Government affair question and answer and matter handling method and system, and related medium and product

Abstract

The application discloses a government affair question and matter handling method and system, and related medium and product, relates to the field of artificial intelligence technology. The method comprises the following steps: performing intention recognition on input business request information; if it is a policy consultation intention, recalling clauses in the government affair knowledge base according to the business request information, and performing time limit filtering and weighted rearrangement in combination with metadata such as the level of the issuing organ, the effective or ineffective date and the applicable region, screening candidate clauses, inputting a large language model to generate a reply, and outputting the reply with reference information after passing the semantic consistency verification; if it is a matter handling intention, initiating multi-round interaction according to the corresponding matter template, and guiding the user to complete the materials. The technical scheme can effectively avoid the policy illusion and time limit mismatch problem, realize seamless connection from policy consultation to matter handling, improve the accuracy and timeliness of policy answering, and further improve the reliability of government service and user satisfaction.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, specifically to a government affairs Q&A and handling method, system, and related media and products. Background Technology

[0002] In current government service systems, when using large language models to process policy consultation requests, the models often generate fictitious information that does not match the actual policy content due to incomplete training data coverage or flawed reasoning logic. Furthermore, the failure to effectively identify the expiration status of policy clauses leads to frequent citation of repealed or pre-revision clauses, resulting in a risk of timeliness mismatch and severely undermining the authority and credibility of policy explanations. In addition, government consultation and service processing scenarios generally employ a separate architecture. If users need to further process related matters after completing a policy consultation, they are often forced to repeatedly switch between different functional modules or submit information repeatedly, causing service flow interruptions and a fragmented experience, failing to achieve a seamless transition from consultation to processing. Therefore, existing government service systems constrain the reliability of government services and user satisfaction. Summary of the Invention

[0003] The main purpose of this application is to propose a method, system, and related media and products for government affairs Q&A and handling of matters, which can effectively avoid policy illusion and timeliness mismatch, achieve seamless connection from policy consultation to handling of matters, improve the accuracy and timeliness of policy answers, and thus improve the reliability of government services and user satisfaction.

[0004] To achieve the above objectives, the government affairs Q&A and procedures for handling matters proposed in this application include: Obtain user-inputted business request information and perform intent recognition on the business request information; If the identified intent is policy consultation, recall clauses are retrieved from a pre-set government knowledge base based on the business request information. Based on metadata in the pre-set government knowledge base corresponding to the recall clauses, including the issuing authority level, effective or expired date, and applicable region, the recall clauses are filtered and weighted according to policy timeliness to select candidate clauses. These candidate clauses are then input into a large language model as constraints to generate a response. Clauses used to constrain the response generation are extracted from the candidate clauses as annotation clauses, and semantic consistency is verified between the response content and the corresponding annotation clauses. If the verification passes, a response with citation information extracted from the annotation clauses is output. If the verification fails, the search scope in the pre-set government knowledge base is expanded, and the search, generation, and verification steps are re-executed. If the identified intent is to handle a matter, multiple rounds of interaction are initiated according to the business request information and the matter template corresponding to the business request information to guide the user to supplement the information materials required to handle the matter, and a work order is generated based on the collected information materials for circulation.

[0005] In one embodiment, the semantic consistency check between the response content and the corresponding labeled clause includes: The response content is broken down into independent factual assertions. A natural language reasoning model is called to determine the semantic implication relationship between each independent factual assertion and its corresponding labeled clause, and the implication confidence score of each independent factual assertion is calculated. The implied confidence scores of each independent factual assertion are compared with a preset consistency threshold. When the implied confidence scores of all independent factual assertions are greater than the preset consistency threshold, the verification is deemed to have passed. When the implied confidence score of any independent factual assertion is not greater than the preset consistency threshold, the verification is deemed to have failed. If the response fails to be validated a set number of times consecutively, the current session will be switched to a human agent for processing.

[0006] In one embodiment, obtaining the user-inputted service request information includes: Obtain business request information input by the user via voice; The built-in dialect speech recognition module performs dialect acoustic feature recognition and text normalization processing on the speech input, converts the speech input into standard text, and then performs the intent recognition step.

[0007] In one embodiment, retrieving recall clauses from a preset government knowledge base includes: In the policy and regulation database segmented by document, chapter, article, and clause structure, vector semantic retrieval and keyword retrieval are performed in parallel to obtain the recalled clauses; The policy timeliness filtering of the recall terms includes: The policy expiration filter identifies the expiration date and revision status of recall clauses, removes repealed clauses, and replaces revised clauses with the current effective version with a revision note.

[0008] In one embodiment, after performing intent recognition on the business request information, the government affairs Q&A and service processing method further includes: Based on the identified intent, the business request information is routed and distributed to the corresponding target intelligent agents; wherein, the target intelligent agents include policy consultation intelligent agents, service processing intelligent agents, work order circulation intelligent agents, and public opinion analysis intelligent agents; The process involves retrieving recall clauses from a pre-defined government knowledge base based on the business request information; filtering and weighting the recall clauses based on the metadata in the pre-defined government knowledge base corresponding to the issuing authority level, effective or expired date, and applicable region, to select candidate clauses; inputting the candidate clauses as constraints into a large language model to generate response content, and performing semantic consistency verification between the response content and the corresponding labeled clauses; if the verification passes, outputting a response with citation information extracted from the labeled clauses; if the verification fails, expanding the search scope in the pre-defined government knowledge base and re-executing the retrieval, generation, and verification steps for the policy consultation agent to execute. The process involves initiating multiple rounds of interaction based on the business request information and the corresponding item template, to execute the task handling intelligent agent. The process involves assembling and generating work orders based on the collected information materials, which are then processed by the intelligent agent that handles the work order processing.

[0009] In one embodiment, initiating multi-round interactions based on the business request information and a corresponding item template includes: The intelligent agent for handling matters initiates multi-round, slot-filling interactions based on a preset list of required materials, guiding users to complete the necessary application information and supporting documents; and... The process of assembling and generating work orders based on collected information materials for circulation includes: The work order flow intelligent agent assembles the collected application information and material attachments into a structured work order that includes the type of matter, responsible department, processing time limit, and material attachments; The sensitive personal information in the structured chemical form is desensitized; The desensitized structured chemical orders are transferred to the corresponding department's business system or grid worker's terminal, and the acceptance status and query number are pushed to the user's terminal.

