CN112184133A - Artificial intelligence-based government office system preset approval and division method - Google Patents
Artificial intelligence-based government office system preset approval and division method Download PDFInfo
- Publication number
- CN112184133A CN112184133A CN201910588105.6A CN201910588105A CN112184133A CN 112184133 A CN112184133 A CN 112184133A CN 201910588105 A CN201910588105 A CN 201910588105A CN 112184133 A CN112184133 A CN 112184133A
- Authority
- CN
- China
- Prior art keywords
- model
- training
- office system
- approval
- division
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/10—Office automation; Time management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/374—Thesaurus
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Databases & Information Systems (AREA)
- Entrepreneurship & Innovation (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Computational Linguistics (AREA)
- Quality & Reliability (AREA)
- Biomedical Technology (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Operations Research (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a preset approval and division method for a government office system based on artificial intelligence, which comprises the steps of initial training of a model in the step one and real-time deduction and continuous training in the step two; the initialization training of the first step is composed of three parts, wherein the first part is preprocessing, and the second part is model training for generating the batch. The invention can preset two work achievements needed to be completed by a processor in a government office system before processing, and can be used as a blue book for some modification or directly used as a result, thereby greatly improving the processing speed and accuracy of the link processor, and integrally improving the working efficiency and the working quality of the government office system.
Description
Technical Field
The invention relates to the technical field of natural language data processing, in particular to a preset approval and division method for a government office system based on artificial intelligence.
Background
The government affair office system is a relatively mature system developed in government informatization, and the promotion of the government affair office system enables the government departments to realize paperless office work, greatly improves the working efficiency and also enables the office flow in the government to realize standardization and normalization. Various nodes such as receipt, handling and archiving are arranged in the workflow of the government affair system, each node is a processing link and is provided with a fixed processor, and the processors need to give instructions according to the contents of incoming documents and send tasks to the processors in the next working link.
The work of making approval and next step requires that the processor is very familiar with the department of workers, has wide knowledge of business work, and has rigorous and appropriate speech and fast handling speed. The node also becomes a pain point in the government office process, on one hand, all processes pass through the node, the requirement on the timeliness is high, and once the delay occurs, a lot of subsequent links are influenced; on the other hand, the work requirement is high, and the criticized characters are in accordance with the requirement of government documents, so that few processors can finish the work.
The current government affair office system needs a processor to fill in the approval manually and select the objects for division of work, and both the efficiency and the quality are low.
Disclosure of Invention
Aiming at the situation, in order to overcome the defects of the prior art, the invention provides a method for presetting the approval and the division of labor of the government affair office system based on artificial intelligence, which solves the problems that the existing government affair office system needs to process the manual approval filling and the division of labor selection, and the efficiency and the quality are low.
In order to achieve the purpose, the invention provides the following technical scheme: a government affair office system preset approval and division method based on artificial intelligence comprises the steps of initial training of a model in the step one, and real-time deduction and continuous training in the step two;
the initialization training of the first step consists of three parts, wherein the first part is preprocessing, the second part is model training for generating the approval, and the third part is model training for generating the division of labor;
the preprocessing work needs to convert 'business' data acquired from a government office system into 'vector' data which can be processed by a general machine learning model, the 'vector' conversion needs to separate and code an original text, a word segmentation tool is used according to the characteristics of Chinese, the text is separated by taking 'words' rather than 'words', and the whole names and short names of a unit gate and an individual are taken as a preset dictionary of the word segmentation tool according to the characteristics of government, so that the word segmentation is more accurate, all words in a sample are summarized into a 'dictionary table', each word corresponds to one digit one by one, and the word sequence of the text can be converted into the digit sequence through the mapping of the words and the digits of the dictionary table;
the model training of the generation and the verification is to use a sequence-to-sequence seq2seq model, the input and the output of the model are text sequences, a sequence which is used as input is mapped into a sequence which is used as output through a deep neural network model, and the process consists of two links of encoding input and decoding output;
the model training position for generating the division uses a text classification model, in the technical scheme, mapping of multiple classifications is needed, namely, one text can correspond to multiple classifications, similar to multiple choice questions, and each classification corresponds to a division target, so that the technical scheme uses word vectors and a bidirectional LSTM layer to realize the text classification model;
and step two, real-time deduction and continuous training, namely, by utilizing the model of the nodes obtained by the initial training in the step one, when new unprocessed incoming texts exist, inputting text information into the model, deducting approval and division work information, and presetting the information into a government office system.
According to the above technical scheme, the preprocessing is preprocessing of sample data.
According to the technical scheme, the preset approval and division method further comprises the steps that when a handler handles the work, the handler looks at preset information to work, and can directly submit or submit after modification to form new historical data, and after a certain amount of new historical data are accumulated in the technical scheme, the new historical data are used as a training sample continuous training model.
