in the prior art.

[0006] Other features and advantages of the invention will become apparent from the following detailed description, or may be learned in part by practice of the invention.

[0007] To achieve the above objectives, the embodiments of the present invention provide the following technical solutions:

[0008] According to a first aspect of the present invention, a method for supporting intelligent problem handling is provided, the method comprising:

[0009] Read the pending records that support the issues from the external ledger and parse the field mappings of the pending records;

[0010] For each record to be processed, a preset combination of fields is read from the record to be processed, and the preset combination of fields is concatenated into a combined query text.

[0011] The combined query text is used to search within a specified knowledge base to obtain search results, and the search results are used to generate a solution. The search results include cited fragments, evidence summaries, hit counts, related document links, and / or debugging information.

[0012] Write the solution back to the external ledger.

[0013] Further, read the pending records supporting the issue from the external ledger, and parse the field mappings of the pending records, including:

[0014] Receive agent configuration parameters, which include external ledger address, model configuration, knowledge base retrieval scope, round-robin interval, question field name, answer field name, link field name, progress field name, module field name, registrant field name, feedback field name, confidence field name, hit number segment name, module person in charge rule, fallback person in charge, summary recipient and / or case review switch.

[0015] Read the pending records supporting the issue from the external ledger based on the external ledger address;

[0016] The record to be processed is parsed by mapping the fields according to the preset field names in the agent configuration parameters to obtain the parsed record fields.

[0017] Furthermore, for each record to be processed, a preset combination of fields is read from the record to be processed, and the preset combination of fields is concatenated into a combined query text, including:

[0018] For each pending record, if the current progress field of the pending record is "to be analyzed" or "failed and to be retried", then the pending record is added to the candidate processing set.

[0019] Read the preset combination fields from the record fields, and concatenate the preset combination fields into a combined query text;

[0020] Update the current progress field of the record to be processed to "in analysis";

[0021] The combined query text is input into a preset sub-agent or a preset rule classifier for classification to obtain classification results. The classification results include FAQ type, configuration troubleshooting type, SQL troubleshooting type, environment difference type, or manual upgrade required type.

[0022] Furthermore, the combined query text is used to perform a search within a specified knowledge base to obtain search results, and the search results are used to generate a solution, including:

[0023] Using the combined query text, the classification results, the question field value, the module field value, and / or similar case context, a search is performed within the specified knowledge base to obtain the number of search hits, cited fragments, evidence summaries, related document links, and / or debugging information. The similar case context refers to historical cases with a similarity greater than a preset similarity threshold to the current case.

[0024] If the number of hits is zero, then write a no-hit message in the answer field, write zero in the confidence field, write zero in the hit number field, write "awaiting manual confirmation" in the current progress field, and generate a no-hit result.

[0025] If the number of search hits is greater than zero, the classification results, similar case context, number of search hits, and the retrieved evidence summary are input into the preset draft agent. Relevant document links are extracted from the hit knowledge documents. Based on the link field type of the external ledger, the relevant document link field value is generated using the relevant document link. The link field type includes URL type and text type. A solution draft is generated.

[0026] Furthermore, the method of using the combined query text to perform a search within a specified knowledge base to obtain search results, and using the search results to generate a solution, also includes:

[0027] The solution draft, classification results, problem fields, evidence summary, search hit count, and preset built-in review baseline are input into the preset review sub-Agent, and the review results are output. The review results include review status, confidence level, review reason and / or suggested progress.

[0028] If the review status is passed, the current progress field is updated to "analysis completed", and the solution draft is used as the intelligent solution.

[0029] If the review status is pending manual confirmation, then the current progress field will be updated to pending manual confirmation.

[0030] Obtain the results of manual processing, and generate a final manual solution based on the results of manual processing. The results of manual processing include direct adoption or adoption after modification.

[0031] Furthermore, writing the solution back to the external ledger includes:

[0032] Write the intelligent solution, related document link field value, confidence field, search hit count and / or current progress field back to the external ledger;

[0033] If writing back to the external ledger fails, the current progress field will be updated to a failed retry status.

