CN114638597A

CN114638597A - Intelligent government affair handling application system, method, terminal and medium

Info

Publication number: CN114638597A
Application number: CN202210536245.0A
Authority: CN
Inventors: 蒋红军; 黄海霞; 薛凤; 徐志伟; 杨帆; 王康
Original assignee: Shanghai Pudong New District Administrative Service Center (shanghai Pudong New District Civic Center)
Current assignee: Shanghai Pudong New District Administrative Service Center (shanghai Pudong New District Civic Center)
Priority date: 2022-05-18
Filing date: 2022-05-18
Publication date: 2022-06-17

Abstract

The invention provides an intelligent government affair handling application system and an application method, wherein the system comprises: the material making module is used for making or modifying the declaration material required by item handling and generating declaration material image data; the material uploading module is used for uploading the declared material image data; the examination and approval checking module is used for examining and approving the uploaded image data of the declared material to obtain declared material data with examination and approval results and generating the problems needing to be corrected in the examination and approval results; the problem list module is used for generating and outputting the problem list needing to be corrected; an information storage module; the module is used for storing the declaration material data which completes the examination and approval verification. A corresponding terminal and medium are also provided. The invention supports government affair service by intelligent development technology, changes the traditional pure manual and low-efficiency mode of government affair service, and improves the efficiency of application and approval of administrative affairs.

Description

Intelligent government affair handling application system, method, terminal and medium

Technical Field

The invention relates to the technical field of information, in particular to a government affair intelligent office application system, a method, a terminal and a medium.

Background

With the acceleration of the digital transformation pace, the digital government is entering the development stage of intelligent innovation from networking and platform. However, the current government affair service mode is still in a relatively traditional networking and digitalization stage, the application degree of the new technology is still insufficient, and especially the application of the artificial intelligence technology is less, so that the problems of more running for people, low window approval efficiency and the like are caused.

From the application end, during the transaction process of enterprises and masses, the traditional handwriting mode is still adopted for material filling, or window workers assist in entering declaration information. Therefore, the speed is low, errors are easy to occur, and reporting efficiency is low. When the window is seen from the receiving end, the window staff still carries out material arrangement and information input in a pure manual mode, and the window is time-consuming, labor-consuming and long in time-consuming. In addition, with the reformation of the comprehensive window, window workers not only need to master a large number of examination and approval rules, but also need to check a plurality of examination and approval key points of each material one by one, so that the workload is large, errors are easy to occur, and the administrative examination and approval efficiency is greatly reduced. Taking the change items of the internal capital company as an example, the window staff needs to check the 114 approval points, which is time-consuming and easy to cause rework. Therefore, the current government affair service mode urgently needs to introduce a new intelligent technology to replace a large amount of standardized and repetitive work so as to improve the window approval efficiency and improve the business experience of enterprises and masses.

At present, no explanation or report of the similar technology of the invention is found, and similar data at home and abroad are not collected.

Disclosure of Invention

Aiming at the defects in the prior art, the invention provides an intelligent government affair handling application system, method, terminal and medium.

According to one aspect of the invention, a government affairs intelligent office application system is provided, which comprises:

the material making module is used for making or modifying the declaration material required by item handling and generating declaration material image data;

the material uploading module is used for uploading the declared material image data;

the examination and approval checking module is used for examining and approving the uploaded declared material image data to obtain declared material data with examination and approval results and generating problems needing to be corrected in the examination and approval results;

the problem list module is used for generating and outputting the problem list needing to be corrected;

an information storage module; the module is used for storing the declared material data which completes the examination and approval verification.

Preferably, the material making module comprises:

the field acquisition submodule is used for acquiring the field information of the item information, the enterprise basic information, the declaration information and the sponsor information required by item transaction and performing object attribute definition on the acquired field information;

the attribute association submodule presets a declaration material template, and business fields of the declaration material template are endowed with different object attributes and are associated with the object attributes of the field information;

and the page rendering sub-module is used for rendering the field information into the declaration material template by adopting a template engine based on the result of the object attribute association, and generating declaration material image data required by transaction.

Preferably, in the field obtaining sub-module, the field information of the event information is obtained through a specific situation of an event selected by a user;

the field information of the enterprise basic information is acquired through an access database interface;

the field information of the declaration information is obtained through the relevant content of the items input by the user;

and the field information of the sponsor information is obtained by detecting the sponsor certificate by adopting a certificate detection model and extracting the information of the qualified sponsor certificate by adopting a character recognition model.

Optionally, the detecting the certificate of the sponsor by using the certificate detection model includes:

constructing a certificate detection preliminary model based on an HOG + SVM detection algorithm;

training the certificate detection preliminary model to obtain a certificate detection model;

and detecting the graphic characteristics of the certificate of the sponsor by adopting the certificate detection model, judging whether the certificate is a valid certificate or not, and finishing the detection of the certificate of the sponsor.

Optionally, the training of the license detection preliminary model to obtain a license detection model includes:

testing the test sample set on the certificate detection preliminary model, and taking a negative sample with low confidence level in the testing process as a negative sample difficult case;

and performing data enhancement on the negative sample difficult case, training the certificate detection preliminary model by using a training sample, and obtaining a certificate detection model.

Optionally, the graphical features comprise: the size and aspect ratio of the document.

