Design method of intelligent document indication copying and clearing system based on artificial intelligence

Abstract

The invention relates to a design method of an intelligent document indication copying and clearing system based on artificial intelligence, and belongs to the field of computer software. According to the handwritten endorsement content recognition method, a handwritten recognition algorithm model is established for handwritten endorsement content recognition, firstly, the handwritten endorsement area image is obtained, input is provided for handwritten content recognition, the handwritten endorsement content is recognized and extracted, manual extraction and input work is avoided, and the difficulty and measurement of manual recognition work are reduced. In order to improve the accuracy of content finally obtained by handwritten instructions, a content error correction link is added, a content error correction algorithm model is established, error detection is carried out from the two aspects of word granularity and word granularity through word segmentation, an optimal correction word is obtained through a language model, error correction is completed, and the final accuracy of handwritten content recognition is greatly improved. And through a template engine technology, the document is automatically generated according to the constructed copying and clearing template, so that the work of manually sorting and inputting the document is omitted, and the document circulation speed is greatly improved.

Description

Technical Field

[0001] The invention belongs to the field of computer software, and in particular relates to a design method for an intelligent copying and clearing system for official document approval based on artificial intelligence. Background Art

[0002] With the rapid development of information technology, smart office has become an indispensable part of the office environment. The development of artificial intelligence technology has brought smart office into a new stage. By integrating various office software and tools, a higher level of intelligence can be achieved, which further enhances the human-computer collaboration and helps office personnel to better complete various tasks, greatly save human resources and improve work efficiency. In order to meet the requirements of people's work and life, the research and development of smart office related software has become an important task.

[0003] In daily office work, a large number of official documents such as submissions and handling documents are often generated. These documents need to be circulated at multiple levels and require review and approval by leaders at all levels. After each level of approval, the secretary or relevant personnel needs to organize the circled and approved content, extract it into the system and distribute it to the lower level along with the main text. First, the recognition of handwritten signatures and excerpts greatly increases the workload of the excerpting personnel. Secondly, manual excerpting and inputting work often takes a lot of time, reduces the circulation efficiency of documents, and thus affects the progress of related work.

[0004] For the work of copying official documents and instructions, artificial intelligence technology is used to carry out design and transformation at different stages. The first step is to solve the problem of identifying handwritten signature content. Image cutting technology is used to identify the area of ​​handwritten annotations, and then OCR and handwriting recognition are performed on the area to obtain relevant annotation content; the second step is to use Chinese word segmentation and error correction technology to correct the identified instruction content to improve the final accuracy of content recognition; the third step is to solve the problem of content entry and sorting. Important information is retrieved from the database to obtain necessary elements such as the leader's name and time, and a copy format template is constructed. The obtained element information and annotation content are sorted through the template engine, and finally a document file is generated for output, realizing the intelligentization of the entire process of copying official documents and instructions. Summary of the invention

[0005] 1. Technical issues to be resolved

[0006] The technical problem to be solved by the present invention is how to provide a design method of an intelligent copying system for official document instructions based on artificial intelligence to solve the problems of recognition of handwritten signatures and excerpts of contents, which greatly increases the workload of the excerpting personnel. Secondly, manual excerpting and inputting work often consumes a lot of time, reducing the circulation efficiency of documents.

[0007] (II) Technical solution

[0008] In order to solve the above technical problems, the present invention proposes a design method of an intelligent copying system for official document instructions based on artificial intelligence. The intelligent copying system for official document instructions is used for text view analysis and text document generation, wherein the text view analysis includes handwriting recognition and content error correction, and the service is divided into offline and online modes; the text document generation is supported by a template engine, and text content documents are automatically generated according to different templates; the steps used by the system include:

[0009] First, the original file is obtained and converted into an image format. The region is cropped to obtain an image containing only the handwritten signature content. The image is input into the handwriting recognition model to perform end-to-end text recognition on text sequences of indefinite length.

[0010] The recognized text content is input into the text error correction model. The text error correction model includes error detection and error correction. All suspected error positions are obtained through error detection. Then, through error correction, the words in the error position are replaced with similar words. Finally, the sentence perplexity is calculated through the language model. All candidate set results are compared and sorted to obtain the optimal correction word and the corrected text content.

