Device for automatically filling data in electronic report

Abstract

The invention relates to the technical field of machine learning, in particular to a device for automatically filling data in an electronic report. According to the automatic data filling device for the electronic report, the basic support component provides basic capability support for each function of the system; the data collection management component is accessed to multiple data sources to realize standardization and tagging of data; the report management component realizes report template design management and report task management, and the intelligent filling component realizes automatic filling and automatic calculation of reports by utilizing artificial intelligence and machine learning technologies; the operation management assembly realizes management capability required by daily operation of the system and provides system operation condition analysis monitoring and system operation abnormity early warning analysis capability. According to the device for automatically filling the data based on the electronic report, various reports issued by a superior can be pre-filled, the data source of the filled data can be clearly informed, primary-level workers only need to check and confirm, meanwhile, manual filling is supported, and the workload of the primary-level workers is relieved.

Description

Technical Field

[0001] The present invention relates to the technical field of machine learning, and in particular to a device for automatically filling in data in an electronic report. Background Art

[0002] For a long time, grassroots work in various fields has been faced with problems such as numerous forms, repeated collection, multiple reporting, cross-level, and cross-system data difficulty, which has increased the burden on grassroots management departments and enterprises. The inconsistent statistical standards, methods, and calibers of various departments have formed "data islands" and "data chimneys", and there are problems such as slow progress in information sharing.

[0003] At present, most of the existing reporting systems are based on form filling. For grassroots staff, it is just an additional system, which does not fundamentally reduce the purpose of data filling. The data in the data resource pool that has been established by the management department or the enterprise is not fully utilized.

[0004] In addition, the intelligent reporting system has a slow processing rate for large amounts of data, a slow response speed, and average system stability, which is prone to crashes. Moreover, the data collection of the intelligent reporting system mostly relies on internal enterprise data, and the algorithms and inferences for big data are relatively weak, making it difficult to provide accurate risk predictions for enterprises. Enterprise managers cannot take timely measures, which can easily lead to further expansion of risks. Moreover, there is a risk of leakage when system data is shared, and security cannot be guaranteed, so it is urgently needed to be improved.

[0005] In view of the above problems, the present invention proposes a device for automatically filling in data in electronic reports. Summary of the invention

[0006] In order to make up for the defects of the prior art, the present invention provides a simple and efficient device for automatically filling in data in electronic reports.

[0007] The present invention is achieved through the following technical solutions:

[0008] A device for automatically filling in data in electronic reports, comprising a basic support component, a data collection and management component, a report management component, an intelligent filling component and an operation management component;

[0009] The basic support component includes an organizational structure module, a service gateway module, a process engine module, an identity and authority authentication module, a large model base module and a data encryption module, which are used to provide basic capability support for various functions of the system and ensure stable operation of the system;

[0010] The organizational structure module is responsible for providing the ability to manage the hierarchy and authority of the organization;

[0011] The service gateway module, based on WebService technology, uses microservice architecture to decompose system functions into discrete services step by step to achieve decoupling of the entire system, so that each application module can be independently developed, managed and iterated; using cloud architecture and platform-based deployment, management and service functions in decentralized components makes system delivery and maintenance easier;

[0012] The process engine module is responsible for supporting the task flow of reports, providing the ability to customize processes and free processes, and performing flexible authority control on each process node;

[0013] The identity and authority authentication module is responsible for providing the system with unified identity and authority management capabilities and supporting integration with external systems to improve the compatibility and adaptability of the system;

[0014] The large model base module is based on a lightweight visual large model with 7 billion parameters (7B), which is responsible for supporting semantic recognition and automatic data reporting capabilities. The large model supports 128K long text input and has built-in image processing OCR and recognition. In addition, self-supervised learning fine-tuning (SFT) and QLoRA fine-tuning are performed on the professional terminology data in the reporting report to ensure the recognition accuracy of the large model.

[0015] The data encryption module uses a data encryption module that supports the national encryption algorithm to provide data encryption support for data storage and data display.

[0016] The data collection and management component includes a data source management module, an exchange task management module, a data standard management module, a data processing and management module, a keyword management module, an intelligent reporting database management module, and a data service publishing management module, which are used to access multiple data sources and provide data support for the intelligent reporting system; at the same time, the built-in data standard management module, data processing and management module, and keyword management module are used to realize the standardization and labeling process of data, providing support for intelligent data reporting;

[0017] The data source management module is used to connect to multiple data sources and manage and assign permissions to data sources;

[0018] The exchange task management module includes an exchange middleware management unit, an exchange task configuration unit, and an exchange task monitoring unit. It is responsible for configuring exchange tasks using the exchange middleware after selecting the data source for the collected data, and collecting the data into the original database for storage.