[0034] Furthermore, the method also includes:

[0035] For support issues, read the full table records of support issues in the external ledger, construct feedback facts based on field mapping, and generate feedback history records when the fields in the feedback facts change. The feedback facts include support issues, current progress, intelligent solutions, relevant document links, manual processing results, manual final solutions, feedback notes, confidence level, search hit count, classification results and / or source ledger address.

[0036] If the manual processing result in the feedback fact is direct adoption or adoption after modification, and the final manual solution is not empty, then the problem text and intelligent solution are extracted as historical cases, and candidate cases are generated based on the feedback fact.

[0037] The candidate cases are converted into knowledge base documents, and the knowledge base documents are stored in the knowledge base. The knowledge base documents include the problem phenomenon, scope of application, solution, precautions, links to relevant documents and / or source ledger records.

[0038] According to a second aspect of the present invention, a device for supporting intelligent problem processing is provided, the device comprising: a processor and a memory;

[0039] The memory is used to store one or more program instructions;

[0040] The processor is configured to run one or more program instructions to perform the steps of a problem-supporting intelligent processing method as described in any of the preceding claims.

[0041] According to a third aspect of the present invention, a computer-readable storage medium is provided, on which a computer program is stored, wherein when executed by a processor, the computer program implements the steps of a problem-supporting intelligent processing method as described in any of the preceding claims.

[0042] According to a fourth aspect of the present invention, a computer program product is provided, the computer program product comprising a computing program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions that, when executed by a computer, cause the computer to perform the steps of a problem-supporting intelligent processing method as described in any of the preceding claims.

[0043] This invention provides a method, device, medium, and program product for intelligent problem handling. The method includes: first, reading pending records of supporting problems from an external ledger and parsing the field mappings of the pending records; for each pending record, reading preset combined fields from the pending record and concatenating the preset combined fields into a combined query text; using the combined query text to perform a search within a specified knowledge base to obtain search results, and generating a solution using the search results, which include cited fragments, evidence summaries, hit counts, related document links, and / or debugging information; finally, writing the solution back to the external ledger. This invention effectively improves the processing efficiency of supporting problems, reduces the risk of low credibility of problem answers, and achieves high traceability throughout the entire process. Attached Figure Description

[0044] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0045] Figure 1 A flowchart illustrating a method for supporting intelligent problem handling provided in an embodiment of the present invention;

[0046] Figure 2 A schematic diagram illustrating the principle of a method for supporting intelligent problem handling provided in an embodiment of the present invention;

[0047] Figure 3 A schematic diagram of the processing timing of a problem-solving intelligent processing method provided in an embodiment of the present invention;

[0048] Figure 4 This is a schematic diagram of the intelligent problem-solving architecture provided for embodiments of the present invention. Detailed Implementation

[0049] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0050] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0051] Figure 1 A flowchart of a method for supporting intelligent problem handling according to an embodiment of the present invention is shown.

[0052] like Figure 1 As shown, the intelligent problem-solving method according to an embodiment of the present invention may include steps S100, S200, S300 and S400.

[0053] In step S100, the pending records supporting the issues in the external ledger are read, and the field mappings of the pending records are parsed.

[0054] Specifically, the above steps include:

[0055] Receive agent configuration parameters, including external ledger address, model configuration, knowledge base retrieval scope, round-robin interval, question field name, answer field name, link field name, progress field name, module field name, registrant field name, feedback field name, confidence field name, hit number segment name, module owner rule, fallback owner, summary recipient and / or case review switch.

[0056] Based on the external ledger address and authentication information, the pending records of supported issues in the external ledger are read. Then, based on the preset field names in the agent configuration parameters, the field mapping of the pending records is parsed to obtain the parsed record fields. Field matching can adopt a standardized name comparison method to normalize spaces, underscores, Chinese and English symbols and capitalization to adapt to different ledger field names.

[0057] The external ledger mentioned above is Lark Multidimensional Table, but it can also be replaced with DingTalk Table, WeChat Work Form, Jira, ZenTao, ServiceNow, Salesforce, CRM support ledger, or enterprise-developed work order system, as long as it provides interface for record reading, field updating, and status synchronization.

