CN-122022363-A - Government affair dispatching method and device

CN122022363ACN 122022363 ACN122022363 ACN 122022363ACN-122022363-A

Abstract

The invention relates to a government affair dispatching method and device, and relates to the technical field of government affair dispatching. The government affair dispatching method comprises the steps of obtaining government affair text information to be processed, extracting key information of the government affair text information according to preset rules, matching the key information with knowledge information of a knowledge base to determine first knowledge information obtained by matching, inputting the first knowledge information and the key information into a large language model, analyzing the first knowledge information and the key information through the large language model to determine first dispatching information, checking the first dispatching information, and outputting the first dispatching information under the condition that a check result meets a check standard, wherein the knowledge information comprises government department information, policy information and historical work order information. The first order-sending information is determined progressively through the knowledge base and the large language model, the government affair order-sending decision process is optimized in multiple stages, the conversion from relying on manual experience to automatic intelligent research and judgment is achieved, and the accuracy and efficiency of government affair order-sending are improved.

Inventors

ZHAO JIE
ZHANG TONG
LIU XIN
SUI LIYANG
ZHOU MENGDONG
LIU LEI

Assignees

青岛开锐数据科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260211

Claims (10)

1. A government affair assignment method, characterized in that the government affair assignment method comprises: Acquiring government affair text information to be processed; extracting key information of the government affair text information according to a preset rule; matching the key information with knowledge information of a knowledge base, and determining first knowledge information obtained by matching; Inputting the first knowledge information and the key information into a large language model, and analyzing the first knowledge information and the key information by using the large language model to determine first dispatch information; checking the first form information, and outputting the first form information under the condition that a check result meets a check standard; the knowledge information comprises government department information, policy information and historical work order information.
2. The government affair dispatch method according to claim 1, wherein after checking the first dispatch information and outputting the first dispatch information if the check result meets the check standard, the method further comprises: Under the condition that the first form dispatching information is executed, marking the first form dispatching information, the first knowledge information and the key information which correspond to the first form dispatching information as reference form dispatching samples; the reference bill sample is classified into the historical work bill information for the subsequent government bill process; In the subsequent government service dispatch process, the priority of the reference dispatch sample is greater than that of the history dispatch sample in the process of matching the key information with the history work order information in the knowledge information.
3. The government affair dispatch method according to claim 1, wherein after checking the first dispatch information and outputting the first dispatch information if the check result meets the check standard, the method further comprises: acquiring a return reason of the first form information under the condition that the first form information is returned; According to the return reason of the first form information, matching the key information with the knowledge information, and determining second knowledge information obtained by matching; fine-tuning the large language model according to the return reason of the first form information, the key information and the first knowledge information; Inputting the second knowledge information, the return reason of the first form information and the key information into the trimmed large language model, and analyzing the second knowledge information, the return reason of the first form information and the key information by using the trimmed large language model to determine second form information; and checking the second form information, and outputting the second form information under the condition that the checking result meets the checking standard.
4. The government affairs dispatch method according to claim 3, wherein after checking the second dispatch information and outputting the second dispatch information if the checking result meets the checking standard, the method further comprises: Marking the second form information and the second knowledge information and the key information corresponding to the second form information as reference form samples under the condition that the second form information is executed; the reference bill sample is classified into the historical work bill information for the subsequent government bill process; In the subsequent government service dispatch process, the priority of the reference dispatch sample is greater than that of the history dispatch sample in the process of matching the key information with the history work order information in the knowledge information.
5. The government affair dispatching method according to claim 1, wherein the matching the key information with knowledge information in a knowledge base to determine first knowledge information obtained by matching includes: matching the key information with the historical work order information, and determining at least one target historical information obtained by matching; under the condition that each target historical information only comprises the same agency, matching the key information with government information, and determining at least one target government information obtained by matching; determining a degree of correlation of each of the target government information and each of the target history information; determining each of the target government information and the corresponding target history information, the degree of correlation of which is greater than a threshold value, as the first knowledge information; when each of the target history information includes a plurality of underwriting departments, corresponding target underwriting information is determined from the government information according to each of the underwriting departments; determining each piece of target history information and the corresponding piece of target underwriting information as the first knowledge information; Wherein the government information includes the government agency information and the policy information, and the target underwriting information includes information related to the underwriting agency determined from the government information.
6. The government affair dispatching method according to claim 1, wherein the preset rules comprise a first preset rule and a second preset rule, and the extracting key information in the government affair text information according to the preset rules comprises: Determining target information of the government text information according to the first preset rule; And extracting the key information in the target information according to the second preset rule.
7. The method of government affair dispatch according to claim 6, wherein the determining the target information of the government affair text information according to the first preset rule includes: Cleaning the government affair text information to obtain first text information; coding the first text information according to a set coding standard to obtain second text information; Analyzing the second text information to determine a core element and key parameters; wherein the target information includes the core element and the key parameter.
8. The method for dispatching government affairs according to claim 1, wherein the checking the first dispatching information and outputting the first dispatching information if the checking result meets the checking standard comprises: determining a first score of the first form information according to a first check rule; determining a second score of the first form information according to a second check rule; Weighting the first score and the second score according to the weight information of the first check rule and the second check rule to obtain a target score; and outputting the first dispatch information under the condition that the target score is larger than the verification score of the verification standard.
9. The government dispatch method of any one of claims 1 to 8, wherein the large language model includes a univocal thousand large language model, and wherein prior to the obtaining the government text information to be processed, the method further comprises: acquiring a government work order data set; Fine-tuning the universal thousand-question large language model according to the government work order data set; Inputting the first knowledge information and the key information into a large language model, analyzing the first knowledge information and the key information by the large language model, and determining first dispatch information, wherein the method comprises the following steps: and inputting the first knowledge information and the key information into the fine-tuned universal-meaning and thousand-question large language model, and analyzing the first knowledge information and the key information by using the fine-tuned universal-meaning and thousand-question large language model to determine the first dispatch information.
10. The utility model provides a government affairs list device, its characterized in that, government affairs list device includes: The acquiring module is configured to acquire government affair text information to be processed; The extraction module is configured to extract key information of the government affair text information according to preset rules; the first determining module is configured to match the key information with knowledge information of a knowledge base and determine first knowledge information obtained by matching; The second determining module is configured to input the first knowledge information and the key information into a large language model, analyze the first knowledge information and the key information by the large language model and determine first form information; The output module is configured to verify the first form information, and output the first form information under the condition that a verification result meets a verification standard; the knowledge information comprises government department information, policy information and historical work order information.