[0019] The data standard management module includes a standard rule management unit, a data detection task management unit, and a standardization governance unit, and is responsible for standardizing and processing data collected from different data sources;

[0020] The data processing and governance management module includes a data fusion rule management unit, a data fusion task management unit, and a fusion task monitoring unit. It is responsible for customizing and managing data fusion rules based on data fusion requirements, and storing the data in the standard library after processing and governance.

[0021] The keyword management module, including the keyword library management unit, the intelligent annotation module unit, the manual annotation module unit and the keyword relationship map unit, is responsible for intelligently analyzing the table structure, metadata and data content in the intelligent reporting database to generate data keywords; it supports manual maintenance of keywords and generates keyword relationship chains using the keyword relationship map after keyword generation, providing support for intelligent reporting of reports and data query and retrieval;

[0022] The intelligent reporting database management module includes the database structure management unit, data resource management unit, database permission management unit and database sharing management unit. It is responsible for designing and managing the intelligent reporting database, supporting data table creation and full life cycle management of data resources, and can set field-level permissions for all database tables to ensure data access security. It provides sharing capabilities to users at different levels and departments through the database sharing management unit.

[0023] Data service publishing management, which includes service orchestration unit, service rule management unit, service publishing management unit, service permission management unit and call status monitoring unit, is based on the basic supporting component capabilities of the service gateway module. It orchestrates and publishes interface services for data in the intelligent reporting database according to business demand logic, and provides data services for other functional modules.

[0024] The report management component includes two parts: report template design management and report task management;

[0025] The report template design and management part includes the report template design module, report template management module, report template authorization module, report template editing module, data source association module and report overview module, which are used to implement report template design, data source connection and permission authorization;

[0026] The report task management part includes report task publishing module, report task circulation module, report filling module, report task review module, report sharing authorization module and report query module, which are used to realize report publishing, circulation and authorization sharing;

[0027] Report template design module, using JavaScript technology to develop and design report template designers for Excel tables;

[0028] The report template management module is responsible for managing and editing the designed report templates to achieve daily data maintenance. It is linked with the report data to automatically identify the templates that already have report data, only supports conditional editing, and performs template deletion protection to protect the security of report data.

[0029] The report template authorization module is used to authorize and manage the usage scope of the report template. Authorized users can view, use and modify the report template according to the authorized permissions.

[0030] The report template editing module is responsible for editing report templates that require editing;

[0031] The data source association module is responsible for associating the existing data in the report system and realizing intelligent pre-filling in combination with the intelligent filling component;

[0032] Report overview module, used to view and preview the designed report templates;

[0033] The report task publishing module is responsible for helping users authorized by the publishing unit to publish report tasks. Users can customize the departments that need to receive report tasks, and specify the reporting requirements and reporting time. At the same time, users are allowed to view the report circulation status, report reporting status, and report data after filing at any time.

[0034] The report task circulation module is responsible for the process circulation according to the reporting scope set when the report task is released by the report task release module, and publishes pending messages to the circulation receiving units. After checking the requirements, the receiving units will distribute the information to the specific reporting unit accounts for reporting. At the same time, users are allowed to view the circulation status and report reporting status of the distributed reports at any time.

[0035] The report filling module is responsible for combining the intelligent filling component to realize intelligent data matching, and intelligently matches and fills data according to the data owned by the user; at the same time, it supports manual matching of data sources, and visualizes the corresponding relationship of data through the drag-and-drop mode on the interface to realize automatic filling of data; and fills the report part without data support through manual mode;

[0036] The report task review module is responsible for helping the circulation review unit to review the report content after the report filling module completes the filling and submission. It supports the auditor to introduce the filling standard for auxiliary review. If the review fails, the report will be returned to the report filling post for further filling. If the review passes, the report will be automatically collected to the publishing unit, and the publishing unit will see the archived report data in the report task publishing module;

[0037] The report sharing authorization module is responsible for helping the publishing unit to authorize the sharing of archived report data to other units with demand; the publishing unit customizes the fields, data ranges and desensitized fields allowed to be shared in the report, as well as the permission for the shared unit to download the form;

[0038] The report query module is responsible for helping report publishing units and units authorized to share to query report data. It supports multi-condition queries for single report data and compound multi-condition queries for multiple reports. It combines with the intelligent reporting component to realize intelligent query display based on voice and text question and answer.

[0039] The intelligent reporting component includes a semantic recognition module, a semantic parsing module, an intelligent association module, an intelligent filling module, an intelligent calculation module and an intelligent recommendation module, which are responsible for realizing automatic filling and automatic calculation of reports by using artificial intelligence and machine learning technology;

[0040] The semantic recognition module is responsible for using the deep learning model to capture the semantic information in the text, understand the context of the sentence, and thus identify the key information. It converts the text to be recognized into the corresponding word embedding vector to form the input matrix X, and then creates three new weight matrices, namely W Q , W K and W V , multiply the matrix X by the matrix W Q , W K and W V After obtaining the query matrix Q, key matrix K and value matrix V, calculate the attention matrix Z:

[0041]

[0042] Among them, d k represents the dimension of the key vector in the key matrix K;

[0043] The semantic parsing module is responsible for converting natural language into a structured data format using syntactic analysis and semantic role labeling technology, helping the system understand sentence components and their relationships, and thus extracting meaningful information;

[0044] Specifically, we use machine learning models to learn sentence structures and determine the relationship between words in a sentence, including determining noun phrases NP and verb phrases VP. We use the BiLSTM-CRF model to label the arguments of the predicates by minimizing the loss function:

[0045] L = -log(P(b|a))

[0046] Determine the label sequence b corresponding to the input sequence a, which helps understand who did what and to whom the action occurred.