According to the technical scheme, the historical data of the node in the government office system is divided into the information of the incoming text and the work result of a processor, and the initialization training is carried out by using the process, so that a model suitable for the node is obtained.
According to the technical scheme, the deep neural network model is a plurality of LSTMs, namely long and short memory networks.
According to the above technical solution, the encoding input is several layers of LSTM or RNN.
According to the above technical scheme, the decoding output is several layers of LSTM or RNN.
The invention can preset two work achievements needed to be completed by a processor in a government office system before processing, and can be used as a blue book for some modification or directly used as a result, thereby greatly improving the processing speed and accuracy of the link processor, and integrally improving the working efficiency and the working quality of the government office system.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention. In the drawings:
FIG. 1 is a schematic diagram of the first step of the present invention.
FIG. 2 is a schematic diagram of the second step of the present invention.
Detailed Description
The following describes in further detail embodiments of the present invention with reference to fig. 1-2.
The embodiment is given by figures 1 and 2, and the government office system preset approval and division method based on artificial intelligence comprises the steps of initial training of a model in the step one, and real-time deduction and continuous training in the step two;
the initialization training of the first step consists of three parts, wherein the first part is preprocessing, the second part is model training for generating the approval, and the third part is model training for generating the division of labor;
the preprocessing work needs to convert 'business' data acquired from a government office system into 'vector' data which can be processed by a general machine learning model, the 'vector' conversion needs to separate and code an original text, a word segmentation tool is used according to the characteristics of Chinese, the text is separated by taking 'words' rather than 'words', and the whole names and short names of a unit gate and an individual are taken as a preset dictionary of the word segmentation tool according to the characteristics of government, so that the word segmentation is more accurate, all words in a sample are summarized into a 'dictionary table', each word corresponds to one digit one by one, and the word sequence of the text can be converted into the digit sequence through the mapping of the words and the digits of the dictionary table;
the model training of the generation and the verification is to use a sequence-to-sequence seq2seq model, the input and the output of the model are text sequences, a sequence which is used as input is mapped into a sequence which is used as output through a deep neural network model, and the process consists of two links of encoding input and decoding output;
the model training position for generating the division uses a text classification model, in the technical scheme, mapping of multiple classifications is needed, namely one text can correspond to multiple classifications, similar to multiple choice questions, and each classification corresponds to a division target, so that the technical scheme uses word vectors and a bidirectional LSTM layer to realize the text classification model so as to obtain better accuracy and operation efficiency;
and step two, real-time deduction and continuous training, namely, by utilizing the model of the nodes obtained by the initial training in the step one, when new unprocessed incoming texts exist, inputting text information into the model, deducting approval and division work information, and presetting the information into a government office system.
According to the above technical scheme, the preprocessing is preprocessing of sample data.
According to the technical scheme, the preset approval and division method further comprises the steps that when a processor transacts, the processor looks at preset information to work, and can directly submit or submit after modification to form new historical data, and after a certain amount of new historical data are accumulated in the technical scheme, the new historical data are used as a training sample continuous training model, so that the model can more accurately track and simulate the real work of the processor.
According to the technical scheme, the historical data of the node in the government office system is divided into the information of the incoming text and the work result of a processor, and the initialization training is carried out by using the process, so that a model suitable for the node is obtained.
According to the technical scheme, the deep neural network model is a plurality of LSTMs, namely long and short memory networks.
According to the above technical solution, the encoding input is several layers of LSTM or RNN.
According to the above technical scheme, the decoding output is several layers of LSTM or RNN.
The invention is based on the incoming information accumulated in the history of the processing link and the result made by the processing person at that time, including the approval and the division of labor, and these historical data are used as samples, the technical proposal sets the previous incoming information as x and the made result as y, and trains the model of the rule of x- > y by using the existing sample data, and when there is new incoming, it is new x, and new y can be derived by the model as the preset approval and the division of labor.
In the case, more than forty thousand pieces of sample data are extracted from a business office system, and a relatively accurate model is trained, wherein each piece of sample data comprises incoming information including time, title, incoming department, text number, text and the like, and results made by a processor, including batch display and division work.
Preprocessing is an essential step before training, the purpose of the preprocessing is to convert 'business' data acquired from a business office system into 'vector' data which can be processed by a general machine learning model, the preprocessing can adopt a Hadamard 1tp word segmentation tool or other word segmentation tools supporting a preset dictionary, and a single part gate and a full name and a short name of an individual can be preset in the dictionary, so that word segmentation is more accurate. The words after word segmentation are de-duplicated, one word in one line is induced into one text file, the line number becomes the unique correspondence of the word, so that a dictionary table containing all the words in the sample data is induced, and the word sequence of all the samples can be converted into the number sequence through the mapping of the words and the line number of the dictionary table.