Optionally, the extracting information of the qualified certificate of the sponsor by using the character recognition model includes:

constructing an OCR character recognition preliminary model based on a DBNet + CRNN + CTC algorithm;

training the OCR character recognition preliminary model to obtain an OCR character recognition model;

positioning and marking field image information in the certificate of the sponsor by adopting the OCR character recognition model; and performing text recognition on the field image information of the marked position, extracting corresponding characters and obtaining corresponding field information.

Optionally, the material uploading module further includes signing a part of the declaration material image data that needs to be signed, and uploading the part of the declaration material image data after the signing processing; wherein the signature processing comprises:

the system is accessed to a CA interface, and a user uses Ukey and inputs a key to carry out company signature;

the user inputs the personal name, the ID card number and the mobile phone number, and carries out personal signature after personal authorization.

Uploading the signed declaration material image data.

Preferably, the approval verification module includes:

a classification submodule for classifying a plurality of declaration material image data belonging to the same declaration material;

the extraction submodule is used for performing targeted extraction on key field information and position information of the classified reporting material according to a preset extraction rule to obtain preprocessed reporting material data;

and the checking sub-module is used for examining and approving the preprocessed declared material data to obtain an examination and approval preliminary result, and checking the examination and approval preliminary result to obtain a final examination and approval result.

Preferably, the classification submodule includes:

an image correction unit for performing direction and tilt correction on the declaration material image data;

the OCR recognition unit is used for recognizing the position information of characters in the corrected reporting material image data and extracting all character field information on the reporting material image data based on the position information;

and the rule matching unit is used for matching all the character field information on the reporting material image data with the classification rules through a regular expression according to the preset classification rules, and classifying the reporting material image data with consistent primary keywords and excluding the set keywords into the same reporting material.

Optionally, in the extracting sub-module, performing targeted extraction of key field information and location information on the classified declaration material, including:

and according to the classified declaration material, performing targeted extraction on the key field information and the position information in the declaration material by adopting a regular expression or NLP (non-line-of-sight) recognition algorithm.

Preferably, the check submodule includes:

a document element detection unit that judges a required content on the extracted position information by using a document element detection model;

the examination and approval rule checking unit is used for comparing the key field information extracted from the declared material data with an examination and approval rule base to obtain a checking result;

the examination and approval result output unit divides each piece of declared material data into examination and approval results which are passed, to be confirmed and need to be corrected according to the verification result;

and the examination and approval correcting unit is used for manually rechecking the examination and approval result obtained by the examination and approval result output unit and correcting the examination and approval result.

Optionally, the determining, by using the document element detection model, the required content on the extracted position information includes:

constructing a document element detection initial model based on a Yolov5 algorithm;

constructing a training data set by adopting a data enhancement mode aiming at the category, and training the document element detection initial model to obtain a document element detection model;

and judging whether the position contains signature, date, check, certificate and signature required by examination and approval by utilizing the document element detection model according to the extracted position information, and finishing corresponding judgment.

Optionally, the data enhancement manner for the category includes:

and calculating the quantity of various elements in the calibrated data set, acquiring a target slice aiming at the elements with the quantity distribution lower than a set threshold value a, combining the target slice with background pictures of other elements and reconstructing a label, and ensuring that the maximum overlapping rate between different targets does not exceed a set threshold value b, thereby enriching the data set.

Optionally, the acquiring a target slice and combining and tag reconstructing with a background picture of other elements includes:

determining the element types needing to be enhanced according to the quantity distribution condition of various elements in the data set;

selecting a target picture containing corresponding element categories, and acquiring a target slice for enhancement according to the calibrated target position and category;

randomly selecting background pictures of other elements as a combined picture background, inserting a single or a plurality of target slices to obtain a combined picture, preprocessing the combined picture, and generating coordinates of target slice insertion in the combined picture background;

IoU calculation is carried out on each inserted coordinate and the original calibrated target position in the background of the combined picture, when the obtained intersection and overlapping threshold value exceeds a set threshold value, the inserted coordinates of the target slices are regenerated until the inserted coordinates of all the target slices meet the requirements;

and generating a new picture and a label thereof according to the target slice and the generated inserted coordinates, and adding the new picture and the label into the data set.

Wherein the tag comprises: signature, date, check, certificate, element category and coordinate area of the signature.

Optionally, the approval rule base includes:

the material completeness inspection rule is used for inspecting whether the submitted material meets the requirement of transaction, including material category and material page number;

the text segment attribute consistency checking rule is used for checking the basic attributes in the submitted materials;

and (3) cross-material consistency checking rules for checking consistency of the same field among different materials.

Optionally, the correcting the approval result includes:

modifying the classification of declared materials with classification errors;

judging the approval result to be confirmed;

and manually marking and/or modifying the approval result which is judged to be wrong or not identified.

According to another aspect of the invention, a government affair intelligent application method is provided, which comprises the following steps:

making or modifying a declaration material required for item transaction, and generating declaration material image data;

uploading the declaration material image data;

examining and approving the uploaded image data of the declaration material to obtain declaration material data with an examination and approval result, and generating a problem needing to be corrected in the examination and approval result;

generating a list of the problems needing to be corrected;

and storing the declared material data which is approved and verified.