[0058] Next, in step S200, for each record to be processed, the preset combination fields in the record to be processed are read, and the preset combination fields are concatenated into a combined query text.

[0059] Specifically, the above steps include:

[0060] For each record to be processed, if the current progress field of the record to be processed is "to be analyzed" or "failed and to be retried", then the record to be processed is added to the candidate processing set, and other records are not included in the current row-level processing; if there is no candidate processing set in the current round, then an unchanged running record is generated.

[0061] Furthermore, historical cases are retrieved from the feedback field and the reviewed case library. For records where the manual processing result is direct adoption or adoption after modification and the final solution is not empty, the problem and the final solution are extracted as historical cases. The similarity between the current problem and the historical case problems is calculated, and the top few similar cases are selected as auxiliary context.

[0062] Generate scope record IDs with table and record identifiers, and read issue, module, feedback, and supplementary fields.

[0063] Read the preset combination fields (including question, supplementary information, remarks, phenomenon, reproduction steps, background, description, investigation record, etc.) from the record fields, and concatenate the preset combination fields into a combined query text.

[0064] The current progress field of the records to be processed will be updated to "in analysis" to prevent duplicate processing.

[0065] The combined query text is input into the preset classification sub-Agent or preset rule classifier for classification. The classification result is then input into the calibrated combined query text. The classification result includes FAQ type, configuration troubleshooting type, SQL troubleshooting type, environment difference type, or type requiring manual upgrade.

[0066] The categorization results are used to select different answer prompt templates. For example, SQL troubleshooting questions will prioritize outputting executable SQL troubleshooting steps, while configuration troubleshooting questions will prioritize outputting configuration paths and verification order.

[0067] In step S300, the combined query text is used to search within the specified knowledge base to obtain search results, and a solution is generated using the search results.

[0068] Specifically, the above steps include:

[0069] By combining query text, classification results, question field values, module field values, and / or similar case context, a search is performed within a specified knowledge base to obtain the number of search hits, cited fragments, evidence summaries, related document links, and / or debugging information. The search scope can be the global knowledge base or a node in the knowledge tree and its child nodes.

[0070] If the number of hits is zero, then write a no-hit message in the answer field, write zero in the confidence field, write zero in the hit number field, write "awaiting manual confirmation" in the current progress field, generate a no-hit result, and then trigger a support notification according to the preset module manager rules or the backup manager.

[0071] If the number of search hits is greater than zero, the classification results, similar case context, number of search hits, and the retrieved evidence summary are input into the preset draft agent. Relevant document links are extracted from the hit knowledge documents. Based on the link field type of the external ledger, relevant document link field values ​​are generated using the relevant document links. The link field types include URL type and text type. A solution draft is then generated.

[0072] Input the draft solution, classification results, issue fields, evidence summary, search hit count, and preset built-in review baseline into the preset review sub-Agent, and output the review results, which include review status, confidence level, review reason, and / or suggested progress.

[0073] If the review status is "Passed," the current progress field will be updated to "Analysis Completed," and the solution draft will be designated as the intelligent solution. If the review status is "Pending Human Confirmation," the current progress field will be updated to "Pending Human Confirmation," and a support person notification will be triggered. This step is used to prevent the automatic completion of low-reliability answers.

[0074] After manual processing is completed, the results of manual processing are obtained, and a final manual solution is generated based on the results of manual processing. The aforementioned results of manual processing include direct adoption or adoption after modification.

[0075] Preferably, the enhanced retrieval generation can be implemented using keyword retrieval, vector retrieval, hybrid retrieval, graph retrieval, or a full-text search engine. The vector index can be implemented using Chroma, Milvus, Elasticsearch, OpenSearch, FAISS, or database vector extensions. The classifier agent, draft agent, and final agent can be implemented using the same large language model, or they can be implemented using different models, rule engines, traditional machine learning classifiers, or template engines.

[0076] In step S400, the solution is written back to the external ledger.

[0077] Specifically, the above steps include:

[0078] The intelligent solution, related document link field values, confidence field, search hit count and / or current progress field are written back to the external ledger through a unified write-back node;

[0079] If writing back to the external ledger fails, the current progress field will be updated to a failed and pending retry status. If detailed failure information also cannot be written back, at least try updating the progress field separately to ensure that it can be retried and investigated later.

