A BERT-based intelligent allocation method for grassroots governance events

Abstract

The present invention relates to a method for intelligent distribution of grassroots governance events based on BERT, comprising the following steps: S1: Collect grassroots event data, perform data preprocessing, and use the built-in word segmentation tool of the BERT model to perform word segmentation processing on the event description text; S2: Select the pre-trained BERT model as the basis, perform domain fine-tuning and model evaluation on it; S3: Use the fine-tuned BERT model to perform event analysis, intelligent matching and make distribution decisions on grassroots events; S4: Collect the processing results and feedback after distribution, evaluate the distribution effect, and regularly update the BERT model. The present invention has the advantages of automating the processing flow, improving the efficiency and accuracy of event processing, and reducing the burden of manual distribution.

Description

Technical Field

[0001] The present invention relates to the field of grassroots governance technology, and in particular to a BERT-based grassroots governance event intelligent distribution method. Background Art

[0002] Grassroots governance is an important manifestation of governance capacity. As an important part of the national governance system, grassroots governance institutions bear the important responsibility of implementing policies, regulations, and systems. There are some drawbacks in manually handling grassroots issues and incidents in grassroots governance, which may affect the efficiency and effectiveness of governance and may even aggravate the complexity of social problems and contradictions. They mainly include:

[0003] 1. Inefficiency; Layered distribution: In the past, manual distribution of grassroots governance events often required a multi-level distribution process such as "city-district-street", which not only increased the circulation links, but also prolonged the processing time of the event; incorrect distribution: In the case of a large number of event fields, manual distribution is prone to errors, resulting in the event being distributed to inappropriate departments or personnel, further increasing the difficulty and time cost of processing;

[0004] 2. High subjectivity; judgment bias: manual allocation often relies on the experience and subjective judgment of the allocation personnel, which may be biased, resulting in misjudgment of events or inaccurate allocation; unfairness: in some cases, manual allocation may be affected by human factors, such as relationships, interests, etc., resulting in unfair allocation results;

[0005] 3. Waste of resources; waste of human resources: due to the low efficiency of manual distribution, more manpower may be required to be invested in the distribution work, resulting in a waste of human resources; waste of time resources: long-term circulation and waiting not only increase the time cost of event processing, but also may cause some emergency events to not be handled in time, resulting in greater losses;

[0006] 4. Technical limitations; difficulty in responding to large-scale incidents: When faced with large-scale and complex grassroots governance incidents, manual distribution may be difficult to handle all incidents in a timely and accurate manner; lack of intelligent support: Compared with intelligent distribution, manual distribution lacks the support of intelligent technology and is difficult to achieve rapid identification, classification and distribution of incidents. Summary of the invention

[0007] In view of the above-mentioned deficiencies in the prior art, the technical problem to be solved by this patent application is how to provide an intelligent distribution method for grassroots governance events based on BERT to provide an automated processing flow, improve the efficiency and accuracy of event processing, and reduce the burden of manual distribution.

[0008] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0009] A method for intelligent allocation of grassroots governance events based on BERT, comprising the following steps:

[0010] S1: Collect grassroots event data, perform data preprocessing, and use the built-in word segmentation tool of the BERT model to segment the event description text;

[0011] S2: Select the pre-trained BERT model as the basis, perform domain fine-tuning and model evaluation on it;

[0012] S3: Use the fine-tuned BERT model to perform event analysis, intelligent matching, and allocation decisions on grassroots events;

[0013] S4: Collect the processing results and feedback after allocation, evaluate the allocation effect, and regularly update the BERT model.

[0014] As an optimization, in step S1, the grassroots event data collected includes but is not limited to event description, time, location, and persons involved; the channels for collecting the event data include but are not limited to social media, hotlines, and government websites.

[0015] As an optimization, in step S1, the data preprocessing includes cleaning the collected data, removing noise and duplicates, and formatting the data to ensure the consistency and accuracy of the data.

[0016] As an optimization, in step S2, the pre-trained BERT model includes one of BERT-Base and BERT-Large; when fine-tuning it, event classification and keyword extraction are used as tasks as adjustment strategies.

[0017] As an optimization, when evaluating the model in step S2, the performance of the fine-tuned model is evaluated by cross-validation to ensure its accuracy and stability in event classification and allocation tasks.

[0018] As an optimization, when performing intelligent matching in step S3, the time analysis is used to extract information on event type, urgency, and geographic location, and the information is matched with the preset distribution rules in combination with the domain knowledge graph or rule base to determine the distribution object.