[0010] In one embodiment, after the desensitized structured application form is transferred to the corresponding departmental business system or grid worker terminal, the government affairs Q&A and application processing method further includes: The public opinion analysis intelligent agent performs hotspot clustering analysis on the consultation data and work order data after the cases are closed, and generates a public opinion analysis report. Based on the public opinion analysis report, frequently missed issues and hot public needs are extracted, and the content of the preset government knowledge base is updated and the search weight parameters are optimized.

[0011] This application also provides a government affairs Q&A and service processing system for implementing the above-mentioned government affairs Q&A and service processing methods, wherein the government affairs Q&A and service processing system includes: The request acquisition and intent recognition module is used to acquire business request information input by the user and to perform intent recognition on the business request information; The storage module stores a preset government knowledge base, which includes metadata about the issuing authority level, effective or expiration date, and applicable region corresponding to the recall clauses. The policy consultation processing module is used to retrieve recall clauses from a preset government knowledge base based on the business request information when the identified intent is policy consultation; based on the metadata in the preset government knowledge base corresponding to the issuing authority level, effective or expired date, and applicable region of the recall clauses, the module performs policy timeliness filtering and weighted ranking on the recall clauses to select candidate clauses; input the candidate clauses as constraints into a large language model to generate response content, and perform semantic consistency verification between the response content and the corresponding labeled clauses; if the verification passes, output a response with citation information extracted from the labeled clauses; if the verification fails, expand the search scope in the preset government knowledge base and re-execute the search, generation, and verification steps. The task processing module is used to initiate multiple rounds of interaction based on the business request information and the corresponding task template when the identified intent is to process a task. This is done to guide the user to complete the information materials required for processing the task and to assemble and generate a work order based on the collected information materials for processing.

[0012] This application also provides a storage medium, which is a computer-readable storage medium, and stores a computer program thereon. When the computer program is executed by a processor, it implements the above-described government affairs Q&A and matter handling methods.

[0013] This application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the government affairs Q&A and matter handling methods described above.

[0014] As can be seen from the above, the government affairs Q&A and service handling method provided in this application includes: obtaining business request information input by the user and identifying the intent of the business request information; if the identified intent is policy consultation intent, retrieving recall clauses from a preset government affairs knowledge base based on the business request information; filtering and weighting the recall clauses based on the metadata of the issuing authority level, effective or expired date, and applicable region corresponding to the recall clauses in the preset government affairs knowledge base to select candidate clauses; inputting the candidate clauses as constraints into a large language model to generate response content, and outputting a response with reference information extracted from the marked clauses after the semantic consistency verification between the response content and the corresponding marked clauses passes; if the verification fails, expanding the search scope in the preset government affairs knowledge base and re-executing the search, generation, and verification steps; if the identified intent is service handling intent, initiating multiple rounds of interaction according to the service request information and the corresponding service template to guide the user to supplement the information materials required for service handling, and assembling and generating a work order based on the collected information materials for circulation. This application effectively intercepts expired and inapplicable policies by filtering and weighting them based on metadata such as the issuing authority's level, timeliness, and applicable region during the recall phase, thus solving the timeliness mismatch problem in traditional retrieval. Secondly, it uses the selected candidate clauses as hard constraints for the generation of the large model, and combines semantic consistency verification and a closed-loop error correction mechanism to expand the search, physically limiting the blind association of the large model, effectively curbing "policy illusion," and providing traceable citation sources. Finally, by dynamically diverting consultation and processing flows through unified intent recognition, and automatically collecting materials to assemble work orders through multi-round template interaction, it eliminates the service gap between consultation and processing, achieving a seamless one-stop connection of "consultation-guidance-processing," significantly improving the accuracy, timeliness, and user experience of government affairs Q&A.

[0015] In summary, this application effectively addresses the issues of policy illusion, timeliness mismatch, and service process interruption, achieving a seamless connection from policy consultation to handling of matters, improving the accuracy and timeliness of policy answers, and thus enhancing the reliability of government services and user satisfaction. Attached Figure Description

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

[0017] Figure 1 A flowchart of the first embodiment of the government affairs Q&A and handling method provided in this application; Figure 2 A flowchart of the second embodiment of the government affairs Q&A and handling method provided in this application; Figure 3 A flowchart of the third embodiment of the government affairs Q&A and handling method provided in this application; Figure 4 A flowchart of the fourth embodiment of the government affairs Q&A and handling method provided in this application; Figure 5 The flowchart of the fifth embodiment of the government affairs Q&A and handling method provided in this application.

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

[0019] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0020] Furthermore, if the embodiments of this application involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the use of "and / or" or "and / or" throughout the text includes three parallel solutions. For example, "A and / or B" includes solution A, solution B, or a solution that simultaneously satisfies A and B. Furthermore, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed in this application.

[0021] In current government service systems, when using large language models to process policy consultation requests, the models often generate fictitious information that does not match the actual policy content due to incomplete training data coverage or flawed reasoning logic. Furthermore, the failure to effectively identify the expiration status of policy clauses leads to frequent citation of repealed or pre-revision clauses, resulting in a risk of timeliness mismatch and severely undermining the authority and credibility of policy explanations. In addition, government consultation and service processing scenarios generally employ a separate architecture. If users need to further process related matters after completing a policy consultation, they are often forced to repeatedly switch between different functional modules or submit information repeatedly, causing service flow interruptions and a fragmented experience, failing to achieve a seamless transition from consultation to processing. Therefore, existing government service systems constrain the reliability of government services and user satisfaction.

[0022] In response, this application provides a method for handling government affairs inquiries and matters, which can effectively avoid policy illusions and timeliness mismatches, achieve seamless connection from policy consultation to matter handling, improve the accuracy and timeliness of policy answers, and thus improve the reliability of government services and user satisfaction.