According to a third aspect of the present invention, there is provided a terminal comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor when executing the program being operable to execute the system of any one of the above, or to perform the method of any one of the above.

According to a fourth aspect of the invention, there is provided a computer readable storage medium having stored thereon a computer program which, when executed by a processor, is operable to run a system as described in any one of the above, or to perform a method as described above.

Due to the adoption of the technical scheme, compared with the prior art, the invention has the following beneficial effects:

the government affair intelligent office application system, method, terminal and medium provided by the invention support government affair services by an intelligent development technology based on the business process and the examination and approval rule of the government affair services, change the traditional pure manual and low-efficiency government affair service mode, and improve the efficiency of application and examination and approval of administrative affairs.

According to the government affair intelligent handling application system, the method, the terminal and the medium, the intelligent manufacturing material is used for replacing the manual filling declaration material, the possibility of error filling is reduced, the problems of low speed and high possibility of error caused by the manual filling material are solved, and the handling efficiency of enterprises and the masses is improved.

According to the government affair intelligent office application system, the method, the terminal and the medium, the traditional manual examination and approval mode is innovated to be the machine examiner, the workload of window workers is reduced, the problems of time and labor waste and low efficiency caused by manual examination and approval are solved, and the examination and approval efficiency of the window workers is improved.

Drawings

Other features, objects and advantages of the invention will become more apparent upon reading of the detailed description of non-limiting embodiments with reference to the following drawings:

fig. 1 is a schematic structural diagram of a government affairs intelligent office application system according to an embodiment of the present invention.

Fig. 2 is a schematic diagram of a system architecture of a government affairs intelligent office application system according to a preferred embodiment of the present invention.

FIG. 3 is a schematic diagram of an operation interface of a material manufacturing module according to an embodiment of the present invention; wherein (a) to (e) are respectively schematic diagrams of operation interfaces made of different declaration materials.

FIG. 4 is a schematic diagram of an operation interface of a material upload module in an embodiment of the present invention; wherein (a) - (c) are respectively functional interface schematic diagrams in different stages.

FIG. 5 is a schematic diagram of an operation interface of an approval verification module according to an embodiment of the present invention.

FIG. 6 is a diagram illustrating an operation interface of a problem list module according to an embodiment of the present invention.

FIG. 7 is a diagram illustrating an interface for completing submission of a transaction piece in an exemplary embodiment of the present invention.

Fig. 8 is a flowchart illustrating a method for applying government affairs intelligence in an embodiment of the present invention.

Detailed Description

The following examples illustrate the invention in detail: the embodiment is implemented on the premise of the technical scheme of the invention, and a detailed implementation mode and a specific operation process are given. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention.

The application aims at providing an application system with intelligent approval function, mainly according to national laws and regulations, policy documents, relevant regulations and the like followed by government administration approval work, by utilizing artificial intelligence technologies such as OCR (optical character recognition), document element detection and the like, carrying out four-side automatic detection on declared materials by constructing an intelligent approval model so as to complete approval work, assisting workers to rapidly complete business handling, and specifically comprising the following steps of:

firstly, the OCR technology is used for identifying and extracting the information of the related declared materials, and the same key fields of different materials, such as business licenses and enterprise addresses in application books, are compared to check whether the related information in the application materials is consistent.

And secondly, judging whether the declared material has signature, date, check, seal and the like according to the requirements by using a document element detection technology, comparing the signed material with a preset approval rule (administrative approval requirement), and automatically detecting whether the approval elements are complete.

Thirdly, detecting whether the declared material has relevant contradiction places by using an intelligent approval model, for example, identifying and extracting relevant material information by using an OCR technology in a business scene with relevant requirements on the enterprise operation area, comparing the relevant material information with a preset approval rule (relevant legal and regulatory requirements), and judging whether the application material violates the corresponding approval requirements.

And fourthly, identifying and extracting field information of the material based on an OCR technology, such as the business certificate date of the enterprise, and comparing the field information with a preset approval rule to examine whether the information in the application material is effective or not.

The following describes specific embodiments and details of the intelligent government application system of the present invention.

As shown in fig. 1, the government affairs intelligent office application system provided by an embodiment of the present invention provided by this embodiment may include the following modules:

the material uploading module is used for uploading declared material image data;

the examination and approval checking module is used for examining and approving the uploaded image data of the declared material to obtain the declared material data with an examination and approval result and generating the problem needing to be corrected in the examination and approval result;

the problem list module is used for generating and outputting a problem list needing to be corrected;

an information storage module; the module is used for storing the declaration material data which completes the examination and approval verification.

In this embodiment, the generated list of questions that need to be corrected may be presented to the user for the user to modify the declared material. According to the embodiment, government affair service can be supported by an intelligent development technology, the traditional pure manual and low-efficiency government affair service mode is changed, the efficiency of application and approval of administrative affairs is improved, and the automation of government affair office approval is further realized.

As a preferred embodiment, in order to meet the requirement of material fabrication, the material fabrication module comprises: the system comprises a field acquisition sub-module, an attribute association sub-module and a page rendering sub-module, wherein: the field acquisition submodule is used for acquiring the field information of the item information, the enterprise basic information, the declaration information and the sponsor information required by item handling, and performing object attribute definition on the acquired field information; the attribute association submodule presets a declaration material template, and business fields of the declaration material template are endowed with different object attributes and are associated with the object attributes of field information; and the page rendering submodule renders the field information into the declaration material template by adopting a template engine based on the result of the object attribute association to generate declaration material image data required by transaction. By the aid of the device, the hand filling declaration material can be replaced by the intelligent manufacturing material, the possibility of error filling is reduced, and the problems of low speed and high error probability caused by the hand filling material are solved.