[0047] The intelligent association module is responsible for using the knowledge graph to store and manage the relationships between entities. It uses graph neural network technology to find patterns and associations in the knowledge graph and perform the following calculations:

[0048]

[0049] Among them, h l (v) represents the feature vector of node v at layer l, W l represents the weight matrix of the lth layer, N(v) represents the set of neighbor nodes of node v, and u represents the neighbor nodes of node v;

[0050] Update the feature vectors of nodes in each layer of the graph neural network, use entity linking technology to identify entities in the text, and link them to the corresponding entities in the knowledge graph;

[0051] The intelligent filling module is responsible for combining the structured information obtained by the semantic recognition module, semantic parsing module and intelligent association module, and realizing automatic data filling by using keyword search, fuzzy matching, prediction algorithm and context perception;

[0052] The intelligent calculation module is responsible for combining the structured information obtained by the semantic recognition module, the semantic parsing module and the intelligent association module, calling statistical tools and calculation tools to process and calculate data, and realizing intelligent filling of the total item and average item fields;

[0053] The intelligent recommendation module is responsible for using collaborative filtering technology to recommend supplementary content for the form based on the historical behavior of the reporting object and the similarity of other reporting objects. The similarity of the reporting objects in collaborative filtering is calculated by the Pearson correlation coefficient r xy To measure:

[0054]

[0055] Among them, x i and i are the scores of the two respondents on the project, and x and y are their average scores.

[0056] The operation management component includes two parts: system operation and maintenance management and operation monitoring and analysis;

[0057] The system operation and maintenance management part includes the organization management module, user management module, permission management module and role management module, which are used to realize the management capabilities required for the daily operation of the system;

[0058] The operation monitoring and analysis part includes the system log module, data fault warning module, multi-source data conflict warning module, data usage analysis module, user access analysis module and data detection and correction module, and is responsible for providing system operation status analysis and monitoring and system operation abnormality warning and analysis capabilities.

[0059] The organization management module is responsible for managing the organizational structure of the organization, supports unlimited hierarchical organizational trees, and supports integration with the organizational structure of existing systems;

[0060] The user management module is associated with the organization management module and is responsible for creating and setting users within the organization;

[0061] The authority management module is responsible for setting the authority of the user;

[0062] The role management module is responsible for setting and managing user roles and supporting centralized authorization of roles.

[0063] The system log module is responsible for displaying and querying the log information recorded by all system operations;

[0064] The data failure warning module is responsible for warning and displaying abnormal situations including data connection failure and data access failure, and promptly reminding the system administrator of data source connection failure and abnormal query return value;

[0065] The multi-source data conflict warning module is responsible for visually displaying and reminding the data inconsistency in the collected multi-source data content found in data detection;

[0066] The data usage analysis module is responsible for analyzing the usage of data managed in the system, and supports display by day, week, month, quarter or year, and also supports classified display by report or department dimension;

[0067] The user access analysis module is responsible for analyzing the system usage of each user, including user login analysis, user function usage analysis and user data query analysis;

[0068] The data detection and correction module is responsible for displaying and reminding problems that occur during the detection process of the data collected in the system, including overdue data updates, abnormal data changes and data standards that are inconsistent.

[0069] The beneficial effect of the present invention is that the device for automatically filling in data based on electronic reports can form data connection with the intelligent reporting system by combing, managing and connecting the data resources of the established data resource pool, labeling and data standard management of the data, and utilizing the retrieval enhancement generation technology of large models and the high-order thinking chain technology of intelligent body planning. It can pre-fill various reports issued by superiors and clearly inform the data source of the filled data. Grassroots staff only need to check and confirm, and the staff will fill in the parts that are not covered by the data manually, which greatly reduces the workload of grassroots staff. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0071] Attached Figure 1 Schematic diagram of the device architecture for automatically filling in data in electronic reports of the present invention.

[0072] Attached Figure 2 Schematic diagram of the data collection and management method of the present invention.

[0073] Attached Figure 3 It is a schematic diagram of the report circulation process of the present invention.