[0023] In one embodiment of this application, as Figure 1 As shown, the government affairs Q&A and procedures for handling matters include: Step S110: Obtain the service request information input by the user, and perform intent recognition on the service request information; Step S120: If the identified intent is policy consultation, retrieve the recall clauses from the preset government knowledge base based on the business request information; based on the metadata of the issuing authority level, effective or expired date, and applicable region corresponding to the recall clauses in the preset government knowledge base, filter and weight the recall clauses according to policy timeliness to select candidate clauses; input the candidate clauses as constraints into the large language model to generate response content, and extract the clauses used to constrain the response generation from the candidate clauses as annotation clauses, and perform semantic consistency verification between the response content and the corresponding annotation clauses; if the verification passes, output the response with the citation information extracted from the annotation clauses; if the verification fails, expand the search scope in the preset government knowledge base and re-execute the search, generation, and verification steps. Step S130: If the identified intent is to handle a matter, then according to the business request information, initiate multiple rounds of interaction according to the matter template corresponding to the business request information to guide the user to complete the information materials required to handle the matter, and assemble and generate a work order based on the collected information materials for circulation.

[0024] Business request information refers to the original information submitted by users to the government service system through various channels (such as text input, voice input, etc.), containing their consultation or processing needs. This information is the starting point for the system to carry out subsequent processing.

[0025] Intent recognition refers to the system's analysis of user-input business request information to determine the fundamental purpose or category of the user's request. For example, it can identify whether a user wants policy consultation or to handle a specific transaction.

[0026] Recall clauses refer to a set of policies or service clauses related to the user's query content, initially retrieved from a pre-set government knowledge base based on the user's business request information. These clauses may contain some relevance, but have not yet undergone fine screening.

[0027] The labeled clauses refer to policy or operational clauses extracted or determined from the candidate clauses that are specifically used to constrain the generation of responses from large language models and serve as the basis for generation.

[0028] Policy timeliness filtering refers to screening recall clauses based on their metadata (such as effective or expiration dates, revision status, etc.), removing expired or repealed clauses, and replacing them with currently valid versions to ensure the timeliness of policy responses.

[0029] Weighted ranking refers to reordering the selected clauses based on the policy timeliness filtering, according to the metadata of the recalled clauses (such as the issuing authority level, applicable region, etc.) and their relevance to the user request, assigning different weights to different clauses, so as to prioritize the display of the most relevant and authoritative candidate clauses.

[0030] Semantic consistency verification refers to the semantic comparison and verification between the response content generated by the large language model and the labeled clauses used as the basis for generation, in order to ensure that the response content accurately reflects the meaning of the labeled clauses and avoid the generation of "policy illusion".

[0031] The government affairs Q&A and service processing method in this embodiment first obtains the service request information input by the user and then performs intent recognition on the service request information. For example, the user can input their inquiry content through a text box. After receiving the text information, the system determines whether the user wants to learn about policies or process services through preset rule matching or a shallow machine learning model. As an optional implementation, the system can receive the request type selected by the user through a preset drop-down menu or button and perform intent recognition accordingly.

[0032] If the identified intent is policy consultation, the system retrieves relevant clauses from a pre-defined government knowledge base based on the business request information. Specifically, this can be done by using keyword matching to search the knowledge base for clauses containing the user's query keywords. Alternatively, it can be done by using vector similarity calculation to compare the semantic vector of the user's request with the semantic vectors of clauses in the knowledge base to obtain semantically relevant clauses.

[0033] Subsequently, based on metadata from the pre-defined government knowledge base regarding the issuing authority level, effective or expired date, and applicable region corresponding to the recalled clause, the system performs policy timeliness filtering and weighted ranking on the recalled clause to select candidate clauses. For example, the system can check the effective date of the recalled clauses and eliminate all clauses that have not yet taken effect or have been clearly expired. Simultaneously, it can perform preliminary sorting based on the issuing authority level (e.g., national, provincial, municipal), or adjust the weights based on the matching degree between the user's location and the applicable region of the clause.

[0034] The selected candidate clauses are input as constraints into a large language model to generate the response. This model is configured to refer to the provided candidate clauses during response generation to avoid generating content that is inconsistent with the facts or lacks basis. For example, the model can be instructed to extract information only from the candidate clauses for reorganization and polishing, or to continuously compare the relevance between the generated content and the candidate clauses during the generation process. During this process, the system extracts specific clause paragraphs from the candidate clauses used to constrain the response generation as labeled clauses, providing clear technical support for subsequent verification and citation.

[0035] After generating the response, the system performs a semantic consistency check between the response and the corresponding labeled clause. If the check passes, the system outputs a response containing the citation information extracted from the labeled clause. For example, the system can use a rule-based text comparison method to check whether there are any statements in the response that directly contradict the labeled clause. If the check fails, the system expands the search scope in the preset government knowledge base and re-executes the search, generation, and verification steps, i.e., re-executes step S120. For example, the system can relax the search criteria to obtain more recalled clauses or try different search strategies to find more accurate evidence.

[0036] If the identified intent is to process a request, the system initiates multiple rounds of interaction based on the corresponding service template to guide the user in providing the necessary information and materials. For example, the system can ask the user questions one by one according to a preset workflow, requiring basic information such as name, ID number, and contact information, and prompting the user to upload relevant supporting documents. This interaction process can be a fixed question-and-answer sequence.

[0037] Finally, work orders are generated based on the collected information and materials and then processed. For example, the system can simply package the information provided by the user and the uploaded materials into a work order to be processed, and send it to a general to-do queue to await manual processing or further system distribution.

[0038] The government affairs Q&A and service processing method in this embodiment effectively solves the problem of the disconnect between government affairs consultation and service processing scenarios through a unified intent recognition mechanism, thus achieving service continuity. Simultaneously, by introducing policy timeliness filtering, weighted ranking, labeled clause extraction, and semantic consistency verification and closed-loop error correction mechanisms, it reduces the risk of "policy illusion" and timeliness mismatch in policy responses generated by the large language model, and improves the accuracy and traceability of the generated content.