[0080] Generate row-level results for each pending record, including record identifier, issue, processing status, answer, related links, message, search hit count, confidence level, review status, review reason, issue category, number of similar cases, and execution trajectory of each sub-node.

[0081] Meanwhile, after all the pending records in this round have completed row-level processing, the number of rows read in this round, the number of candidate processing rows, the number of AI-generated rows completed, the number of rows awaiting manual confirmation, the number of no hits, and the number of failures are counted. Based on the failure status, the status of this round is determined as successful, partially successful, failed, or unchanged; the running record and running trajectory are saved.

[0082] In addition, the intelligent problem-solving method provided in this embodiment of the invention further includes:

[0083] The feedback synchronization subgraph reads all records of the supporting issues from the external ledger for each issue, constructs feedback facts based on field mapping, and generates feedback history records when fields in the feedback facts change. Feedback facts include supporting issues, current progress, intelligent solutions, relevant document links, manual processing results, manual final solutions, feedback notes, confidence level, search hit count, classification results, and / or source ledger address.

[0084] Obtain a snapshot of the current feedback and the previous feedback, and compare fields such as progress, AI solution, manual processing results, final manual solution, and feedback remarks. Feedback history is only added when a field has actually changed; otherwise, the original creation and update times are retained to avoid duplicate history and notifications.

[0085] Furthermore, check whether all records awaiting manual confirmation have already been notified to a support person; if not, resend them. Simultaneously, check whether the record progress has changed from "first completion without manual confirmation" to "completion with manual confirmation"; if the conditions are met, send a completion notification to the registrant.

[0086] If the manual processing result in the feedback facts is direct adoption or adoption after modification, and the final manual solution is not empty, then the problem text and intelligent solution are extracted as historical cases, and candidate cases are generated based on the feedback facts; if there are already candidate cases but the content has changed, then the candidates are refreshed and reset to pending review; if the candidate cases have been published and the content has changed subsequently, then the old knowledge document is deleted and the case is awaited for re-review.

[0087] The final solutions and notes for candidate cases are edited. When a case is selected and added to the database, it is converted into a knowledge base document and saved to the knowledge base. This knowledge base document can be used for subsequent search enhancement and matching. The knowledge base document includes the problem phenomenon, scope of application, solution, precautions, links to relevant documents, and / or source records. Candidate cases can be manually reviewed before being added to the database, or they can be automatically added when conditions such as high confidence, high adoption rate, and multiple repeated hits are met.

[0088] Feedback facts and candidate cases are loaded according to a preset time window, and statistics are collected on the processing volume, AI completion volume, volume awaiting manual confirmation, number of no hits, number of failures, adoption rate, rejection rate, low confidence rate, no hit rate, manual rewriting rate, high-frequency classification, and high-frequency no hit topics.

[0089] Based on the statistical results, suggestions for knowledge gaps are generated, such as prioritizing the filling of high-frequency, unanswered questions in documents, accumulating and revising adopted samples, and reviewing the search terms and answer templates of rejected samples. A summary email is then sent. Periodic summaries can be sent via email, or generated as dashboards, daily reports, weekly reports, management dashboards, or automatically pushed to collaborative documents.

[0090] Figure 2 The diagram illustrates the principle flow of a problem-solving intelligent processing method provided by an embodiment of the present invention.

[0091] Figure 3 The diagram shows a processing timing diagram of a problem-solving intelligent processing method provided by an embodiment of the present invention.

[0092] Figure 4A schematic diagram of the intelligent problem-solving architecture provided by an embodiment of the present invention is shown.

[0093] In addition, embodiments of the present invention also provide a device for supporting intelligent problem processing, the device comprising: a processor and a memory; the memory for storing one or more program instructions; the processor for executing one or more program instructions to perform the steps of a method for supporting intelligent problem processing as described above.

[0094] In addition, embodiments of the present invention also provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of a problem-supporting intelligent processing method as described above.