[0074] Attached Figure 4 This is a schematic diagram of the architecture of the intelligent reporting module of the present invention. DETAILED DESCRIPTION

[0075] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0076] The device for automatically filling in data in electronic reports includes a basic support component, a data collection and management component, a report management component, an intelligent filling component and an operation management component;

[0077] The basic support component includes an organizational structure module, a service gateway module, a process engine module, an identity and authority authentication module, a large model base module and a data encryption module, which are used to provide basic capability support for various functions of the system and ensure stable operation of the system;

[0078] The organizational structure module is responsible for providing the ability to manage the levels and permissions of legal organizations such as management departments, enterprises, and social organizations;

[0079] The service gateway module, based on WebService technology, uses microservice architecture to decompose system functions into discrete services step by step to achieve decoupling of the entire system, so that each application module can be independently developed, managed and iterated; using cloud architecture and platform-based deployment, management and service functions in decentralized components makes system delivery and maintenance easier;

[0080] The process engine module is responsible for supporting the task flow of reports, providing the ability to customize processes and free processes, and performing flexible authority control on each process node;

[0081] The identity and authority authentication module is responsible for providing the system with unified identity and authority management capabilities and supporting integration with external systems to improve the compatibility and adaptability of the system;

[0082] The large model base module is based on a lightweight visual large model with 7 billion parameters (7B), which is responsible for supporting semantic recognition and automatic data reporting capabilities. The large model supports 128K long text input and has built-in image processing OCR and recognition. In addition, self-supervised learning fine-tuning (SFT) and QLoRA fine-tuning are performed on the professional terminology data in the reporting report to ensure the recognition accuracy of the large model.

[0083] It can be based on large model bases such as Tongyi Qianwen, Wenxin Yiyan, Doubao, etc., or it can be based on the open source transformers architecture.

[0084] The data encryption module uses a data encryption module that supports the national encryption algorithm to provide data encryption support for data storage and data display.

[0085] The data collection and management component includes a data source management module, an exchange task management module, a data standard management module, a data processing and management module, a keyword management module, an intelligent reporting database management module, and a data service publishing management module, which are used to access multiple data sources and provide data support for the intelligent reporting system; at the same time, the built-in data standard management module, data processing and management module, and keyword management module are used to realize the standardization and labeling process of data, providing support for intelligent data reporting;

[0086] The data source management module is used to connect to multiple data sources and manage and assign permissions to data sources;

[0087] The exchange task management module includes an exchange middleware management unit, an exchange task configuration unit, and an exchange task monitoring unit. It is responsible for configuring exchange tasks using the exchange middleware after selecting the data source for the collected data, and collecting the data into the original database for storage.

[0088] The data standard management module includes a standard rule management unit, a data detection task management unit, and a standardization governance unit. It is responsible for standardizing and processing data collected from different data sources to avoid the "data dialect" problem.

[0089] The data processing and governance management module includes a data fusion rule management unit, a data fusion task management unit, and a fusion task monitoring unit. It is responsible for customizing and managing data fusion rules based on data fusion requirements, and storing the data in the standard library after processing and governance.

[0090] The keyword management module, including the keyword library management unit, the intelligent annotation module unit, the manual annotation module unit and the keyword relationship map unit, is responsible for intelligently analyzing the table structure, metadata and data content in the intelligent reporting database to generate data keywords; it supports manual maintenance of keywords and generates keyword relationship chains using the keyword relationship map after keyword generation, providing support for intelligent reporting of reports and data query and retrieval;

[0091] The intelligent reporting database management module includes the database structure management unit, data resource management unit, database permission management unit and database sharing management unit. It is responsible for designing and managing the intelligent reporting database, supporting data table creation and full life cycle management of data resources, and can set field-level permissions for all database tables to ensure data access security. It provides sharing capabilities to users at different levels and departments through the database sharing management unit.

[0092] Data service publishing management, which includes service orchestration unit, service rule management unit, service publishing management unit, service permission management unit and call status monitoring unit, is based on the basic supporting component capabilities of the service gateway module. It orchestrates and publishes interface services for data in the intelligent reporting database according to business demand logic, and provides data services for other functional modules.

[0093] The report management component includes two parts: report template design management and report task management;

[0094] The report template design and management part includes the report template design module, report template management module, report template authorization module, report template editing module, data source association module and report overview module, which are used to implement report template design, data source connection and permission authorization;

[0095] The report task management part includes report task publishing module, report task circulation module, report filling module, report task review module, report sharing authorization module and report query module, which are used to realize report publishing, circulation and authorization sharing;

[0096] The report template design module uses JavaScript technology to develop and design a report template designer for Excel tables. This report template designer is in line with the habits of most people using Excel tables. The following is a simple Excel report design implemented using Node.JS. At present, the built-in functional modules of this module are completely low-code customized, and functions can be realized by quickly dragging and dropping:

[0097] / / Load table data

[0098] app.get(" / data",(req,res)=>{

[0099] fs.readFile("data.json","utf8",(err,data)=>{

[0100] if(err)return res.status(500).send("Error reading data.");

[0101] res.json(JSON.parse(data));

[0102] });

[0103] });

[0104] / / Save table data

[0105] app.post(" / data",(req,res)=>{

[0106] fs.writeFile("data.json",JSON.stringify(req.body),(err)=>{

[0107] if(err)return res.status(500).send("Error saving data.");

[0108] res.send("Data saved.");

[0109] });

[0110] });

[0111] app.listen(3000,()=>console.log("Server running on http: / / localhost:3000"));

[0112] The report template management module is responsible for managing and editing the designed report templates to achieve daily data maintenance. It is linked with the report data to automatically identify the templates that already have report data, only supports conditional editing, and performs template deletion protection to protect the security of report data.