[0039] It is understandable that after completing the intent recognition, step S110 can directly execute step S120 or step S130 in parallel or sequentially according to the recognition result; and if the verification fails in step S120, the search scope can be automatically expanded and the "search-generation-verification" steps can be re-executed in a loop, thereby building a closed-loop processing mechanism that integrates policy consultation and matter handling.

[0040] The government affairs Q&A and service handling method in this embodiment effectively solves the problem of the separation between government affairs consultation and service handling scenarios through a unified intent recognition mechanism, realizing a coherent and seamless connection from consultation to handling. At the same time, by introducing policy timeliness filtering, weighted ranking, and semantic consistency verification and closed-loop error correction mechanisms, the risk of "policy illusion" and timeliness mismatch in policy responses generated by the large language model is reduced, improving the accuracy and traceability of the generated content, thereby improving the reliability of government services and user satisfaction.

[0041] In one embodiment, such as Figure 2 As shown, the semantic consistency check between the response content and the corresponding labeled clauses includes: Step S210: Decompose the response content into independent factual assertions, call the natural language inference model to determine the semantic implication relationship between each independent factual assertion and its corresponding labeled clause, and calculate the implication confidence score of each independent factual assertion. Step S220: Compare the implied confidence scores of each independent factual assertion with a preset consistency threshold; when the implied confidence scores of all independent factual assertions are greater than the preset consistency threshold, the verification is deemed to have passed; when the implied confidence score of any independent factual assertion is not greater than the preset consistency threshold, the verification is deemed to have failed. Step S230: If the response content fails to be verified consecutively a set number of times, the current session will be switched to a human agent for processing.

[0042] After generating the response, the system first breaks it down into multiple independent factual assertions. The response is typically coherent text containing multiple information points; to achieve accurate verification, it needs to be decomposed into the smallest, independently verifiable units of fact. For example, syntactic analysis, information extraction, or rule-based methods can be used to identify the subject-verb-object structure in the response, or punctuation and conjunctions can be used for segmentation, thus breaking down complex statements into a series of phrases or sentences expressing a single fact. Subsequently, the system invokes a natural language inference model to determine the semantic implication relationship between each independent factual assertion and its corresponding labeled clause. A natural language inference model is an artificial intelligence model capable of evaluating the logical relationship (such as implication, contradiction, or neutrality) between two text fragments; for example, a pre-trained model based on a Transformer architecture (such as BERT, RoBERTa, or DeBERTa) can be used. This natural language inference model takes the labeled clause as premises and the independent factual assertions as hypotheses, outputting a probability value representing the likelihood that the hypothesis is logically derived from the premises, i.e., the implication confidence score. This score quantifies the degree to which each fact in the response is consistent with the original source of knowledge (the marked clause).

[0043] Next, the system compares the calculated implied confidence score with a preset consistency threshold. This preset consistency threshold is a configurable parameter, typically ranging from 0 to 1, and can be optimized through experiments on a validation set based on the accuracy requirements of actual business operations. For example, it can be set to 0.85. The system performs assertion-by-assertion judgment: when the implied confidence scores of all independent factual assertions are greater than the preset consistency threshold, the entire response content is deemed to have passed verification, indicating that the response content is highly consistent with and accurate with the labeled clauses. At this time, a response with citation information extracted from the labeled clauses is output. Conversely, when the implied confidence score of any independent factual assertion is not greater than the preset consistency threshold, the verification is deemed to have failed. In this case, the system will trigger a reprocessing mechanism (e.g., expanding the search scope in the preset government knowledge base and re-executing the search, generation, and verification steps).

[0044] Furthermore, to address potential persistent "illusions" in large language models or situations where the knowledge base cannot meet user needs, this application also incorporates a fault-tolerance mechanism. If the response fails to validate a set number of times consecutively, the system will switch the current session to a human agent. This set number is a configurable integer parameter, such as 3 or 5 times. When the system continues to fail after multiple attempts to generate and validate the response, it indicates that the current problem may exceed the capabilities of automated processing, or that there are unresolved semantic ambiguities or missing knowledge base information. In this case, the system will send a message to the user informing them that the problem will be transferred to a human agent, and will package and send the context information of the current session (including the user's request, the system's attempted responses, validation failure records, etc.) to the human agent system so that the agent can quickly understand the situation and take over the processing.

[0045] The above technical solution enables more refined and objective verification of the accuracy of responses. This verification mechanism, based on implied confidence scores, effectively avoids inaccurate responses caused by the illusion of large language models, thus enhancing the reliability and authority of government consultation services.

[0046] In one embodiment, such as Figure 3 As shown, obtaining the user-inputted business request information includes: Step S310: Obtain the service request information input by the user via voice; Step S320: The built-in dialect speech recognition module performs dialect acoustic feature recognition and text normalization processing on the speech input, converts the speech input into standard text, and then performs the intent recognition step.

[0047] For example, users can submit their business requests verbally to the government affairs Q&A and service processing system via microphone interfaces integrated on various terminal devices (e.g., smartphones, tablets, smart speakers, or government service kiosks). Upon receiving the user's voice commands, the system converts them into digital audio signals for processing.

[0048] On this basis, to effectively address the dialect accent problem that may exist in speech input, the present application further proposes a step of performing dialect acoustic feature recognition and text normalization processing on the speech input through a built-in dialect speech recognition module, and converting the speech input into standard text before executing the intent recognition. Specifically, the dialect speech recognition module may be a combination of an acoustic model and a language model constructed based on deep learning technologies (for example, based on a Transformer architecture or a recurrent neural network). The module is trained with a large amount of dialect speech data covering different regions and different accents, so that it can accurately identify the dialect-specific acoustic features contained in the speech signal. For example, for initials, finals or tones that are significantly different from the pronunciations of Mandarin in some dialects, the module can perform targeted recognition and processing. After completing the dialect acoustic feature recognition and initially generating dialect text, the system will further perform text normalization processing. This processing process aims to convert the recognized dialect text into standard Mandarin text to eliminate the influence of dialect differences on subsequent processing. The text normalization processing may include operations such as vocabulary mapping (converting dialect vocabularies into corresponding Mandarin vocabularies), grammatical structure adjustment, and disambiguation of homographs in combination with the context. For example, "搞么事" is normalized to "do what", or the dialect pronunciation of "哪里" is converted into the standard text "where". After the dialect acoustic feature recognition and text normalization processing, the user's speech input is accurately and reliably converted into standard text. Subsequently, the standard text is input into the intent recognition module for subsequent intent recognition, so as to ensure that no matter what dialect the user uses for speech input, the service request can be accurately understood and processed by the system.