As a preferred embodiment, in the field obtaining submodule, the field information of the event information is obtained through the specific situation of the event selected by the user; the field information of the enterprise basic information is acquired by accessing a database interface; acquiring field information of the declaration information through related contents of items input by a user; the field information of the sponsor information is obtained by adopting a license detection model to detect the sponsor certificate and adopting a character recognition model to extract the information of the qualified sponsor certificate.

As a preferred embodiment, the method for detecting the certificate of the sponsor by adopting the certificate detection model comprises the following steps: constructing a certificate detection preliminary model based on an HOG + SVM detection algorithm; training the certificate detection preliminary model to obtain a certificate detection model; and detecting the graphic characteristics of the certificate of the dealer by adopting a certificate detection model, judging whether the certificate is a valid certificate, and finishing the detection of the certificate of the dealer. Specifically, in some embodiments, the graphical features include: the size and aspect ratio of the document.

Further, in some embodiments, training the preliminary license detection model to obtain the license detection model includes: testing the test sample set on a certificate detection preliminary model, and taking a negative sample with low confidence level in the testing process as a negative sample difficult case; and performing data enhancement on the negative sample difficultly, and training the preliminary certificate detection model by training the sample to obtain the certificate detection model. So that a more accurate model can be obtained.

In this embodiment, as a preferred embodiment, the information extraction of the qualified certificate of the sponsor by using the character recognition model includes: constructing an OCR character recognition preliminary model based on a DBNet + CRNN + CTC algorithm; training the OCR character recognition preliminary model to obtain an OCR character recognition model; positioning and marking field image information in the certificate of a manager by adopting an OCR character recognition model; and performing text recognition on the field image information of the marked position, extracting corresponding characters and obtaining corresponding field information. Therefore, automatic extraction of information is realized, and the efficiency of window workers and the accuracy of the information are improved.

In this embodiment, as a preferred embodiment, the material uploading module further includes signing a part of the declaration material image data that needs to be signed, and uploading the part; wherein, the signature processing comprises: the system is accessed to a CA interface, and a user uses Ukey and inputs a key to carry out company signature; the user inputs the personal name, the ID card number and the mobile phone number, and carries out personal signature after personal authorization. Uploading the signed declaration material image data. In practice, the declaration material part needs to be signed, the declaration material part does not need to be signed, and for the part needing to be signed, a corresponding electronic signature interface can be provided in the system, and the signature is uploaded after the signature so as to meet the requirement of government affair approval materials, complete affairs are handled at one time, the phenomenon that a user runs for many times is avoided, and the user experience is improved.

The approval checking module is an important module of the invention, innovates the approval mode from the traditional manual approval to an auditor, reduces the workload of window workers, and solves the problems of time and labor waste and low efficiency caused by manual approval. In order to implement automation of the approval verification, as a preferred embodiment, the approval verification module may include: the classification submodule, the extraction submodule and the check submodule, wherein: the classification submodule classifies a plurality of declared material image data belonging to the same declared material; the extraction submodule performs targeted extraction on key field information and position information of the classified reporting material according to a preset extraction rule to obtain preprocessed reporting material data; and the checking submodule examines and approves the preprocessed declaration material data to obtain an examination and approval preliminary result, and checks the examination and approval preliminary result to obtain a final examination and approval result.

In this embodiment, as a preferred embodiment, the classification sub-module may include: the system comprises an image correction unit, an OCR recognition unit and a rule matching unit, wherein the image correction unit is used for correcting the direction and the inclination of the image data of the declaration material; the OCR recognition unit carries out position information recognition of characters on the corrected image data of the declaration material, and extracts all character field information on the image data of the declaration material based on the position information; the rule matching unit matches all character field information on the reporting material image data with the classification rules through the regular expression according to the preset classification rules, and classifies the reporting material image data with the consistent main keywords and the set keywords excluded as the same reporting material. The automatic classification and matching of the materials to be examined and approved can be realized through the classification submodule, and the workload of window workers is greatly reduced.

In this embodiment, as a preferred embodiment, in the extraction sub-module, the targeted extraction of the key field information and the location information of the classified declaration material includes: and according to the classified declaration material, pertinently extracting key field information and position information in the declaration material by adopting a regular expression or NLP (non line segment) recognition algorithm. The key field information is set according to various approval items.

In this embodiment, as a preferred embodiment, the check submodule includes:

the examination and approval rule checking unit is used for comparing the key field information extracted from the declared material data with the examination and approval rule base to obtain a checking result;

In this embodiment, as a preferred embodiment, the determining the required content on the extracted position information by using the document element detection model includes:

constructing a training data set by adopting a data enhancement mode aiming at the category, and training a document element detection initial model to obtain a document element detection model;

In this embodiment, as a preferred embodiment, the data enhancement mode for the category includes:

In this embodiment, as a preferred embodiment, acquiring a target slice and combining and tag reconstructing with a background picture of other elements includes:

IoU calculation is carried out on each inserted coordinate and the original calibrated target position in the background of the combined picture, when the obtained intersection and the overlapping threshold exceed the set threshold, the coordinates inserted into the target slices are regenerated until the coordinates inserted into all the target slices meet the requirements;

In this embodiment, the above tag includes: any one or more of signature, date, check, certificate, and element category and coordinate region of the signature.