[0011] When splicing data, relevant element data is obtained from the database; in the online mode, after each handwritten approval is completed, the approval content is identified through the service interface call, and the approval time and name of the approver are obtained through the approval system, and stored in the database and bound to the corresponding file for persistence; in the offline mode, the files already on the object storage server are selected for identification and the element data is spliced;

[0012] Before the document is circulated for review, approval and circulation, the text content after error correction is obtained, and a cleared content document is generated according to the cleared format template. The document is circulated together with the original document, and the document content is supplemented and updated each time it is circulated. When viewing the original text, you can choose to view the approval opinions and display the approved content collection in the corresponding text format.

[0013] (III) Beneficial effects

[0014] The present invention proposes a design method for an intelligent copying system for official document instructions based on artificial intelligence. Compared with the prior art, the present invention proposes an intelligent system design for the entire process of copying official document instructions based on artificial intelligence, and selects appropriate algorithms for each processing flow of the task to optimize the framework. For the recognition of handwritten signature content, a handwriting recognition algorithm model is established. First, the image of the handwritten annotation area is obtained to provide input for handwritten content recognition, and the handwritten instruction content is identified and extracted, avoiding manual excerpting and inputting work, and reducing the difficulty and measurement of manual recognition work. In order to improve the accuracy of the final content of handwritten instructions and increase the content correction link, a content correction algorithm model is established. Error detection is performed from both word granularity and word granularity through word segmentation, and the optimal correction word is obtained through the language model to complete the error correction, which greatly improves the final accuracy of handwritten content recognition; through the template engine technology, the document is automatically generated according to the constructed copying template, eliminating the manual sorting and document input work, and greatly improving the file circulation speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is the overall architecture diagram of the present invention;

[0016] Figure 2 A flow chart for use of the system of the present invention;

[0017] Figure 3 is a flow chart of text recognition of the present invention;

[0018] Figure 4 It is a flowchart of error correction of the content of the present invention;

[0019] Figure 5 Generate a flow chart for documentation of the invention. DETAILED DESCRIPTION

[0020] In order to make the purpose, content and advantages of the present invention more clear, the specific implementation methods of the present invention are further described in detail below in conjunction with the drawings and examples.

[0021] The purpose of the present invention is to propose an intelligent document copying system design that can utilize artificial intelligence technology to realize the entire process of copying and clearing official documents in the office field, from identifying the signed content to generating copied documents.

[0022] 1. Overall architecture

[0023] Figure 1 It is the overall architecture diagram of the present invention, such as Figure 1 As shown, the technical architecture of this system can be divided into four layers: presentation layer, transport layer, service layer and data layer.

[0024] The presentation layer is mainly the user-oriented Web end. The system implements Web-based online handwriting content recognition, facilitates content display, export, document generation and other operations, and is built using the React development framework.

[0025] The transport layer is responsible for data transmission between the presentation layer and the service layer. This system mainly uses HTTP network request related technologies for data transmission, while maintaining the RESTful style in interface design;

[0026] The service layer is responsible for the implementation of all business functions of this system. It adopts SpringBoot technology and uses SpringAOP to implement the system's logging, exception handling and permission control functions. MyBatis is used as the persistence layer framework, and TensorFlow and Opencv related technologies are used to implement handwritten text recognition, text error correction, and text generation algorithms. At the same time, the FlaskPythonWeb framework is used to deploy the algorithm online. Finally, ResTemplate is used to realize the communication between the business server and the algorithm server.

[0027] The data layer is responsible for storing data such as instruction records and template information. This system uses MySQL database to store system-related data, and uses OOS object storage service to store various types of documents, copy format templates, generated copy content documents and other files.

[0028] 2. System Usage

[0029] The system is used for text view parsing and text document generation. Text view parsing includes handwriting recognition and content error correction. The service is divided into offline and online modes. The text view file can be uploaded to the storage service and stored using object storage technology to achieve offline recognition. The online recognition interface can also be called to provide real-time related capabilities through interface services. Text document generation is supported by the template engine, which automatically generates text content documents based on different templates.

[0030] The steps used by the system are as follows Figure 2 shown.

[0031] First, obtain the original file, convert it into image format, crop the area, obtain the area image containing only the handwritten signature content, input it into the handwriting recognition model, and perform end-to-end text recognition on text sequences of indefinite length. For details, see the "Handwriting Recognition" section.