[0049] Through the above technical solutions, the problem of recognition accuracy of traditional government service systems when processing speech input, especially when facing complex dialect accents, is effectively solved. Users can conveniently submit service requests through voice, which not only improves the intelligent level of government services, but also enables the services to better cover user groups with different language habits, realizing the inclusiveness of the services.

[0050] In one embodiment, the step of retrieving recalled clauses in a preset government service knowledge base includes: executing vector semantic retrieval and keyword retrieval in parallel in a policy and regulation library segmented according to the structure of document, chapter, article, and paragraph to obtain recalled clauses. And the step of filtering the policy timeliness of the recalled clauses includes: identifying the expiration date and revision status of the recalled clauses through a policy timeliness filter, eliminating the abolished clauses, replacing the revised clauses with the currently effective version with a revision annotation attached.

[0051] The policies and regulations in the government knowledge base are preprocessed and finely segmented and organized according to their inherent hierarchical structure, such as documents, chapters, articles, and clauses. This structured storage method ensures that each part of the policy content has a clear context and independent identifier, facilitating accurate positioning and retrieval by the system and avoiding the inefficiency of fuzzy matching in large documents. To comprehensively and accurately obtain the recalled clauses, this application adopts a dual retrieval mechanism. Vector semantic retrieval converts the user's input business request information and the clause content in the policy and regulation base into high-dimensional vector representations respectively, and calculates the semantic similarity between them to discover potentially relevant clauses, even if these clauses do not contain the exact keywords in the user's query. Keyword retrieval focuses on matching specific words or phrases in the user's query to ensure accurate capture of key information. These two retrieval methods run in parallel, and their results are fused to combine the breadth of semantic understanding and the accuracy of keyword matching, thereby generating a more comprehensive and relevant set of recalled clauses.

[0052] The policy expiration filter is a component specifically designed to assess the validity of policy provisions. It accesses metadata associated with each recalled provision, typically including its effective date, expiration date, and current revision status (e.g., whether it has been repealed, revised, or remains valid). By parsing this metadata, the filter accurately determines whether the recalled provision still has legal effect or applicability at the current point in time. After identifying the expiration date or revision status of a recalled provision, the policy expiration filter removes all provisions explicitly marked as "repealed" or whose expiration date has expired from the recall results. This step ensures that subsequent processing and the final response do not contain any policy information that has lost its validity, thus maintaining the accuracy and authority of government information. For recalled provisions identified as being in a "revised" state, the policy expiration filter does not simply remove them but proactively searches for and retrieves the latest, currently effective version of the provision from the government knowledge base. Subsequently, the system replaces the old version with the new version and adds a revision label to the new version. This label can indicate to the user or subsequent processing module that the terms have been revised, and may include the revision date or a description of the revision to provide a more comprehensive informational context and ensure that the user always receives the latest policy guidance.

[0053] In one embodiment, after the intent recognition of the business request information, the government affairs Q&A and matter handling method further includes: routing and distributing the business request information to the corresponding target intelligent agent according to the recognized intent; wherein, the target intelligent agent includes a policy consultation intelligent agent, a matter handling intelligent agent, a work order circulation intelligent agent, and a public opinion analysis intelligent agent.

[0054] The process involves: retrieving recall clauses from a pre-defined government knowledge base based on the business request information; filtering and weighting the recall clauses based on the metadata of the issuing authority level, effective or expired date, and applicable region corresponding to the recall clauses in the pre-defined government knowledge base to select candidate clauses; inputting the candidate clauses as constraints into a large language model to generate response content, and performing semantic consistency verification between the response content and the corresponding labeled clauses; outputting a response with citation information extracted from the labeled clauses if the verification passes; expanding the search scope in the pre-defined government knowledge base and re-executing the retrieval, generation, and verification steps for the policy consultation agent; initiating multi-round interactions based on the business request information and the corresponding item template for the item handling agent; and assembling and generating work orders based on collected information materials for circulation for the work order circulation agent.

[0055] After identifying the intent behind a business request, the system precisely distributes the request to a pre-defined agent responsible for handling that intent, based on the specific intent identified, such as a policy consultation intent or a service request intent. This routing and distribution mechanism ensures that different types of business requests are processed by the most specialized modules, thereby improving processing efficiency and accuracy. For example, a rule-based routing strategy can be used to directly match the corresponding agent based on the intent label; or a machine learning-based classification model can be used to dynamically assign the request to the most suitable agent based on the semantic features of the intent.

[0056] The target intelligent agents are specialized processing units designed for different functional areas in government services. Among them, the policy consultation intelligent agent is specifically responsible for handling users' policy consultation requests. Its core function is to accurately retrieve, filter, and generate responses that comply with policies and regulations from massive amounts of government knowledge. The service processing intelligent agent focuses on guiding users through the processing procedures of various government services, collecting necessary information through multiple rounds of interaction to ensure users can successfully submit their applications. The work order flow intelligent agent is responsible for standardizing and structuring the collected user processing information and transforming it into work orders that can be processed by back-end departments, achieving automated submission and tracking of work orders. The public opinion analysis intelligent agent focuses on in-depth analysis of data generated during the government service process, mining user needs, identifying hot issues, and providing data support for the optimization of government services.