[0095] In addition, embodiments of the present invention also provide a computer program product, which includes computer program instructions that, when executed by a processor, implement the steps of a problem-supporting intelligent processing method as described above.

[0096] Compared with the prior art, the embodiments of the present invention have the following technical advantages:

[0097] The embodiments of the present invention effectively improve the efficiency of support problem processing. By automatically reading the records to be analyzed, searching the knowledge base, generating answers and writing them back to the ledger, the repetitive operations of manually copying questions, searching for information and organizing answers are reduced, and the response time of routine support problems is shortened.

[0098] This invention improves the traceability of answers. By using enhanced retrieval within a defined knowledge scope to generate and write back relevant document links, answers are no longer freely generated content that cannot be traced, but instead include cited sources and knowledge links, facilitating verification by support personnel and registrants.

[0099] The embodiments of the present invention reduce the risk of low-confidence answers being mistakenly entered into the business process. By setting up review nodes and confidence level diversion mechanisms, low-hit, no-hit, or low-confidence questions will be transferred to manual confirmation, avoiding the generative model from directly completing the ledger processing when there is insufficient evidence.

[0100] This invention effectively improves the stability of human collaboration workflow, matches responsible persons according to module fields, and automatically notifies the corresponding personnel when manual confirmation is required, reducing the cost of repeated assignment, follow-up and confirmation by humans in group chats.

[0101] This invention enables the accumulation of experience and continuous updating of the knowledge base. By synchronizing human feedback and generating candidate cases, the final human solution can be reviewed and entered into the formal knowledge base. Subsequent similar problems can be automatically retrieved and reused, forming a knowledge feedback loop.

[0102] This invention improves process observability and troubleshootability. By recording node trajectories at the run, line, feedback, and summary levels, managers and developers can view the execution status, time consumption, failure reasons, and summary payload of each node, improving the efficiency of anomaly localization.

[0103] This invention also supports the identification of knowledge gaps on the management side. By statistically analyzing the miss rate, manual rewriting rate, adoption rate, rejection rate, and high-frequency classification, managers can identify which problem areas have insufficient knowledge, thereby guiding document completion, training, and product optimization.

[0104] In this embodiment of the invention, the processor can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in this embodiment of the invention. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in this embodiment of the invention can be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software modules can be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The processor reads information from the storage medium and, in conjunction with its hardware, completes the steps of the above methods. The storage medium can be memory, for example, volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DRRAM).The storage media described in the embodiments of this invention are intended to include, but are not limited to, these and any other suitable types of memory. Those skilled in the art will recognize that the functions described in the above examples can be implemented using a combination of hardware and software. When applied software, the corresponding functions can be stored in a computer-readable medium or transmitted as one or more instructions or code on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any medium that facilitates the transmission of computer programs from one place to another. Storage media can be any available medium accessible to general-purpose or special-purpose computers. Although the invention has been described in detail above with general description and specific embodiments, modifications or improvements can be made to it, which will be apparent to those skilled in the art. Therefore, such modifications or improvements made without departing from the spirit of the invention are all within the scope of protection claimed by the invention.

[0105] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any simple modifications, equivalent changes, or alterations made by those skilled in the art using the disclosed technical content shall fall within the protection scope of the present invention.

Claims

1. A method for supporting intelligent problem processing, characterized in that, The method includes: Read the pending records that support the issues from the external ledger and parse the field mappings of the pending records; For each record to be processed, a preset combination of fields is read from the record to be processed, and the preset combination of fields is concatenated into a combined query text. The combined query text is used to search within a specified knowledge base to obtain search results, and the search results are used to generate a solution. The search results include cited fragments, evidence summaries, hit counts, related document links, and / or debugging information. Write the solution back to the external ledger.

2. The intelligent problem-solving method according to claim 1, characterized in that, Read the pending records that support the issue from the external ledger, and parse the field mappings of the pending records, including: Receive agent configuration parameters, which include external ledger address, model configuration, knowledge base retrieval scope, round-robin interval, question field name, answer field name, link field name, progress field name, module field name, registrant field name, feedback field name, confidence field name, hit number segment name, module person in charge rule, fallback person in charge, summary recipient and / or case review switch. Read the pending records supporting the issue from the external ledger based on the external ledger address; The record to be processed is parsed by mapping the fields according to the preset field names in the agent configuration parameters to obtain the parsed record fields.