[0113] The report template authorization module is used to authorize and manage the usage scope of the report template. Authorized users can view, use and modify the report template according to the authorized permissions.

[0114] The report template editing module is responsible for editing report templates that require editing. Its functions are similar to those of the report design module.

[0115] The data source association module is different from traditional reporting software. In the report design stage of this system, in addition to designing report templates, the data source association module can also be used to associate the existing data in the report system, and combine it with the intelligent reporting component to realize intelligent pre-filling, thereby reducing the burden of grassroots data reporting from the source;

[0116] Report overview module, used to view and preview the designed report templates;

[0117] The report task publishing module is responsible for helping users authorized by the publishing unit to publish report tasks to other departments or subordinate departments. Users can customize the departments that need to receive report tasks, and specify the reporting requirements and reporting time. At the same time, users are allowed to view the report circulation status, report reporting status, and report data after filing at any time;

[0118] The report task circulation module is responsible for the process circulation according to the reporting scope set when the report task is released by the report task release module, and publishes pending messages to the circulation receiving units. After checking the requirements, the receiving units will distribute the information to the specific reporting unit accounts for reporting. At the same time, users are allowed to view the circulation status and report reporting status of the distributed reports at any time.

[0119] The report filling module is responsible for combining the intelligent filling component to realize intelligent data matching, and intelligently matches and fills in data according to the data owned by the user; at the same time, it supports manual matching of data sources, and visualizes the corresponding relationship of operating data through the drag-and-drop mode on the interface to realize automatic filling of data; it also fills in the report part without data support through manual mode; this function conforms to the operation habits of tables such as Excel, which greatly reduces the difficulty of use.

[0120] The report task review module is responsible for helping the circulation review unit to review the report content after the report filling module completes the filling and submission. It supports the auditor to introduce the filling standard for auxiliary review. If the review fails, the report will be returned to the report filling post for further filling. If the review passes, the report will be automatically collected to the publishing unit, and the publishing unit will see the archived report data in the report task publishing module;

[0121] The report sharing authorization module is responsible for helping the publishing unit to authorize the sharing of archived report data to other units in need; the publishing unit customizes the fields, data ranges and desensitized fields allowed to be shared in the report, as well as the permission for the shared unit to download the form; while maximizing data sharing, it ensures data security.

[0122] The report query module is responsible for helping report publishing units and units authorized to share to query report data. It supports multi-condition queries for single report data and compound multi-condition queries for multiple reports. It combines with the intelligent reporting component to realize intelligent query display based on voice and text question and answer.

[0123] The intelligent reporting component includes a semantic recognition module, a semantic parsing module, an intelligent association module, an intelligent filling module, an intelligent calculation module and an intelligent recommendation module, which are responsible for realizing automatic filling and automatic calculation of reports by using artificial intelligence and machine learning technology;

[0124] The semantic recognition module is responsible for using deep learning models, especially recurrent neural networks (RNNs), long short-term memory networks (LSTMs), transformers, etc., to capture the semantic information in the text, understand the context of the sentence, and thus identify key information, such as numbers, dates, names, etc., convert the text to be recognized into the corresponding word embedding vector to form the input matrix X, and then create three new weight matrices, namely W Q , W K and W V , multiply the matrix X by the matrix W Q , W K and W V After obtaining the query matrix Q, key matrix K and value matrix V, calculate the attention matrix Z:

[0125]

[0126] Among them, d k represents the dimension of the key vector in the key matrix K;

[0127] The semantic parsing module is responsible for converting natural language into a structured data format using syntactic analysis and semantic role labeling technology, helping the system understand sentence components and their relationships, and thus extracting meaningful information;

[0128] Specifically, we use machine learning models (such as maximum entropy models, conditional random fields, etc.) to learn sentence structures and determine the relationship between words in a sentence, including determining noun phrases NP and verb phrases VP. We use the BiLSTM-CRF model to label the arguments of the predicates and minimize the loss function:

[0129] L = -log(P(b|a))

[0130] Determine the label sequence b corresponding to the input sequence a, which helps understand who did what and to whom the action occurred.