[0057] Upon receiving a policy consultation request from the routing system, the policy consultation agent takes over the entire policy consultation processing flow. Specifically, it searches a pre-defined government knowledge base to retrieve relevant recall clauses based on the request information. Then, using metadata associated with the recall clauses in the knowledge base, such as the issuing authority level, effective or expiration date, and applicable region, it filters and weights the recall clauses based on policy timeliness to select the candidate clauses that best meet the user's current needs. Next, the policy consultation agent uses these candidate clauses as constraints and inputs them into a large language model to generate accurate and standardized responses. After generating the response, the agent performs semantic consistency verification between the response content and the corresponding labeled clauses to ensure the accuracy and authority of the response. Only responses that pass verification are output to the user, along with citation information. If verification fails, the policy consultation agent automatically expands the search scope and re-executes the above search, generation, and verification steps until a high-quality response is obtained.

[0058] Upon receiving a service request from the routing system, the intelligent agent for handling service requests will initiate and manage multiple rounds of interaction with the user. Based on the service request type, the agent automatically matches the appropriate service template and guides the user through a dialogue-based process, providing all necessary information and materials according to the pre-defined list of materials and information requirements. This interaction method ensures the completeness and accuracy of the processing information, laying the foundation for the smooth generation of subsequent work orders.

[0059] After the task handling intelligent agent completes the collection of information materials, it receives this information and is responsible for standardizing and assembling it to generate a structured work order. This intelligent agent ensures that the work order content is complete and the format is standardized, and submits the generated work order to the corresponding back-end business system or staff according to the preset flow rules, thereby realizing the automated flow and processing of government affairs.

[0060] By introducing the target intelligent agent and realizing the routing and distribution of business requests through the above technical solution, this application effectively solves the problems of high coupling, maintenance difficulties and difficulty in professional processing of complex and diverse business scenarios that may exist in traditional government service systems.

[0061] In one embodiment, initiating multi-round interactions based on the business request information and a corresponding item template includes: The intelligent agent for handling matters initiates multiple rounds of interactive slot filling based on a preset list of required materials, guiding users to complete the application information and supporting documents needed for the matter.

[0062] The intelligent agent for handling administrative matters uses a pre-defined template of required documents. This template defines structured slots for all application information and supporting documents needed for a specific government service. The agent then engages in multi-round, slot-filling interactions with the user, asking for or prompting the user to provide the necessary information item by item and guiding them to upload the corresponding supporting documents. For example, for "applying for a business license," the template might include slots for "company name," "registered address," "business scope," and "copy of legal representative's ID card." The agent will sequentially ask the user for this information and provide an interface for uploading attachments. This interaction method ensures the comprehensiveness, accuracy, and standardization of information collection, avoiding the omission of key information or the collection of non-standard materials.

[0063] In one embodiment, such as Figure 4 As shown, the process of assembling and generating work orders based on collected information materials for circulation includes: Step S410: The collected application information and material attachments are assembled into a structured work order containing the matter type, responsible department, processing time limit and material attachments through the work order flow intelligent agent; Step S420: Desensitize the personal sensitive information in the structured chemical form; Step S430: Transfer the desensitized structured chemical order to the corresponding department's business system or grid worker's terminal, and push the acceptance status and query number to the user terminal.

[0064] After receiving all application information and supporting documents collected by the intelligent agent for handling work orders, the agent will structurally integrate this information according to a pre-set work order template. Specifically, the agent will identify and populate key fields in the work order. For example, based on the user's initial business request or matters confirmed in multiple rounds of interaction, it will determine the "matter type"; automatically match and assign the "responsible department" to handle the matter based on the matter type; set a "processing time limit" for the matter according to relevant regulations or internal rules; and uniformly archive all "supporting documents" provided by the user. This structured assembly ensures the integrity, standardization, and machine-processable nature of the work order information, laying the foundation for subsequent automated workflow and manual approval.

[0065] After the structured work order is assembled, the work order workflow intelligent agent will de-identify the sensitive personal information contained therein. For example, for information such as ID card numbers, mobile phone numbers, and bank card numbers, partial masking (e.g., displaying the first and last few digits and replacing the middle with asterisks), encrypted storage, or replacement with anonymous identifiers can be used. This measure aims to protect users' personal privacy and data security to the greatest extent possible, preventing the risk of sensitive information leakage during work order workflow, storage, or access by unauthorized personnel, while ensuring the compliance of business processing.

[0066] The anonymized structured work orders will be automatically routed to their corresponding government department's business system, such as the administrative approval system or social security system, through pre-defined interfaces or integration mechanisms, or directly sent to the grid worker's terminal responsible for the area or matter, so that relevant staff can conduct subsequent review and processing. Simultaneously, to enhance user experience and transparency, the system will immediately push a notification of the work order's processing status to the user's terminal device (such as a mobile app, WeChat mini-program, or SMS), along with a unique query number. Users can use this query number to check the processing progress and results of the work order at any time, thereby achieving transparency and traceability throughout the entire government service process.

[0067] In one embodiment, such as Figure 5 As shown, after the desensitized structured application form is transferred to the corresponding departmental business system or grid worker terminal, the government affairs Q&A and application processing method further includes: Step S510: The public opinion analysis intelligent agent performs hotspot clustering analysis on the completed consultation data and work order data to generate a public opinion analysis report. Step S520: Based on the public opinion analysis report, extract high-frequency unresolved issues and hot public needs, and use them to update the content of the preset government knowledge base and optimize the retrieval weight parameters.

[0068] The aforementioned public opinion analysis intelligent agent is an intelligent module used to collect, analyze, and mine government service data (including consultation data and work order data). Its main function is to discover patterns, identify hot topics, and gain insights into public opinion from massive amounts of user interaction data, providing data support for the continuous optimization of government services. This intelligent agent can be deployed as an independent software service, receiving and processing data from other modules through interfaces.

[0069] The aforementioned hotspot clustering analysis of completed consultation and work order data aims to identify common themes, frequently occurring issues, or points of concern from a large number of completed consultations and work orders. First, the system collects completed consultation data (such as user questions, system responses, and user feedback) and work order data (such as matter type, submitted materials, processing results, and user evaluations). Then, this text data undergoes preprocessing, such as word segmentation, stop word removal, and part-of-speech tagging, transforming it into analyzable structured features. Next, unsupervised learning algorithms, such as K-means, DBSCAN, or hierarchical clustering, are used to cluster the processed data, grouping data with similar semantics or consistent themes into one category. After clustering, clusters with large data volumes and high keyword frequencies are identified as hotspots by statistically analyzing indicators such as the amount of data and keyword frequency within each cluster.