In this embodiment, as a preferred embodiment, the approval rule base includes the following contents:

(1) the material completeness inspection rule is used for inspecting whether the submitted material meets the requirement of transaction, and comprises a material type and a material page number;

(2) the document attribute consistency check rule is used for checking the basic attribute in the submitted material;

(3) and (3) cross-material consistency checking rules for checking consistency of the same field among different materials.

In this embodiment, as a preferred embodiment, the correcting the approval result includes: modifying the classification of declared materials with classification errors; judging the approval result to be confirmed; and manually marking and/or modifying the approval result which is judged to be wrong or not identified. Thereby further ensuring the accuracy of the approval result.

Fig. 2 is a schematic system architecture diagram of a government affairs intelligent office application system according to a preferred embodiment of the present invention.

As shown in fig. 2, the government affairs intelligent office application system of the preferred embodiment is an application system that uses technologies such as OCR (character recognition), license detection and extraction, document element detection, NLP (natural language processing) and the like to realize functions such as intelligent material making and intelligent pre-auditing.

The following describes each functional module of the government affairs intelligent office application system provided in the preferred embodiment in detail.

Function module

The government affair intelligent office application system provided by the preferred embodiment comprises 5 functional modules in total, and comprises:

1. the material making module is used for making declaration materials required by transaction, and is mainly realized by three functional modules, namely a field acquisition sub-module, an attribute association sub-module and a page rendering sub-module. The operation interface of the material manufacturing module can be as shown in (a) - (e) of fig. 3.

A field acquisition submodule, which is mainly used for acquiring item information, enterprise basic information, declaration information and manager information required by item handling:

A. the item information mainly comprises information such as a commission office, item types, transaction situations and the like; the system obtains the information through the selection of the event situation by the user.

B. The basic information of the enterprise mainly comprises field information such as an enterprise name, an operation place address, a region of the enterprise and the like; the system acquires the information by accessing a database interface such as a big data center.

C. The declaration information is mainly distinguished according to different matters; the system acquires the declaration information after the user inputs the information.

D. The information of the operator mainly comprises information such as name, identification card number, address and the like; the system adopts the license detection and extraction and OCR technology to obtain the information:

a. the identification card detection and extraction technology is mainly used for judging whether an uploaded identification card picture is an identification card or not according to the size, the length-width ratio and other graphic characteristics of the identification card;

b. OCR will locate the position of the key field information in the digital image and perform position labeling. And identifying the specific content of the text according to the marked position, converting a string of characters in the image into corresponding characters, and acquiring the information of the name and the license number.

And the attribute association submodule presets a declaration material template corresponding to the event situation in the system based on the rule of the material application, and endows different object attributes to the service fields needing to be filled in the material template. After the fields are acquired, the system defines the object attribute of each acquired field and associates the object attribute with the object attribute of the service field in the material template.

And thirdly, the page rendering sub-module renders the key business fields to the template page by adopting Handlebars (template engine) based on the result of the object attribute association, and generates the material for transaction.

2. And the material uploading module is used for uploading materials required by transaction. And based on the item transaction rule, signature processing needs to be carried out on part of the materials needing to be signed before the materials are uploaded. The module provides an electronic signature mode, namely a system is accessed to a CA interface, and a user is allowed to use Ukey to input a key and then carry out company signature; the user is allowed to input personal name, ID card number and mobile phone number, and personal signature is carried out after personal authorization. The operation interface of the material uploading module can be as shown in (a) - (c) of fig. 4.

3. The examination and approval checking module is used for examining and approving the uploaded declared materials and is mainly realized by three functional modules, namely a classification submodule, an extraction submodule and a checking submodule. And the operation interface of the approval and verification module can be shown in FIG. 5.

The classification submodule is mainly realized based on an algorithm of image direction correction and OCR recognition:

A. and the image correction unit is used for carrying out direction correction on the material picture, including direction correction and inclination correction.

B. The OCR recognition unit is used for recognizing the material picture by OCR and acquiring the position information of the characters; and identifying and extracting all field information on the material picture based on the position information.

C. And the rule matching unit is used for configuring a classification rule by the system based on the service logic, matching the material field with the classification rule through a regular expression, and classifying the materials with consistent main keywords and excluding partial keywords into the same material.

And secondly, extracting a submodule, which is mainly used for extracting key field information and position information thereof:

A. for most materials, the regular expression performs targeted extraction on the OCR-identified fields according to the extraction rules obtained by the business logic.

B. And for materials such as the chapters of companies, the chapters of the companies and the like, performing targeted extraction on the fields identified by the OCR by adopting an NLP algorithm according to the extraction rules obtained by the business investigation.

The verification submodule judges the approval result of the declared material mainly through document element detection and approval rule verification:

A. the document element detection unit judges whether a certain position on a piece of material has a signature, a date, a check, an identity card and a seal by a document element detection algorithm.