[0131] The intelligent association module is responsible for using the knowledge graph to store and manage the relationships between entities. It uses graph neural network technology to find patterns and associations in the knowledge graph and perform the following calculations:

[0132] h l (v) = σ(∑ u∈N(v) W l ·h l-1 (u))

[0133] Among them, h l (v) represents the feature vector of node v at layer l, W l represents the weight matrix of the lth layer, N(v) represents the set of neighbor nodes of node v, and u represents the neighbor nodes of node v;

[0134] Update the feature vectors of nodes in each layer of the graph neural network, use entity linking technology to identify entities in the text, and link them to the corresponding entities in the knowledge graph;

[0135] The intelligent filling module is responsible for combining the structured information obtained by the semantic recognition module, semantic parsing module and intelligent association module, and realizing automatic data filling by using keyword search, fuzzy matching, prediction algorithm and context perception;

[0136] The intelligent calculation module is responsible for combining the structured information obtained by the semantic recognition module, the semantic parsing module and the intelligent association module, calling statistical tools and calculation tools to process and calculate data, and realizing intelligent filling of the total item and average item fields;

[0137] The intelligent recommendation module is responsible for using collaborative filtering technology to recommend supplementary content for the form based on the historical behavior of the reporting object and the similarity of other reporting objects. The similarity of the reporting objects in collaborative filtering is calculated by the Pearson correlation coefficient r xy To measure:

[0138]

[0139] Among them, x i andi They are the scores given to the projects by the two respondents. and is their average rating.

[0140] The operation management component includes two parts: system operation and maintenance management and operation monitoring and analysis;

[0141] The system operation and maintenance management part includes the organization management module, user management module, permission management module and role management module, which are used to realize the management capabilities required for the daily operation of the system;

[0142] The operation monitoring and analysis part includes the system log module, data fault warning module, multi-source data conflict warning module, data usage analysis module, user access analysis module and data detection and correction module, and is responsible for providing system operation status analysis and monitoring and system operation abnormality warning and analysis capabilities.

[0143] The organization management module is responsible for managing the organizational structure of management departments, enterprises, social groups and other organizations, supports unlimited hierarchical organizational trees, and supports integration with the organizational structure of existing systems;

[0144] The user management module is associated with the organization management module and is responsible for creating and setting users within the organization;

[0145] The authority management module is responsible for setting the authority of the user;

[0146] The role management module is responsible for setting and managing user roles and supporting centralized authorization of roles.

[0147] The system log module is responsible for displaying and querying the log information recorded by all system operations. Other monitoring and analysis modules mostly rely on the log data in the system log module.

[0148] The data failure warning module is responsible for warning and displaying abnormal situations including data connection failure and data access failure, and promptly reminding the system administrator of data source connection failure and abnormal query return value;

[0149] The multi-source data conflict warning module is responsible for visually displaying and reminding the data inconsistency in the collected multi-source data content found in data detection;

[0150] The data usage analysis module is responsible for analyzing the usage of data managed in the system, and supports display by day, week, month, quarter or year, and also supports classified display by report or department dimension;

[0151] The user access analysis module is responsible for analyzing the system usage of each user, including user login analysis, user function usage analysis and user data query analysis;

[0152] The data detection and correction module is responsible for displaying and reminding problems that occur during the detection process of the data collected in the system, including overdue data updates, abnormal data changes and data standards that are inconsistent.

[0153] The above is a detailed introduction to a device for automatically filling in data in an electronic report in an example of the present invention. This section uses specific examples to illustrate the principles and implementation methods of the invention. The above examples are only used to help understand the core ideas of the present invention. Without departing from the principles of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

Claims

1. A device for automatically filling in data in electronic reports, characterized in that: Including basic support components, data collection and governance components, report management components, intelligent reporting components and operation management components; The basic support component includes an organizational structure module, a service gateway module, a process engine module, an identity and authority authentication module, a large model base module and a data encryption module, which are used to provide basic capability support for various functions of the system and ensure stable operation of the system; The data collection and management component includes a data source management module, an exchange task management module, a data standard management module, a data processing and management module, a keyword management module, an intelligent reporting database management module, and a data service publishing management module, which are used to access multiple data sources and provide data support for the intelligent reporting system; at the same time, the built-in data standard management module, data processing and management module, and keyword management module are used to realize the standardization and labeling process of data, providing support for intelligent data reporting; The report management component includes two parts: report template design management and report task management; The report template design and management part includes the report template design module, report template management module, report template authorization module, report template editing module, data source association module and report overview module, which are used to implement report template design, data source connection and permission authorization; The report task management part includes report task publishing module, report task circulation module, report filling module, report task review module, report sharing authorization module and report query module, which are used to realize report publishing, circulation and authorization sharing; The intelligent reporting component includes a semantic recognition module, a semantic parsing module, an intelligent association module, an intelligent filling module, an intelligent calculation module and an intelligent recommendation module, which are responsible for realizing automatic filling and automatic calculation of reports by using artificial intelligence and machine learning technology; The operation management component includes two parts: system operation and maintenance management and operation monitoring and analysis; The system operation and maintenance management part includes the organization management module, user management module, permission management module and role management module, which are used to realize the management capabilities required for the daily operation of the system; The operation monitoring and analysis part includes the system log module, data fault warning module, multi-source data conflict warning module, data usage analysis module, user access analysis module and data detection and correction module, and is responsible for providing system operation status analysis and monitoring and system operation abnormality warning and analysis capabilities.