3. The method for supporting intelligent problem processing according to claim 2, characterized in that, For each record to be processed, a preset combination of fields is read from the record, and the preset combination of fields is concatenated into a combined query text, including: For each pending record, if the current progress field of the pending record is "to be analyzed" or "failed and to be retried", then the pending record is added to the candidate processing set. Read the preset combination fields from the record fields, and concatenate the preset combination fields into a combined query text; Update the current progress field of the record to be processed to "in analysis"; The combined query text is input into a preset sub-agent or a preset rule classifier for classification to obtain classification results. The classification results include FAQ type, configuration troubleshooting type, SQL troubleshooting type, environment difference type, or manual upgrade required type.

4. The method for supporting intelligent problem processing according to claim 3, characterized in that, The method involves using the combined query text to perform a search within a specified knowledge base, obtaining search results, and generating a solution using the search results, including: Using the combined query text, the classification results, the question field value, the module field value, and / or similar case context, a search is performed within the specified knowledge base to obtain the number of search hits, cited fragments, evidence summaries, related document links, and / or debugging information. The similar case context refers to historical cases with a similarity greater than a preset similarity threshold to the current case. If the number of hits is zero, then write a no-hit message in the answer field, write zero in the confidence field, write zero in the hit number field, write "awaiting manual confirmation" in the current progress field, and generate a no-hit result. If the number of search hits is greater than zero, the classification results, similar case context, number of search hits, and the retrieved evidence summary are input into the preset draft agent. Relevant document links are extracted from the hit knowledge documents. Based on the link field type of the external ledger, the relevant document link field value is generated using the relevant document link. The link field type includes URL type and text type. A solution draft is generated.

5. The intelligent problem-solving method according to claim 4, characterized in that, The method further includes: using the combined query text to perform a search within a specified knowledge base, obtaining search results, and generating a solution using the search results; The solution draft, classification results, problem fields, evidence summary, search hit count, and preset built-in review baseline are input into the preset review sub-Agent, and the review results are output. The review results include review status, confidence level, review reason and / or suggested progress. If the review status is passed, the current progress field is updated to "analysis completed", and the solution draft is used as the intelligent solution. If the review status is pending manual confirmation, then the current progress field will be updated to pending manual confirmation. Obtain the results of manual processing, and generate a final manual solution based on the results of manual processing. The results of manual processing include direct adoption or adoption after modification.

6. The method for supporting intelligent problem processing according to claim 5, characterized in that, Writing the solution back to the external ledger includes: Write the intelligent solution, related document link field value, confidence field, search hit count and / or current progress field back to the external ledger; If writing back to the external ledger fails, the current progress field will be updated to a failed retry status.

7. The method for supporting intelligent problem processing according to claim 5, characterized in that, The method further includes: For support issues, read the full table records of support issues in the external ledger, construct feedback facts based on field mapping, and generate feedback history records when the fields in the feedback facts change. The feedback facts include support issues, current progress, intelligent solutions, relevant document links, manual processing results, manual final solutions, feedback notes, confidence level, search hit count, classification results and / or source ledger address. If the manual processing result in the feedback fact is direct adoption or adoption after modification, and the final manual solution is not empty, then the problem text and intelligent solution are extracted as historical cases, and candidate cases are generated based on the feedback fact. The candidate cases are converted into knowledge base documents, and the knowledge base documents are stored in the knowledge base. The knowledge base documents include the problem phenomenon, scope of application, solution, precautions, links to relevant documents and / or source ledger records.

8. A device for supporting intelligent problem processing, characterized in that, The device includes: a processor and a memory; The memory is used to store one or more program instructions; The processor is configured to run one or more program instructions to perform the steps of a problem-supporting intelligent processing method as described in any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of a problem-supporting intelligent processing method as described in any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes computer program instructions that, when executed by a processor, implement the steps of a problem-supporting intelligent processing method as described in any one of claims 1 to 7.

Images