B. And the examination and approval rule checking unit is used for comparing the extracted field information and the document detection result with the examination and approval rule base. Wherein, the examination and approval rule base mainly defines the following rules:

a. and (4) material completeness inspection, namely inspecting whether the submitted material meets the requirement of transaction, including material category and material page number.

b. And (5) checking consistency of the attributes of the text segments, namely checking basic attributes such as array length, date, seal and the like.

c. Cross-material consistency checks, i.e., checks for consistency of the same field across different materials.

C. And the examination and approval result output unit divides each material into three examination and approval results, namely passed, to be confirmed and required to be corrected, according to the consistency check condition:

a. passed means that any key field in the material is completely consistent with the verification of the approval rule base.

b. The correction is needed, which means that any key field in the material is not completely consistent with the verification of the approval rule base, namely, partial difference exists. For the fields inconsistent with the examination and approval rule base, the front end annotates the material in the form of text based on the examination and approval information returned by the back end; meanwhile, the system draws a curve according to the position information of the error field for marking.

c. To be confirmed, there are two cases: one is that the rule type is set as 'temporary support' because the rule engine temporarily does not support the verification of a certain rule; secondly, unknown errors occur in the classification and extraction processes.

D. And the approval correcting unit is used for correcting the approval result after the window worker manually rechecks the approval result: firstly, modifying the classification of the materials with the wrong classification; judging the examination and approval key points to be determined; thirdly, manually marking and/or modifying the approval points which are judged to be wrong or not identified by the system.

4. And the problem list module is mainly used for outputting a problem list needing to be corrected (examination and approval verification). The question list comprises a serial number, a material name and contents to be corrected. The operation interface of the question list module can be shown in fig. 6.

5. The information storage module is mainly used for storing the transaction information, including transaction information, enterprise information, declaration information, sponsor information, examination and approval results and the like.

And the submission and handling of the parts are completed through the functional modules. The interface for submitting the office is completed as shown in FIG. 7.

(II) Key algorithm model

1. The certificate detection model comprises: for certificate detection and extraction

The basic principle is as follows: the license detection model is obtained mainly based on the open-source HOG + SVM detection algorithm.

The license detection model has the advantages that: in an actual scene, the model for license detection and extraction needs to be tested and optimized to match the actual application scene. In the mode optimization, a test sample set is tested on a license identification model, and negative samples with low confidence in the test process are used as negative sample difficult cases; through the small-angle rotation, adjust luminance, saturation, contrast parameter, modes such as increase gaussian noise, salt and pepper noise carry out data enhancement to the difficult case of negative sample, retrain the model of license discernment again.

Thirdly, the license detection model effect: through the optimization of the model, the accuracy of the identity card detection reaches more than 90%.

2. Document element detection model

The basic principle is as follows: the document element detection model is obtained mainly based on an open source Yolov5 algorithm.

The document element detection model has the advantages that: in an actual scene, various elements in the document are not uniformly distributed. Therefore, in order to solve the influence of element class imbalance on the training model, a data enhancement mode aiming at the classes is adopted: the quantity of various elements in the calibrated data set is calculated, and aiming at the license elements with less distribution, the marking area and other picture backgrounds of the business scene are extracted for combination and label reconstruction, so that the maximum overlapping rate between different targets is not more than 10%, the effect of enriching the data set is realized, the element category distribution balance is achieved, and the feature extraction and generalization capability of the model to the document elements is improved.

Thirdly, detecting the model effect of the document elements: by enriching the data set, the detection mode of the document elements is continuously optimized, and the accuracy exceeds 98.6%.

3. OCR character recognition model

The basic principle is as follows: the OCR character recognition model is obtained mainly based on an open-source DBNet + CRNN + CTC algorithm.

OCR character recognition model has the advantages that: generally, OCR character recognition can only recognize one line or one segment of characters and coordinates thereof, and the model can accurately recognize single characters and coordinates thereof after multiple tests and optimizations.

(iii) model effect:

A. through model training and optimization, the OCR character recognition accuracy is over 93%.

B. Because OCR can accurately identify a single character and the coordinate thereof, in the examination and approval verification module, the front end can identify the character in which the problem point is located and the coordinate thereof.

Fig. 8 is a flowchart illustrating a method for applying government affairs intelligence according to an embodiment of the present invention.

As shown in fig. 8, the method for applying government affairs intelligence provided by this embodiment may include the following steps:

s100, making or modifying a declaration material required by item handling, and generating declaration material image data;

s200, uploading the declaration material image data;

s300, examining and approving the uploaded image data of the declaration material to obtain declaration material data with an examination and approval result, and generating a problem needing to be corrected in the examination and approval result;

s400, generating a list of problems needing to be corrected;

and S500, storing the declared material data which is approved and verified.

It should be noted that, the steps in the method provided by the present invention can be implemented by using corresponding modules, units, and the like in the system, and those skilled in the art can implement the step flow of the method by referring to the technical scheme of the system, that is, the embodiment in the system can be understood as a preferred example of the implementation method, and details are not described herein.

An embodiment of the present invention provides a terminal, including a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor is configured to execute the system in any one of the above embodiments or perform the method in any one of the above embodiments when executing the program.

An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, is operable to run the system of any one of the above embodiments or to perform the method of any one of the above embodiments.