[0019] As an optimization, when making a distribution decision in step S3, a distribution decision is automatically generated based on the result of intelligent matching, including specifying a processing department and assigning task priority.

[0020] As an optimization, in step S4, when evaluating the distribution effect, the evaluation is performed by comparing the accuracy and efficiency indicators of manual distribution and intelligent distribution.

[0021] As an optimization, the overall architecture of the intelligent distribution system is designed when making distribution decisions, including data input layer, model processing layer, and result output layer; the data input layer is responsible for receiving data input of grassroots governance events; the model processing layer uses the BERT model to classify and process events; the result output layer presents the classification results to users or relevant processing departments in an appropriate manner.

[0022] The present invention has the following advantages:

[0023] 1. Efficiently handle a large number of events; With its powerful natural language processing capabilities, the BERT model can quickly understand and analyze the text content of grassroots events, and achieve fast and accurate classification and allocation, which is particularly important for handling a large number of complex grassroots events and can significantly improve processing efficiency.

[0024] 2. Reduce human errors. Compared with manual allocation, the BERT model can avoid judgment bias and erroneous allocation caused by human factors. It automatically handles events through algorithms, reduces human intervention, and improves the accuracy and fairness of allocation.

[0025] 3. Improve service quality; the intelligent distribution system can ensure that events are promptly and accurately assigned to the corresponding processing departments or personnel, thereby speeding up processing and improving service quality, which is of great significance for improving public satisfaction and trust.

[0026] 4. Support multi-field applications: The BERT model has wide transferability and can be applied to grassroots event distribution in different fields. Whether it is urban management, community services or public safety, the BERT model can be used to achieve intelligent distribution and improve governance efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 This is a flow chart of a method for intelligent distribution of grassroots governance events based on BERT described in the present invention. DETAILED DESCRIPTION

[0028] The present invention is further described in detail below in conjunction with the accompanying drawings. In the description of the present invention, it should be understood that the directions or positional relationships indicated by directional words such as "upper, lower" and "top, bottom" are usually based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description. Unless otherwise stated, these directional words do not indicate or imply that the devices or components referred to must have a specific direction or be constructed and operated in a specific direction, and therefore cannot be understood as limiting the scope of protection of the present invention; the directional words "inside and outside" refer to the inside and outside relative to the contours of each component itself.

[0029] like Figure 1As shown, a method for intelligent distribution of grassroots governance events based on BERT includes the following steps: S1: Collect grassroots event data, perform data preprocessing, and use the built-in word segmentation tool of the BERT model to segment the event description text; S2: Select the pre-trained BERT model as the basis, perform domain fine-tuning and model evaluation on it; S3: Use the fine-tuned BERT model to parse, intelligently match and make distribution decisions for grassroots events; S4: Collect the processing results and feedback after distribution, evaluate the distribution effect, and update the BERT model regularly.

[0030] In this embodiment, in step S1, the grassroots event data collected includes but is not limited to event description, time, location, and persons involved; the channels for collecting the event data include but are not limited to social media, hotlines, and government websites.

[0031] In this embodiment, in step S1, the data preprocessing includes cleaning the collected data, removing noise and duplicates, and formatting the data to ensure the consistency and accuracy of the data.

[0032] Specifically, text cleaning includes removing irrelevant symbols, stop words, word segmentation, and tagging.

[0033] In this embodiment, in step S2, the pre-trained BERT model includes one of BERT-Base and BERT-Large; when fine-tuning it, event classification and keyword extraction are used as tasks as adjustment strategies.

[0034] Specifically, during the training process, the classification accuracy and generalization ability of the model are improved by adjusting model parameters, optimizing training strategies, etc.

[0035] In this embodiment, when the model is evaluated in step S2, the performance of the fine-tuned model is evaluated by cross-validation to ensure its accuracy and stability in event classification and allocation tasks.

[0036] In this embodiment, when performing intelligent matching in step S3, the time analysis is used to extract information on event type, urgency, and geographic location, and the information is matched with the preset distribution rules in combination with the domain knowledge graph or rule base to determine the distribution object.

[0037] In this embodiment, when making a distribution decision in step S3, a distribution decision is automatically generated based on the result of intelligent matching, including specifying a processing department and assigning a task priority.