2. The device for automatically filling in data in electronic reports according to claim 1, characterized in that: The organizational structure module is responsible for providing the ability to manage the hierarchy and authority of the organization; The service gateway module, based on WebService technology, uses microservice architecture to decompose system functions into discrete services step by step to achieve decoupling of the entire system, so that each application module can be independently developed, managed and iterated; using cloud architecture and platform-based deployment, management and service functions in decentralized components makes system delivery and maintenance simpler; The process engine module is responsible for supporting the task flow of reports, providing the ability to customize processes and free processes, and performing flexible authority control on each process node; The identity and authority authentication module is responsible for providing the system with unified identity and authority management capabilities and supporting integration with external systems to improve the compatibility and adaptability of the system; The large model base module is based on a lightweight visual large model with 7 billion parameters, which is responsible for providing support for semantic recognition and automatic data reporting capabilities; the large model supports 128K long text input, as well as built-in image processing OCR and recognition; in addition, self-supervised learning fine-tuning and QLoRA fine-tuning are performed on the professional terminology data in the reporting report to ensure the recognition accuracy of the large model; The data encryption module adopts a data encryption module that supports the national encryption algorithm to provide data encryption support for data storage and data display.

3. The device for automatically filling in data in electronic reports according to claim 1, characterized in that: The data source management module is used to connect to multiple data sources and manage and assign permissions to data sources; The exchange task management module includes an exchange middleware management unit, an exchange task configuration unit and an exchange task monitoring unit, which is responsible for configuring the exchange task using the exchange middleware after selecting the data source for the collected data, and collecting the data into the original library for storage; The data standard management module includes a standard rule management unit, a data detection task management unit and a standardization governance unit, and is responsible for standardization detection and processing of data collected from different data sources; The data processing and governance management module includes a data fusion rule management unit, a data fusion task management unit and a fusion task monitoring unit, which are responsible for customizing and managing data fusion rules based on data fusion requirements, and storing the data in the standard library after processing and governance; The keyword management module includes a keyword library management unit, an intelligent annotation module unit, a manual annotation module unit and a keyword relationship map unit, which is responsible for intelligently parsing the table structure, metadata and data content in the intelligent reporting database to generate data keywords; supports manual maintenance of keywords, and generates keyword relationship chains using the keyword relationship map after keyword generation, providing support for intelligent reporting of reports and data query and retrieval; The intelligent reporting database management module includes a database structure management unit, a data resource management unit, a database permission management unit, and a database sharing management unit, which is responsible for designing and managing the intelligent reporting database, supporting data table building and full life cycle management of data resources, and can set field-level permissions for all database tables to ensure data access security, and provide sharing capabilities to users at different levels and departments through the database sharing management unit; The data service publishing management internally includes a service orchestration unit, a service rule management unit, a service publishing management unit, a service authority management unit and a call status monitoring unit. Based on the basic supporting component capabilities of the service gateway module, the data in the intelligent reporting database is intelligently filled in and published according to the business demand logic, providing data services for other functional modules.

4. The device for automatically filling in data in electronic reports according to claim 1, characterized in that: Report template design module, using JavaScript technology to develop and design report template designers for Excel tables; The report template management module is responsible for managing and editing the designed report templates to achieve daily data maintenance. It is linked with the report data to automatically identify the templates that already have report data, only supports conditional editing, and performs template deletion protection to protect the security of report data. The report template authorization module is used to authorize and manage the usage scope of the report template. Authorized users can view, use and modify the report template according to the authorized permissions. The report template editing module is responsible for editing report templates that require editing; The data source association module is responsible for associating the existing data in the report system and realizing intelligent pre-filling in combination with the intelligent filling component; Report overview module, used to view and preview the designed report templates; The report task publishing module is responsible for helping users authorized by the publishing unit to publish report tasks. Users can customize the departments that need to receive report tasks, and specify the reporting requirements and reporting time. At the same time, users are allowed to view the report circulation status, report reporting status, and report data after filing at any time. The report task circulation module is responsible for the process circulation according to the reporting scope set when the report task is released by the report task release module, and publishes pending messages to the circulation receiving units. After checking the requirements, the receiving units will distribute the information to the specific reporting unit accounts for reporting. At the same time, users are allowed to view the circulation status and report reporting status of the distributed reports at any time. The report filling module is responsible for combining the intelligent filling component to realize intelligent data matching, and intelligently matches and fills data according to the data owned by the user; at the same time, it supports manual matching of data sources, and visualizes the corresponding relationship of data through the drag-and-drop mode on the interface to realize automatic filling of data; and fills the report part without data support through manual mode; The report task review module is responsible for helping the circulation review unit to review the report content after the report filling module completes the filling and submission. It supports the auditor to introduce the filling standard for auxiliary review. If the review fails, the report will be returned to the report filling post for further filling. If the review passes, the report will be automatically collected to the publishing unit, and the publishing unit will see the archived report data in the report task publishing module; The report sharing authorization module is responsible for helping the publishing unit to share the archived report data with other units in need; The publishing unit customizes the fields, data ranges, and masked fields that are allowed to be shared in the report, as well as the permission for the shared unit to download the form; The report query module is responsible for helping report publishing units and units authorized to share to query report data. It supports multi-condition queries for single report data and compound multi-condition queries for multiple reports. It combines with the intelligent reporting component to realize intelligent query display based on voice and text question and answer.