[0032] The recognized text content is input into the text correction model. The text correction model includes error detection and error correction. All suspected error positions are obtained through error detection. Then, through error correction, the words in the error position are replaced with similar words. Finally, the sentence perplexity is calculated through the language model. All candidate set results are compared and sorted to obtain the optimal correction word and the corrected text content. For details of the process, see the "Content Correction" section.

[0033] When splicing data, relevant element data is obtained from the database. In the online mode, after each handwritten approval is completed, the approval content is identified through the service interface call, and the approval time and the name of the approver are obtained through the approval system, and stored in the database and bound to the corresponding file for persistence. In the offline mode, files already on the object storage server can be selected for identification and element data can be spliced.

[0034] Before the document is circulated for review and approval, the text content after error correction is obtained, and a copy-clearance content document is generated according to the copy-clearance format template. The document is circulated together with the original document, and the document content is updated each time it is circulated. When viewing the original text, you can choose to view the approval opinions, and display the collection of approval content in the corresponding text format.

[0035] 3. Handwriting recognition

[0036] Handwriting recognition first requires obtaining the original document with handwritten signature content, converting it into an image, obtaining the handwritten signature content area as the input of the handwriting recognition model, and obtaining the handwritten text content after recognition.

[0037] 3.1 Region Identification

[0038] The document text and handwritten signature content are distinguished by different colors, and OpenCV is used to realize color-based target recognition. First, the read-in image is grayscale processed, and then filtered to convert the image into an HSV image. The target color is recognized through the OpenCV internal function library, and then the area outline is extracted. Finally, the target's circumscribed rectangle is drawn and cropped to obtain an image containing only the handwritten signature content.

[0039] 3.2 Content Identification

[0040] a. Algorithm selection

[0041] Due to personal writing habits, the font size and number of words in the handwritten signature content are not fixed. In order to avoid manual cutting, a sequence learning model is introduced. When the number of words to be decoded in the picture is not fixed, there is no essential difference between decoding audio into Chinese characters and decoding pictures into Chinese characters. Therefore, it is completely feasible to build an algorithm network guided by LSTM and CTC, which are widely used in the field of speech recognition.

[0042] b. Dataset

[0043] The training set and validation set use the open source handwritten Chinese character dataset CASIA-HWD. When reading data, first get the path of all images, add the corresponding label to each image, convert the path and label list into the corresponding vector, and slice it, then read and preprocess the graphics (including image inversion, brightness change, etc.), and the unique hot encoding representation of the label, form a shuffle batch, and throw it into the model.

[0044] c. Algorithm network construction

[0045] The CNN+RNN+CTC method is used to recognize handwritten Chinese characters. CNN is responsible for extracting image features, and RNN uses a bidirectional LSTM network (BiLSTM) to continue extracting text sequence features based on convolutional features. CTCLoss is used instead of Softmax Loss, so that the training samples do not need to be aligned, solving the problem of inconsistent output and label lengths and avoiding manual strict alignment.

[0046] d. Handwriting recognition process

[0047] Handwriting recognition first detects the text position, then defines the text block slicing function, and finally performs character recognition on the sliced ​​single word images.

[0048] The steps of handwriting recognition are as follows Figure 4 As shown. First, build a handwriting recognition algorithm model for calling. Input a text image, slice it into text blocks, perform perspective transformation based on four points, and transform the quadrilateral represented by the four points into a horizontal rectangular image. Grayscale the image that needs to be recognized, perform predictive recognition, and obtain the recognition result.

[0049] 4. Content Correction

[0050] a. Algorithm selection

[0051] Compared with RNN and LSTM, BERT can read the entire text sequence at one time, extract text information at multiple levels, and interpret text semantics more comprehensively. Therefore, the Soft-MaskedBert algorithm is used for modeling.

[0052] b. Dataset

[0053] When training a model, using a self-built dataset can often achieve customization of the corpus. Compared with general public datasets, this has unique advantages and can relatively accurately detect the performance of the model in a specific field.

[0054] Therefore, we crawled the text content of public documents in different fields, and used the jieba word segmentation library and the hit_stopwords stop word list to segment the original corpus according to the word frequency, forming a dictionary that can be used for deep network training. Through word segmentation, we performed error detection from both the character granularity and the word granularity, integrated the suspected error results of the two granularities, formed a candidate set of suspected error positions, and then corrected the errors by replacing the words in the wrong position with similar words. Finally, by calculating the sentence perplexity, we compared and sorted all the candidate set results, obtained the optimal correction word, and obtained the text content after error recognition. The original corpus was divided into short sentences with punctuation marks as interruptions, forming 10,000 text sequences as the real input corpus for deep network training.