[0038] In this embodiment, in step S4, when evaluating the distribution effect, the evaluation is performed by comparing the accuracy and efficiency indicators of manual distribution and intelligent distribution.

[0039] In this embodiment, the overall architecture of the intelligent distribution system is designed when making distribution decisions, including a data input layer, a model processing layer, and a result output layer; the data input layer is responsible for receiving data input of grassroots governance events; the model processing layer uses the BERT model to classify and process events; and the result output layer presents the classification results to users or relevant processing departments in an appropriate manner.

[0040] Specifically, it also includes the automated event distribution process. When new event data is input into the system, the system automatically calls the BERT model for classification and distribution, and distributes the event to the corresponding processing department or personnel based on the classification results.

[0041] The present invention has the following advantages:

[0042] 1. Efficiently handle a large number of events; With its powerful natural language processing capabilities, the BERT model can quickly understand and analyze the text content of grassroots events, and achieve fast and accurate classification and allocation, which is particularly important for handling a large number of complex grassroots events and can significantly improve processing efficiency.

[0043] 2. Reduce human errors. Compared with manual allocation, the BERT model can avoid judgment bias and erroneous allocation caused by human factors. It automatically handles events through algorithms, reduces human intervention, and improves the accuracy and fairness of allocation.

[0044] 3. Improve service quality; the intelligent distribution system can ensure that events are promptly and accurately assigned to the corresponding processing departments or personnel, thereby speeding up processing and improving service quality, which is of great significance for improving public satisfaction and trust.

[0045] 4. Support multi-field applications: The BERT model has wide transferability and can be applied to grassroots event distribution in different fields. Whether it is urban management, community services or public safety, the BERT model can be used to achieve intelligent distribution and improve governance efficiency.

[0046] Finally, it should be noted that those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for intelligent allocation of grassroots governance events based on BERT, characterized in that: The following steps are involved: S1: Collect grassroots event data, perform data preprocessing, and use the built-in word segmentation tool of the BERT model to segment the event description text; S2: Select the pre-trained BERT model as the basis, perform domain fine-tuning and model evaluation on it; S3: Use the fine-tuned BERT model to perform event analysis, intelligent matching, and allocation decisions on grassroots events; S4: Collect the processing results and feedback after allocation, evaluate the allocation effect, and regularly update the BERT model.

2. According to claim 1, a BERT-based grassroots governance event intelligent allocation method is characterized in that: In step S1, the grassroots event data collected include but are not limited to event description, time, location, and persons involved; the channels for collecting the event data include but are not limited to social media, hotlines, and government websites.

3. According to claim 2, a BERT-based intelligent allocation method for grassroots governance events is characterized in that: In step S1, the data preprocessing includes cleaning the collected data, removing noise and duplicates, and formatting the data to ensure the consistency and accuracy of the data.

4. According to claim 3, a BERT-based grassroots governance event intelligent allocation method is characterized in that: In step S2, the pre-trained BERT model includes one of BERT-Base and BERT-Large; When fine-tuning it, event classification and keyword extraction are used as tasks as adjustment strategies.

5. According to claim 4, a BERT-based intelligent allocation method for grassroots governance events is characterized in that: When evaluating the model in step S2, the performance of the fine-tuned model is evaluated by cross-validation to ensure its accuracy and stability in event classification and allocation tasks.

6. According to claim 5, a method for intelligent allocation of grassroots governance events based on BERT is characterized in that: When performing intelligent matching in step S3, the time analysis is used to extract information on event type, urgency, and geographic location, and the information is matched with the preset distribution rules in combination with the domain knowledge graph or rule base to determine the distribution object.

7. According to claim 6, a BERT-based intelligent allocation method for grassroots governance events is characterized in that: When making a distribution decision in step S3, a distribution decision is automatically generated based on the result of intelligent matching, including specifying a processing department and assigning task priority.

8. According to claim 7, a method for intelligent allocation of grassroots governance events based on BERT is characterized in that: In step S4, when evaluating the distribution effect, the evaluation is performed by comparing the accuracy and efficiency indicators of manual distribution and intelligent distribution.

9. According to claim 8, a BERT-based intelligent allocation method for grassroots governance events is characterized in that: When making allocation decisions, the overall architecture of the intelligent allocation system is designed, including the data input layer, model processing layer, and result output layer; the data input layer is responsible for receiving data input of grassroots governance events; The model processing layer uses the BERT model to classify events; The result output layer presents the classification results to users or relevant processing departments in an appropriate manner.

Images