[0070] The generated public opinion analysis report presents the results of cluster analysis in a structured and visual format, facilitating understanding and decision-making by management personnel. This report may include a list of hot topics, trends in consultation and work order volumes for each topic, a cloud of relevant keywords, user sentiment analysis results (analyzing user feedback through a sentiment analysis model), a list of frequently missed questions, and summaries of key public needs. This report can be automatically generated periodically and pushed to relevant departments via the management backend or email to provide decision support.

[0071] The extraction of frequently missed issues and key public concerns based on the public opinion analysis report involves shifting the focus from a macro-level analysis to specific, actionable optimization points. Frequently missed issues can be identified by analyzing consultation data where the system failed to provide satisfactory answers (e.g., users repeatedly followed up, were transferred to human agents, or received unsatisfactory user feedback) and these issues occur frequently. These problems may indicate a lack of relevant content in the pre-set government knowledge base or that the existing content is inaccurate. Key public concerns can be identified by analyzing work order data for items with high processing volume and user attention, or items frequently mentioned in consultation data but for which no clear policies or services have yet been developed. Extraction methods can combine keyword matching, topic models (such as LDA), or expert rules to automatically identify and extract these issues from the report.

[0072] The process of updating and optimizing retrieval weight parameters in the preset government knowledge base involves applying the analysis results to the system's own optimization, forming a closed loop and improving the intelligence level of government services. For frequently missed questions, the knowledge base administrator can be triggered to manually review and supplement the information, such as adding relevant policy clauses, FAQ entries, or improving the descriptions of existing content. For hot-button public concerns, relevant departments can be encouraged to research and formulate new policies or services and incorporate them into the knowledge base. Simultaneously, for hot-button issues or needs, their retrieval weight in the preset government knowledge base can be increased, making them easier to recall. For example, the relevance weight in the vector semantic retrieval model or the term frequency-inverse document frequency (TF-IDF) weight in keyword retrieval can be adjusted to ensure that this hot-button information is prioritized when users search for relevant content. This can be automatically adjusted through machine learning models or manually configured by administrators based on report suggestions.

[0073] For example, in a specific embodiment where the public opinion analysis agent extracts high-frequency missed questions and feeds back to optimize retrieval weights, the system performs word frequency and semantic clustering on missed questions based on a preset statistical period (such as weekly or monthly) to extract high-frequency missed feature words; then, it achieves dynamic optimization by adjusting the weight factors in the multi-path retrieval fusion algorithm. Specifically, for high-frequency missed feature words, the system automatically increases their word frequency weight factor (IDF / TF weighting) in the BM25 keyword retrieval route, or in the hybrid reranking process of vector retrieval and keyword retrieval, it increases the score weighting coefficient of the inverted index channel containing the high-frequency missed feature word (e.g., from the default 0.4 to 0.6-0.8), thereby improving the recall rate and ranking priority of subsequent business request information containing the same or similar feature words, realizing the automatic self-evolution and feedback loop of retrieval performance.

[0074] Through the above technical solution, this application can perform hotspot clustering analysis on the massive amount of consultation data and work order data after completion, deeply explore user needs and system shortcomings, and generate a public opinion analysis report with guiding significance.

[0075] This application also provides a government affairs Q&A and service processing system for implementing the above-mentioned government affairs Q&A and service processing methods, wherein the government affairs Q&A and service processing system includes: The request acquisition and intent recognition module is used to acquire business request information input by the user and to perform intent recognition on the business request information; The storage module stores a preset government knowledge base, which includes metadata about the issuing authority level, effective or expiration date, and applicable region corresponding to the recall clauses. The policy consultation processing module is used to retrieve recall clauses from a preset government knowledge base based on the business request information when the identified intent is policy consultation; based on the metadata in the preset government knowledge base corresponding to the issuing authority level, effective or expired date, and applicable region of the recall clauses, the module performs policy timeliness filtering and weighted ranking on the recall clauses to select candidate clauses; input the candidate clauses as constraints into a large language model to generate response content, and perform semantic consistency verification between the response content and the corresponding labeled clauses; if the verification passes, output a response with citation information extracted from the labeled clauses; if the verification fails, expand the search scope in the preset government knowledge base and re-execute the search, generation, and verification steps. The task processing module is used to initiate multiple rounds of interaction based on the business request information and the corresponding task template when the identified intent is to process a task. This is done to guide the user to complete the information materials required for processing the task and to assemble and generate a work order based on the collected information materials for processing.

[0076] In one embodiment, the request acquisition and intent recognition module is used to acquire business request information input by the user and to perform intent recognition on the business request information. This module receives text or voice input submitted by the user through an API interface, and uses a pre-trained intent classification model to analyze the semantic features of the business request information, accurately distinguishing between policy consultation intent and matter handling intent. For example, when a user inputs "How to handle social security transfer," the system recognizes it as a matter handling intent; when a user inputs "What is the latest individual income tax threshold?", the system recognizes it as a policy consultation intent.

[0077] The storage module stores a pre-defined government affairs knowledge base, which includes metadata related to the issuing authority level, effective or expiration date, and applicable region corresponding to the recall clauses. This knowledge base is constructed using a distributed database, with each policy and regulation data segmented according to document, chapter, article, and clause structure, and associated with metadata fields. For example, a policy clause concerning housing provident funds has metadata including the issuing authority level (provincial level), effective date (January 1, 2023), expiration date (none), and applicable region (the entire province), ensuring the integrity and searchability of the data structure.