[0055] c. Algorithm network construction

[0056] The error correction model is constructed with reference to the Soft-MaskedBert algorithm. The main body is divided into two parts: the error detection network and the error correction network. The error detection network is a bidirectional GRU (Gate Recurrent Unit) network, that is, a bidirectional gated recurrent unit. Its function is to predict the probability of errors in each position of the character. The error correction network is a correction network based on BERT, which corrects and replaces the characters at the position where the error detection is successful.

[0057] The error detection network and the error correction network are connected into a whole through Soft-Masked technology, and the output information of the error detection network is the input information of the BERT correction network.

[0058] d. Model training

[0059] During training, a MASK strategy based on a confusion set is used to perform word replacement based on glyph confusion, random replacement, and original word unchanged on 15% of the tokens in the training set. The original text sentence is converted into a word vector through the embedding mechanism of BERT, and the vector is input into the bidirectional GRU of the error detection network. The network regards each text data character as a word, and uses the bidirectional GRU to capture context information, learn the error probability of each character, and mask the most likely erroneous characters. The error correction network uses BERT to encode the masked character sequence, capture context information and generate appropriate outputs, and generate a corrected character sequence based on the prediction results. Multiple epochs are performed continuously during training to obtain the final training model.

[0060] e. Error correction process

[0061] The steps for text correction are as follows: Figure 4 shown.

[0062] The text output from the handwriting recognition phase is segmented by word frequency to form a word sequence that can be used for deep network training. It is input into the error correction model, and the error detection network is used to obtain all suspected error positions. Then the error correction network is used to correct the errors, and the words in the wrong position are replaced with similar words. Finally, the sentence perplexity is calculated, and all candidate set results are compared and sorted to obtain the optimal correction word and the text content after correction.

[0063] 5. Text Generation

[0064] Use freemarker technology through SpringBoot to generate text files according to templates.

[0065] Create a word template and define placeholders, such as the name of the reviewer, review time, review content, etc. Use ${string} to fill in the placeholders. The placeholder field name should be consistent with the review record instance attribute name. For different copying formats, different review-related attribute data needs to be spliced ​​to form a standard copying statement, so different template files need to be constructed and uploaded. For application management of different templates, a template can be selected to be applied to the review process of a certain type of file; different versions of the same template can be saved and restored to the previous version.

[0066] After uploading the template file, convert the word file into an XML document, and then save the XML file into FTL format. Because the approval process is multi-node, there are multiple approval data under the same file, so the freemarker dependency is introduced into the project, and the freemarker list tag is added to support the circular display of the approval content.

[0067] The steps to generate the document are as follows Figure 5 shown.

[0068] Read the template file and Freemarker parses the corresponding tags. After the handwritten signature is submitted, the review file identifier is bound to form an instruction record in the database. Get the list data of the instruction record from the database, assemble the data, replace the variables according to the FTL template file, form a file stream, and generate a word document. The document content is consistent with the selected template, and the field placeholder will be replaced with the corresponding form field content.

[0069] Compared with the prior art, the present invention proposes an intelligent system design for the entire process of copying and clarification of official documents based on artificial intelligence, and selects appropriate algorithms for each processing flow of the task to optimize the framework. For the recognition of handwritten signature content, a handwriting recognition algorithm model is established. First, the image of the handwritten annotation area is obtained to provide input for handwritten content recognition, and the handwritten instruction content is identified and extracted, avoiding the manual excerpting and entry work, and reducing the difficulty and quantity of manual recognition work. In order to improve the accuracy of the final content of handwritten instructions and increase the content correction link, a content correction algorithm model is established. Error detection is performed from both the character granularity and the word granularity through word segmentation, and the optimal correction word is obtained through the language model to complete the error correction, which greatly improves the final accuracy of handwritten content recognition; through the template engine technology, the document is automatically generated according to the constructed copying template, eliminating the manual sorting and document entry work, and greatly improving the file circulation speed.