The generation of the criticized model adopts a seq2seq model, and the constructed seq2seq model has satisfactory results in terms of efficiency and accuracy after testing by using two layers of LSTMs in encoding and decoding respectively and using an attention mechanism.
The generated division model adopts a text classification model, and through comparison test, word vectors and a bidirectional LSTM layer are used for realizing the text classification models, so that better accuracy and operation efficiency can be obtained.
And step two, the real-time deduction firstly needs to load a previously trained model, then deduction is carried out by using incoming information to be processed, the deduction result is preset in a government office system, the continuous training process is similar to the initial training process of step one, the previously trained model needs to be loaded firstly only because the original training result needs to be reserved, then training is carried out, and the model is stored for next deduction and continuous training after training is finished.
The invention can preset two work achievements needed to be completed by a processor in a government office system before processing, and can be modified by a blueprint or directly used as a result, thereby greatly improving the processing speed and accuracy of a link processor, integrally improving the working efficiency and the working quality of the government office system.
Finally, it should be noted that: although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that changes may be made in the embodiments and/or equivalents thereof without departing from the spirit and scope of the invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (7)
1. A preset approval and division method for a government office system based on artificial intelligence is characterized by comprising the steps of initial training of a model in the step one and real-time deduction and continuous training in the step two;
the initialization training of the first step consists of three parts, wherein the first part is preprocessing, the second part is model training for generating the approval, and the third part is model training for generating the division of labor;
the preprocessing work needs to convert 'business' data acquired from a government office system into 'vector' data which can be processed by a general machine learning model, the 'vector' conversion needs to separate and code an original text, a word segmentation tool is used according to the characteristics of Chinese, the text is separated by taking 'words' rather than 'words', and the whole names and short names of a unit gate and an individual are taken as a preset dictionary of the word segmentation tool according to the characteristics of government, so that the word segmentation is more accurate, all words in a sample are summarized into a 'dictionary table', each word corresponds to one digit one by one, and the word sequence of the text can be converted into the digit sequence through the mapping of the words and the digits of the dictionary table;
the model training of the generation and the verification is to use a sequence-to-sequence seq2seq model, the input and the output of the model are text sequences, a sequence which is used as input is mapped into a sequence which is used as output through a deep neural network model, and the process consists of two links of encoding input and decoding output;
the model training position for generating the division uses a text classification model, in the technical scheme, mapping of multiple classifications is needed, namely, one text can correspond to multiple classifications, similar to multiple choice questions, and each classification corresponds to a division target, so that the technical scheme uses word vectors and a bidirectional LSTM layer to realize the text classification model;
and step two, real-time deduction and continuous training, namely, by utilizing the model of the nodes obtained by the initial training in the step one, when new unprocessed incoming texts exist, inputting text information into the model, deducting approval and division work information, and presetting the information into a government office system.
2. The method for prearranged approval and division of labor for government offices based on artificial intelligence as claimed in claim 1, wherein said preprocessing is a preprocessing of sample data.
3. The preset approval and division method for government offices based on artificial intelligence according to claim 1, wherein the preset approval and division method further comprises the steps that a processor looks at preset information to work when handling, and can directly submit or modify and submit the preset information to form new historical data, and the technical scheme is used as a training sample continuous training model after accumulating a certain amount of new historical data.
4. The method for prearranged approval and division of labor force for government offices based on artificial intelligence as claimed in claim 1, wherein the historical data of the node in the government office system is divided into information of the incoming text and the work result of the handler, and the model suitable for the node is obtained by performing the initial training through the above process.
5. The artificial intelligence based government office system preset approval and division method according to claim 1, wherein the deep neural network model is a plurality of LSTM — long and short memory networks.
6. The artificial intelligence based government office system preset approval and division method according to claim 1, wherein the code input is a number of layers LSTM or RNN.