In the above two embodiments, optionally, the memory is used for storing a program; a Memory, which may include a volatile Memory (RAM), such as a Random Access Memory (SRAM), a Double Data Rate Synchronous Dynamic Random Access Memory (DDR SDRAM), and the like; the memory may also comprise a non-volatile memory, such as a flash memory. The memories are used to store computer programs (e.g., applications, functional modules, etc. that implement the above-described methods), computer instructions, etc., which may be stored in partition in the memory or memories. And the computer programs, computer instructions, data, etc. described above may be invoked by a processor.

The computer programs, computer instructions, etc. described above may be stored in one or more memories in a partitioned manner. And the computer programs, computer instructions, data, etc. described above may be invoked by a processor.

A processor for executing the computer program stored in the memory to implement the steps of the method according to the above embodiments. Reference may be made in particular to the description relating to the preceding method embodiment.

The processor and the memory may be separate structures or may be an integrated structure integrated together. When the processor and the memory are separate structures, the memory and the processor may be coupled by a bus.

A specific application example is provided below, and the technical solution provided by the above embodiment of the present invention is further explained as follows.

The intelligent government affair handling application system provided by the specific application example mainly comprises the following modules:

1. and the material making module is used for making declaration materials required by item handling. Firstly, a user inputs transaction information, enterprise information, declaration information and operator information in sequence according to transaction. Secondly, the system defines and processes the object attribute of the received field. And finally, automatically manufacturing the material after page rendering by associating the service field, the item handling field and the object attribute of the preset template. Finally, the material making module replaces manual filling with intelligent filling, and the possibility of material filling errors is reduced. Taking the enterprise change operation range as an example, in the past, enterprise clerks often need to fill in 3 materials and 51 elements; through application material preparation module, enterprise's personnel of handling affairs no longer need fill out any material of reporting and just can accomplish the application, have shortened the length of time of reporting, have promoted its experience of handling affairs.

2. And the material uploading module is used for uploading materials required by transaction. Wherein, based on the item transaction rule, part of the material needs to be signed before the material is uploaded. The module provides an electronic signature mode, namely a system is accessed to a CA interface, and a user is allowed to use a Ukey input key to carry out company signature; the user is allowed to input personal name, ID card number and mobile phone number, and then personal signature processing is carried out after personal authorization.

3. And the examination and approval checking module is used for examining and approving the declared materials. Firstly, automatically classifying materials to be approved by a system after OCR recognition; secondly, based on an extraction rule, the system extracts key fields of the material identified by the OCR; at the same time, the system will detect the elements (such as seal, date, signature) of each material; and finally, based on the extracted key fields and element detection results, comprehensively approving the aspects of completeness, document attribute consistency, cross-material consistency and the like of declared materials by the system according to an approval rule base, and finally obtaining approval results. On the whole, the examination and approval checking module replaces manual examination and approval with intelligent examination and approval, so that the examination and approval time length is shortened, and the examination and approval efficiency is improved. Taking the engineering muck approval and approval items as an example, the approval and approval module can finish the verification of 14 approval elements within 2 minutes, so that the time is saved by over 70 percent, and the approval time of window workers is greatly shortened.

4. And the problem list module is mainly used for outputting a problem list needing to be corrected (examination and approval verification). The question list comprises a serial number, a material name and contents to be corrected.

5. The information storage module is mainly used for storing item transaction information, including transaction item information, enterprise information, declaration information, manager information, examination and approval results and the like.

Based on the functional modules, the window staff assist the enterprise staff in handling the internal resource change (change + record). In the specific application example, the interfaces related to the government affair intelligent office application system are shown in FIGS. 3-7; the steps relating to the application example are mainly as follows:

step0, through oral communication of enterprise transacting personnel, the window staff enters transaction item information, enterprise basic information and change information on the system interface.

Step1, the window worker enters the dealer identification card information using the scanner.

Step2, the system automatically generates four materials, namely, a "company registration (filing) application", a "resolution and decision on the revision of the company's chapters", a "notice of prior approval of the company's name", a "application of prior approval of the company's name", and a "revised company's chapters or revision of the company's chapters".

Step3, the business clerk submits a proof of use of the post-change residence, a copy of the relevant approval document or license document to the window clerk.

Step4, the window staff upload all the materials of Step2 and Step 3.

Step5, the system performs one-by-one approval and verification on 114 approval points of the submitted material to generate three approval results.

Step6, the window staff checks the approval results presented by the system.

Step 7: if the material is in question, the window worker will print a list of the questions and hand them to the business clerk.

Step 8: if the material is not in a problem, the window staff can click 'propose and submit' on the system, and finally the process of handling the work is finished.

Step 9: the information of the office procedure will be automatically saved to the history.

As can be seen from the specific application example, the technical solution provided by the above embodiment of the present invention will effectively improve the efficiency of administrative affair application and approval: for the claimant, the material making module replaces the claimant to make 4 parts of materials such as company registration (filing) application book and the like, so that the declaration time is shortened, and the declaration efficiency is improved; for window workers, the examination and approval checking module replaces manual work to check 114 examination and approval points such as ' whether the shareholder at the resolution of the shareholder meeting ' has signed ', so that the workload is greatly reduced, and the examination and approval efficiency is improved by over 80%.