5. The device for automatically filling in data in electronic reports according to claim 1, characterized in that: The semantic recognition module is responsible for using the deep learning model to capture the semantic information in the text, understand the context of the sentence, and thus identify the key information. It converts the text to be recognized into the corresponding word embedding vector to form the input matrix X, and then creates three new weight matrices, namely W Q , W K and W V , multiply the matrix X by the matrix W Q , W K and W V After obtaining the query matrix Q, key matrix K and value matrix V, calculate the attention matrix Z: Among them, d k Represents the dimension of the key vector in the key matrix K.

6. The device for automatically filling in data in electronic reports according to claim 1, characterized in that: The semantic parsing module is responsible for converting natural language into a structured data format using syntactic analysis and semantic role labeling technology, helping the system understand sentence components and their relationships, and thus extracting meaningful information; Specifically, we use machine learning models to learn sentence structures and determine the relationship between words in a sentence, including determining noun phrases NP and verb phrases VP. We use the BiLSTM-CRF model to label the arguments of the predicates by minimizing the loss function: L = -log(P(b|a)) Determine the label sequence b corresponding to the input sequence a, which helps understand who did what and to whom the action occurred.

7. The device for automatically filling in data in electronic reports according to claim 1, characterized in that: The intelligent association module is responsible for using the knowledge graph to store and manage the relationships between entities. It uses graph neural network technology to find patterns and associations in the knowledge graph and perform the following calculations: h l (v)=σ(∑ u∈N(v) W l ·h l-1 (u)) Among them, h l (v) represents the feature vector of node v at layer l, W l represents the weight matrix of the lth layer, N(v) represents the set of neighbor nodes of node v, and u represents the neighbor nodes of node v; The feature vectors of nodes in each layer of the graph neural network are updated, and then the entity linking technology is used to identify entities in the text and link them to the corresponding entities in the knowledge graph.

8. The device for automatically filling in data in electronic reports according to claim 1, characterized in that: The intelligent filling module is responsible for combining the structured information obtained by the semantic recognition module, semantic parsing module and intelligent association module, and realizing automatic data filling by using keyword search, fuzzy matching, prediction algorithm and context perception; The intelligent computing module is responsible for combining the structured information obtained by the semantic recognition module, the semantic parsing module and the intelligent association module, calling statistical tools and calculation tools to process and calculate data, and realizing the intelligent filling of the total item and average item fields.

9. The device for automatically filling in data in electronic reports according to claim 1, characterized in that: The intelligent recommendation module is responsible for using collaborative filtering technology to recommend supplementary content for the form based on the historical behavior of the reporting object and the similarity of other reporting objects. The similarity of the reporting objects in collaborative filtering is calculated by the Pearson correlation coefficient r xy To measure: Among them, x i and i They are the scores given to the projects by the two respondents. and is their average rating.

10. The device for automatically filling in data in electronic reports according to claim 1, characterized in that: The organization management module is responsible for managing the organizational structure of the organization, supports unlimited hierarchical organizational trees, and supports integration with the organizational structure of existing systems; The user management module is associated with the organization management module and is responsible for creating and setting users within the organization; The authority management module is responsible for setting the authority of the user; The role management module is responsible for setting and managing user roles and supporting centralized authorization of roles; The system log module is responsible for displaying and querying the log information recorded by all system operations; The data failure warning module is responsible for warning and displaying abnormal situations including data connection failure and data access failure, and promptly reminding the system administrator of data source connection failure and abnormal query return value; The multi-source data conflict warning module is responsible for visually displaying and reminding the data inconsistency in the collected multi-source data content found in data detection; The data usage analysis module is responsible for analyzing the usage of data managed in the system, and supports display by day, week, month, quarter or year, and also supports classified display by report or department dimension; The user access analysis module is responsible for analyzing the system usage of each user, including user login analysis, user function usage analysis and user data query analysis; The data detection and correction module is responsible for displaying and reminding problems that occur during the detection process of the data collected in the system, including overdue data updates, abnormal data changes and data standards that are inconsistent.

Images