[0070] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A design method for an intelligent copying system for official document instructions based on artificial intelligence, characterized in that: The intelligent copying and clearing system for official document approval is used for text view analysis and text document generation. Among them, text view analysis includes handwriting recognition and content error correction. The service is divided into offline and online modes. Text document generation is supported by the template engine, and text content documents are automatically generated according to different templates. The steps used by the system include: First, the original file is obtained and converted into an image format. The region is cropped to obtain an image containing only the handwritten signature content. The image is input into the handwriting recognition model to perform end-to-end text recognition on text sequences of indefinite length. The recognized text content is input into the text error correction model. The text error correction model includes error detection and error correction. All suspected error positions are obtained through error detection. Then, through error correction, the words in the error position are replaced with similar words. Finally, the sentence perplexity is calculated through the language model. All candidate set results are compared and sorted to obtain the optimal correction word and the corrected text content. When splicing data, relevant element data is obtained from the database; in the online mode, after each handwritten approval is completed, the approval content is identified through the service interface call, and the approval time and name of the approver are obtained through the approval system, and stored in the database and bound to the corresponding file for persistence; in the offline mode, the files already on the object storage server are selected for identification and the element data is spliced; Before the document is circulated for review, approval and circulation, the text content after error correction is obtained, and a cleared content document is generated according to the cleared format template. The document is circulated together with the original document, and the document content is supplemented and updated each time it is circulated. When viewing the original text, you can choose to view the approval opinions and display the approved content collection in the corresponding text format.

2. The design method of the intelligent copying system for official document instructions based on artificial intelligence as claimed in claim 1 is characterized in that: The system uploads the text view file to the storage service and uses object storage technology to store it to achieve offline recognition, or the system calls the online recognition interface to provide real-time related capabilities through interface services.

3. The design method of the intelligent copying system for official document instructions based on artificial intelligence as claimed in claim 1 is characterized in that: The system is divided into 4 layers: presentation layer, transport layer, service layer and data layer; The presentation layer is the user-oriented Web end. The system implements Web-based online handwriting content recognition for content display, export, and document generation operations. It is built using the React development framework. The transport layer is responsible for data transmission between the presentation layer and the service layer. It uses HTTP network request related technologies for data transmission, while maintaining the RESTful style in interface design. The service layer is responsible for the implementation of all business functions of this system. It adopts SpringBoot technology and SpringAOP to realize the system's log, exception handling and permission control functions. MyBatis is used as the persistence layer framework. TensorFlow and Opencv related technologies are used to implement handwritten text recognition, text error correction, and text generation algorithms. At the same time, the FlaskPythonWeb framework is used to deploy the algorithm online. Finally, ResTemplate is used to realize the communication between the business server and the algorithm server. The data layer is responsible for storing data such as instruction records and template information. It uses the MySQL database to store system-related data and the OOS object storage service to store various types of documents, copy format templates, and generated copy content document files.

4. The design method of the intelligent copying system for official document instructions based on artificial intelligence according to any one of claims 1 to 3, characterized in that: When recognizing handwriting, you first need to get the original document with the handwritten signature content, convert it into an image, obtain the handwritten signature content area as the input of the handwriting recognition model, and obtain the text content of the handwriting after recognition; The document text and handwritten signature content are distinguished by different colors. OpenCV is used to realize color-based target recognition: first, the read-in image is grayscale processed, and then filtered to convert the image into an HSV image. The target color is recognized through the OpenCV internal function library, and then the area outline is extracted. Finally, the target's circumscribed rectangle is drawn and cropped to obtain an image containing only the handwritten signature content; The handwriting recognition process is as follows: first detect the text position, then define the text block slicing function, and finally perform character recognition on the cut single-word image; specifically, first build a handwriting recognition algorithm model for calling, input the text image, perform text block slicing on it, perform perspective transformation based on four points, transform the quadrilateral represented by the four points into a horizontal rectangular image, grayscale the image that needs character recognition, perform predictive recognition, and obtain the recognition result.

5. The design method of the intelligent copying system for official document instructions based on artificial intelligence as claimed in claim 4 is characterized in that: The training and validation sets of the handwriting recognition algorithm model use the open source handwritten Chinese character dataset CASIA-HWD. When reading the data, we first get the paths of all images, add corresponding labels to each image, convert the paths and label lists into corresponding vectors, and slice them. Then, we read and preprocess the images, and use the unique hot encoding representation of the labels to form shuffle batches, which are then fed into the model. The handwriting recognition algorithm model uses the CNN+RNN+CTC method to recognize handwritten Chinese characters. CNN is responsible for extracting image features, and RNN uses a bidirectional LSTM network to continue extracting text sequence features based on convolutional features. CTC Loss is used instead of Softmax Loss, so that training samples do not need to be aligned.