According to the above specific application example, in the technical scheme provided by the above embodiment of the invention, the material manufacturing module uses the intelligent manufacturing material to replace the manual filling declaration material, so that the possibility of failure in filling declaration is reduced, the problem of long consumed time caused by manual filling material is solved, and the transaction efficiency of enterprises and masses is improved. Taking the enterprise change operation range as an example, enterprise office staff needs to fill 3 declaration materials in the past; by applying the material manufacturing module, enterprise personnel can finish application without filling any declaration material, thereby shortening declaration time and improving the handling experience. The examination and approval checking module replaces manual examination and approval with intelligent examination and approval, so that the workload of window workers is reduced, and the administrative examination and approval efficiency is improved. Taking the approval items of the engineering muck quasi-transport certificate as an example, the approval and verification module can complete the verification of 14 approval elements within 2 minutes, so that the time is saved by more than 70%, and the efficiency is improved by more than 80%.

The above embodiments of the present invention are not exhaustive of the techniques known in the art.

The foregoing description of specific embodiments of the present invention has been presented. It is to be understood that the present invention is not limited to the specific embodiments described above, and that various changes or modifications may be made by one skilled in the art within the scope of the appended claims without departing from the spirit of the invention.

Claims

1. An intelligent government affair handling application system, comprising:

the examination and approval checking module is used for examining and approving the uploaded image data of the declared material to obtain declared material data with examination and approval results and generating the problems needing to be corrected in the examination and approval results;

an information storage module; the module is used for storing declaration material data for completing examination and approval verification;

the approval verification module comprises:

the extraction submodule is used for performing targeted extraction on key field information and position information of the classified declared materials according to a preset extraction rule to obtain preprocessed declared material data;

the checking sub-module is used for examining and approving the preprocessed declared material data to obtain an examination and approval preliminary result, and checking the examination and approval preliminary result to obtain a final examination and approval result;

the classification submodule includes:

the rule matching unit is used for matching all the character field information on the reporting material image data with the classification rules through a regular expression according to the preset classification rules, and classifying the reporting material image data with consistent primary keywords and excluding the set keywords into the same reporting material;

the verification submodule comprises a document element detection unit which adopts a document element detection model to judge the required content on the extracted position information; the judging the required content on the extracted position information by adopting the document element detection model comprises the following steps:

judging whether the position contains signature, date, check, certificate and signature required by examination and approval by using the document element detection model according to the extracted position information, and finishing corresponding judgment;

the data enhancement mode aiming at the category comprises the following steps:

calculating the quantity of various elements in the calibrated data set, acquiring a target slice aiming at the elements with the quantity distribution lower than a set threshold value a, combining the target slice with background pictures of other elements and reconstructing labels, and ensuring that the maximum overlapping rate between different targets does not exceed a set threshold value b, thereby enriching the data set;

the acquiring of the target slice and the combination and tag reconstruction with the background picture of other elements includes:

2. The government affairs intelligent office application system according to claim 1, wherein the material making module comprises:

3. The government affairs intelligent office application system according to claim 2, wherein in the field obtaining submodule, the field information of the event information is obtained through a specific situation of an event selected by a user;

4. The intelligent government office application system according to claim 3, wherein the certificate detection model is used for detecting certificates of dealers, and comprises:

and detecting the graphic characteristics of the certificate of the manager by adopting the certificate detection model, judging whether the certificate is a valid certificate, and finishing the detection of the certificate of the manager.

5. The government affairs intelligent office application system according to claim 4, wherein the training of the license detection preliminary model to obtain a license detection model comprises:

6. The government affairs intelligent office application system of claim 4, wherein the graphical features include: the size and aspect ratio of the document.

7. The intelligent government affair office application system according to claim 3, wherein the information extraction of qualified certificate of the sponsor by the character recognition model comprises:

8. The intelligent government affair handling application system according to claim 2, wherein the material uploading module further comprises signing and uploading a part of the declaration material image data needing signing; wherein the signature processing comprises:

the user inputs personal name, ID card number and mobile phone number, and carries out personal signature after personal authorization;

uploading the signed declaration material image data.

9. The system according to claim 1, wherein the extraction sub-module performs targeted extraction of key field information and location information of the classified declaration material, and includes:

and according to the classified declaration material, pertinently extracting key field information and position information in the declaration material by adopting a regular expression or NLP (non line segment) recognition algorithm.

10. The system according to claim 1, wherein the extraction sub-module performs targeted extraction of key field information and location information of the classified declaration material, and includes:

11. The government affairs intelligent office application system according to claim 10, wherein the approval rule base includes:

the document attribute consistency check rule is used for checking the basic attribute in the submitted material;

12. The government affairs intelligent office application system according to claim 10, wherein the correcting the approval result comprises:

modifying the classification of the declared material with the classification error;

judging the approval result to be confirmed;

13. An intelligent government application method, which is implemented by using the intelligent government application system according to any one of claims 1 to 12, and specifically comprises the following steps:

uploading the declaration material image data;

generating a list of the problems needing to be corrected;

and storing the declared material data which is approved and verified.

14. A terminal comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the program when executed by the processor is operable to operate the system of any one of claims 1 to 12 or to perform the method of claim 13.

15. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, is adapted to carry out the system of any one of claims 1-12 or to carry out the method of claim 13.