Description

Government affair dispatching method and device Technical Field The invention relates to the technical field of government affair dispatching, in particular to a government affair dispatching method and device. Background Currently, a government affairs dispatch system generally adopts a mode of 'city level-district level-department level' to dispatch. The public submits the appeal to the city interview office through telephone, applet and website, the city interview office generates the work order to be processed with the public appeal, and distributes the interview offices of each district preliminarily, the staff of each district interview office sends the order to the appropriate department according to the specific content of the work order to be processed. However, as the number of work orders to be processed is large, and the description modes of the public appeal are various, the problem of misjudgment exists for staff, and the accuracy of government affair dispatching is low. Disclosure of Invention In order to overcome the problems in the related art, the invention provides a government affair dispatching method and device. According to a first aspect of an embodiment of the present invention, there is provided a government affair dispatching method, including: Acquiring government affair text information to be processed; extracting key information of the government affair text information according to a preset rule; matching the key information with knowledge information of a knowledge base, and determining first knowledge information obtained by matching; Inputting the first knowledge information and the key information into a large language model, and analyzing the first knowledge information and the key information by using the large language model to determine first dispatch information; checking the first form information, and outputting the first form information under the condition that a check result meets a check standard; the knowledge information comprises government department information, policy information and historical work order information. In some embodiments of the present invention, after verifying the first dispatch information and outputting the first dispatch information if the verification result meets the verification standard, the method further includes: Under the condition that the first form dispatching information is executed, marking the first form dispatching information, the first knowledge information and the key information which correspond to the first form dispatching information as reference form dispatching samples; the reference bill sample is classified into the historical work bill information for the subsequent government bill process; In the subsequent government service dispatch process, the priority of the reference dispatch sample is greater than that of the history dispatch sample in the process of matching the key information with the history work order information in the knowledge information. In some embodiments of the present invention, after verifying the first dispatch information and outputting the first dispatch information if the verification result meets the verification standard, the method further includes: acquiring a return reason of the first form information under the condition that the first form information is returned; According to the return reason of the first form information, matching the key information with the knowledge information, and determining second knowledge information obtained by matching; fine-tuning the large language model according to the return reason of the first form information, the key information and the first knowledge information; Inputting the second knowledge information, the return reason of the first form information and the key information into the trimmed large language model, and analyzing the second knowledge information, the return reason of the first form information and the key information by using the trimmed large language model to determine second form information; and checking the second form information, and outputting the second form information under the condition that the checking result meets the checking standard. In some embodiments of the present invention, after verifying the second dispatch information and outputting the second dispatch information if the verification result meets the verification standard, the method further includes: Marking the second form information and the second knowledge information and the key information corresponding to the second form information as reference form samples under the condition that the second form information is executed; the reference bill sample is classified into the historical work bill information for the subsequent government bill process; In the subsequent government service dispatch process, the priority of the reference dispatch sample is greater than that of the history dispatch sample in the process of matching the key info