6. The design method of the intelligent copying system for official document instructions based on artificial intelligence as claimed in claim 4 is characterized in that: When correcting content, the text output from the handwriting recognition stage is segmented according to word frequency to form a word sequence that can be used for deep network training. The sequence is then input into the error correction model, and error detection is performed through the error detection network in the error correction model to obtain all suspected error locations. Error correction is then performed through the error correction network, and words in the wrong positions are replaced with similar words. Finally, the sentence perplexity is calculated, and all candidate set results are compared and sorted to obtain the optimal correction word and the corrected text content.

7. The design method of the intelligent copying system for official document instructions based on artificial intelligence as claimed in claim 6 is characterized in that: When training the error correction model, the use of self-built data sets can often achieve customization of the corpus set; the text content of public documents in different fields is captured, and the original corpus is segmented by word frequency using the jieba word segmentation library and the hit_stopwords stop word list to form a dictionary that can be used for deep network training. Error detection is performed from both character granularity and word granularity through word segmentation, and the suspected error results of the two granularities are integrated to form a candidate set of suspected error positions. Then, through error correction, the words in the wrong position are replaced by similar words. Finally, by calculating the sentence perplexity, all candidate set results are compared and sorted to obtain the optimal correction word and the text content after error recognition. The original corpus is divided into short sentences with punctuation marks as interruptions to form 10,000 text sequences as the real input corpus for deep network training.

8. The design method of the intelligent copying and clearing system for official document instructions based on artificial intelligence as claimed in claim 7 is characterized in that: The error correction model is divided into two parts: the error detection network and the error correction network. The error detection network is a bidirectional GRU network, that is, a bidirectional gated recurrent unit, which is used to predict the probability of errors at various positions of characters. The error correction network is a correction network based on BERT, which corrects and replaces the characters at the positions where the errors are successfully detected. The error detection network and the error correction network are connected into a whole through Soft-Masked technology, and the output information of the error detection network is the input information of the BERT correction network. The MASK strategy based on the confusion set is used when training the error correction model. 15% of the tokens in the training set are replaced with confusing words based on glyphs, randomly replaced, and the original words remain unchanged. The original text sentence is converted into a word vector through BERT's embedding mechanism, and the vector is input into the bidirectional GRU of the error detection network. The network regards each text data character as a word, and uses the bidirectional GRU to capture context information, learn the error probability of each character, and mask the characters that are most likely to be wrong; The error correction network uses BERT to encode the masked character sequence, capture contextual information and generate appropriate outputs, and generates a corrected character sequence based on the prediction results; multiple epochs are performed continuously during training to obtain the final training model.

9. The method for designing an intelligent copying system for official document instructions based on artificial intelligence according to any one of claims 6 to 8, characterized in that: When generating text documents, use freemarker technology through SpringBoot to generate text files according to templates; formulate word templates and define placeholders, including: reviewer name, review time and review content, use ${string} to hold placeholders, and keep the placeholder field name consistent with the review record instance attribute name; for different copy-clearing formats, it is necessary to splice different review-related attribute data to form standard copy-clearing statements, so different template files are constructed and uploaded; application management is performed for different templates, and a certain template is selected to be applied to the review process of a certain type of file; different versions of the same template are saved for restoration to the previous version; After uploading the template file, convert the word file into an XML document, and then save the XML format file into FTL format; introduce freemarker dependency into the project, and support the circular display of instruction content by adding freemarker list tags.

10. The design method of the intelligent copying system for official document instructions based on artificial intelligence as claimed in claim 9 is characterized in that: The text document generation process is as follows: read the template file, parse the corresponding tags with freemarker, bind the review file identifier after the handwritten signature is submitted, and form an instruction record in the database; obtain the list data of the instruction record from the database, assemble the data, replace the variables according to the ftl template file, form a file stream, and generate a word document. The document content is consistent with the selected template, and the field placeholder will be replaced with the corresponding form field content.

Images