[0078] Through the aforementioned technical solutions, this application achieves high-accuracy question-and-answer and seamless business processing automation at the system level. The policy timeliness filtering mechanism ensures responses are based on current and valid policies, the weighted ranking mechanism enhances the authority and regional adaptability of candidate clauses, the semantic consistency verification and closed-loop error correction mechanism significantly reduce the risk of policy illusion, and the dual-track parallel processing mechanism seamlessly connects consultation and processing scenarios, providing users with a one-stop service experience from policy interpretation to business implementation, effectively addressing the shortcomings in government services.

[0079] This application also provides a storage medium, which is a computer-readable storage medium, and stores a computer program thereon. When the computer program is executed by a processor, it implements the above-described government affairs Q&A and matter handling methods.

[0080] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0081] This application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the government affairs Q&A and matter handling methods described above.

[0082] Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the government affairs Q&A and handling methods provided in the above embodiments, and will not be repeated here.

[0083] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for handling government affairs inquiries and matters, characterized in that, The aforementioned government affairs Q&A and procedures for handling matters include: Obtain user-inputted business request information and perform intent recognition on the business request information; If the identified intent is policy consultation, recall clauses are retrieved from a pre-defined government knowledge base based on the business request information. Based on metadata in the pre-defined government knowledge base corresponding to the recall clauses, including the issuing authority level, effective or expired date, and applicable region, the recall clauses are filtered and weighted according to policy timeliness to select candidate clauses. These candidate clauses are then input into a large language model as constraints to generate a response. Clauses used to constrain the response generation are extracted from the candidate clauses as annotation clauses, and semantic consistency is verified between the response content and the corresponding annotation clauses. If the verification passes, a response with citation information extracted from the annotation clauses is output. If the verification fails, the search scope in the pre-defined government knowledge base is expanded, and the search, generation, and verification steps are re-executed. If the identified intent is to handle a matter, multiple rounds of interaction are initiated according to the business request information and the matter template corresponding to the business request information to guide the user to supplement the information materials required to handle the matter, and a work order is generated based on the collected information materials for circulation.

2. The government affairs Q&A and service processing method according to claim 1, characterized in that, The semantic consistency check between the response content and the corresponding labeled clauses includes: The response content is broken down into independent factual assertions. A natural language reasoning model is called to determine the semantic implication relationship between each independent factual assertion and its corresponding labeled clause, and the implication confidence score of each independent factual assertion is calculated. The implied confidence scores of each independent factual assertion are compared with a preset consistency threshold. When the implied confidence scores of all independent factual assertions are greater than the preset consistency threshold, the verification is deemed to have passed. When the implied confidence score of any independent factual assertion is not greater than the preset consistency threshold, the verification is deemed to have failed. If the response fails to be validated a set number of times consecutively, the current session will be switched to a human agent for processing.

3. The government affairs Q&A and service processing method according to claim 1, characterized in that, The process of obtaining user-inputted service request information includes: Obtain business request information input by the user via voice; The built-in dialect speech recognition module performs dialect acoustic feature recognition and text normalization processing on the speech input, converts the speech input into standard text, and then performs the intent recognition step.

4. The government affairs Q&A and service processing method according to claim 1, characterized in that, The recall clauses retrieved from the preset government knowledge base include: In the policy and regulation database segmented by document, chapter, article, and clause structure, vector semantic retrieval and keyword retrieval are performed in parallel to obtain the recalled clauses; The policy timeliness filtering of the recall terms includes: The policy expiration filter identifies the expiration date and revision status of recall clauses, removes repealed clauses, and replaces revised clauses with the current effective version with a revision note.

5. The government affairs Q&A and handling method according to any one of claims 1 to 4, characterized in that, After performing intent recognition on the business request information, the government affairs Q&A and service processing method further includes: Based on the identified intent, the business request information is routed and distributed to the corresponding target intelligent agents; wherein, the target intelligent agents include policy consultation intelligent agents, service processing intelligent agents, work order circulation intelligent agents, and public opinion analysis intelligent agents; The process involves retrieving recall clauses from a pre-defined government knowledge base based on the business request information; filtering and weighting the recall clauses based on the metadata in the pre-defined government knowledge base corresponding to the issuing authority level, effective or expired date, and applicable region, to select candidate clauses; inputting the candidate clauses as constraints into a large language model to generate response content, and performing semantic consistency verification between the response content and the corresponding labeled clauses; if the verification passes, outputting a response with citation information extracted from the labeled clauses; if the verification fails, expanding the search scope in the pre-defined government knowledge base and re-executing the retrieval, generation, and verification steps for the policy consultation agent to execute. The process involves initiating multiple rounds of interaction based on the business request information and the corresponding item template, to execute the task handling intelligent agent. The process involves assembling and generating work orders based on the collected information materials, which are then processed by the intelligent agent that handles the work order processing.

6. The government affairs Q&A and service processing method according to claim 5, characterized in that, The step of initiating multiple rounds of interaction based on the business request information and according to the item template corresponding to the business request information includes: The intelligent agent for handling matters initiates multi-round, slot-filling interactions based on a preset list of required materials, guiding users to complete the necessary application information and supporting documents; and... The process of assembling and generating work orders based on collected information materials for circulation includes: The work order flow intelligent agent assembles the collected application information and material attachments into a structured work order that includes the type of matter, responsible department, processing time limit, and material attachments; The sensitive personal information in the structured chemical form is desensitized; The desensitized structured chemical orders are transferred to the corresponding department's business system or grid worker's terminal, and the acceptance status and query number are pushed to the user's terminal.

7. The government affairs Q&A and service processing method according to claim 6, characterized in that, After the desensitized structured chemical application is transferred to the corresponding departmental business system or grid worker terminal, the government affairs Q&A and application processing method further includes: The public opinion analysis intelligent agent performs hotspot clustering analysis on the consultation data and work order data after the cases are closed, and generates a public opinion analysis report. Based on the public opinion analysis report, frequently missed issues and hot public needs are extracted, and the content of the preset government knowledge base is updated and the search weight parameters are optimized.

8. A government affairs inquiry and service processing system, characterized in that, Used to implement the method according to any one of claims 1 to 7.

9. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the government affairs Q&A and matter handling method as described in any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the government affairs Q&A and matter handling method as described in any one of claims 1 to 7.

Images