7. The artificial intelligence based government office system preset approval and division method according to claim 1, wherein the decoding output is a number of layers of LSTM or RNN.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910588105.6A CN112184133A (en) | 2019-07-02 | 2019-07-02 | Artificial intelligence-based government office system preset approval and division method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910588105.6A CN112184133A (en) | 2019-07-02 | 2019-07-02 | Artificial intelligence-based government office system preset approval and division method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112184133A true CN112184133A (en) | 2021-01-05 |
Family
ID=73914812
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910588105.6A Pending CN112184133A (en) | 2019-07-02 | 2019-07-02 | Artificial intelligence-based government office system preset approval and division method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112184133A (en) |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107832310A (en) * | 2017-11-27 | 2018-03-23 | 首都师范大学 | Structuring argument generation method and system based on seq2seq models |
CN107832400A (en) * | 2017-11-01 | 2018-03-23 | 山东大学 | A kind of method that location-based LSTM and CNN conjunctive models carry out relation classification |
CN108073677A (en) * | 2017-11-02 | 2018-05-25 | 中国科学院信息工程研究所 | A kind of multistage text multi-tag sorting technique and system based on artificial intelligence |
CN108427665A (en) * | 2018-03-15 | 2018-08-21 | 广州大学 | A kind of text automatic generation method based on LSTM type RNN models |
CN108763539A (en) * | 2018-05-31 | 2018-11-06 | 华中科技大学 | A kind of file classification method and system based on parts of speech classification |
CN109299273A (en) * | 2018-11-02 | 2019-02-01 | 广州语义科技有限公司 | Based on the multi-source multi-tag file classification method and its system for improving seq2seq model |
CN109493265A (en) * | 2018-11-05 | 2019-03-19 | 北京奥法科技有限公司 | A kind of Policy Interpretation method and Policy Interpretation system based on deep learning |
CN109636345A (en) * | 2018-12-14 | 2019-04-16 | 武汉烽火信息集成技术有限公司 | A kind of intelligent management and system of business handling workflow |
CN109740169A (en) * | 2019-01-09 | 2019-05-10 | 北京邮电大学 | A kind of Chinese medical book interpretation method based on dictionary and seq2seq pre-training mechanism |
-
2019
- 2019-07-02 CN CN201910588105.6A patent/CN112184133A/en active Pending
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107832400A (en) * | 2017-11-01 | 2018-03-23 | 山东大学 | A kind of method that location-based LSTM and CNN conjunctive models carry out relation classification |
CN108073677A (en) * | 2017-11-02 | 2018-05-25 | 中国科学院信息工程研究所 | A kind of multistage text multi-tag sorting technique and system based on artificial intelligence |
CN107832310A (en) * | 2017-11-27 | 2018-03-23 | 首都师范大学 | Structuring argument generation method and system based on seq2seq models |
CN108427665A (en) * | 2018-03-15 | 2018-08-21 | 广州大学 | A kind of text automatic generation method based on LSTM type RNN models |
CN108763539A (en) * | 2018-05-31 | 2018-11-06 | 华中科技大学 | A kind of file classification method and system based on parts of speech classification |
CN109299273A (en) * | 2018-11-02 | 2019-02-01 | 广州语义科技有限公司 | Based on the multi-source multi-tag file classification method and its system for improving seq2seq model |
CN109493265A (en) * | 2018-11-05 | 2019-03-19 | 北京奥法科技有限公司 | A kind of Policy Interpretation method and Policy Interpretation system based on deep learning |
CN109636345A (en) * | 2018-12-14 | 2019-04-16 | 武汉烽火信息集成技术有限公司 | A kind of intelligent management and system of business handling workflow |
CN109740169A (en) * | 2019-01-09 | 2019-05-10 | 北京邮电大学 | A kind of Chinese medical book interpretation method based on dictionary and seq2seq pre-training mechanism |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109359293B (en) | Mongolian name entity recognition method neural network based and its identifying system | |
CN109885824B (en) | Hierarchical Chinese named entity recognition method, hierarchical Chinese named entity recognition device and readable storage medium | |
CN111046946B (en) | Burma language image text recognition method based on CRNN | |
CN109145290B (en) | Semantic similarity calculation method based on word vector and self-attention mechanism | |
CN109543181B (en) | Named entity model and system based on combination of active learning and deep learning | |
CN111222305A (en) | Information structuring method and device | |
CN110673840B (en) | Automatic code generation method and system based on tag graph embedding technology | |
US11636341B2 (en) | Processing sequential interaction data | |
CN110472688A (en) | The method and device of iamge description, the training method of image description model and device | |
CN107491444B (en) | Parallel word alignment method based on bilingual word embedding technology | |
CN109857846B (en) | Method and device for matching user question and knowledge point | |
CN111078847A (en) | Power consumer intention identification method and device, computer equipment and storage medium | |
CN108829823A (en) | A kind of file classification method | |
CN111125380B (en) | Entity linking method based on RoBERTa and heuristic algorithm | |
CN109241268A (en) | A kind of analog information recommended method, device, equipment and storage medium | |
JP2019091172A (en) | Phrase structure learning device, phrase structure analysis device and method, and program | |
CN112527986A (en) | Multi-round dialog text generation method, device, equipment and storage medium | |
CN114238373A (en) | Method and device for converting natural language question into structured query statement | |
CN116737922A (en) | Tourist online comment fine granularity emotion analysis method and system | |
CN111832303A (en) | Named entity identification method and device | |
CN113239698A (en) | Information extraction method, device, equipment and medium based on RPA and AI | |
CN112036179A (en) | Electric power plan information extraction method based on text classification and semantic framework | |
CN112184133A (en) | Artificial intelligence-based government office system preset approval and division method | |
CN108762523A (en) | Output characters through input method prediction technique based on capsule networks | |
Williams | Neural lexicons for slot tagging in spoken language understanding |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |