Automatic retrieval and automatic form filling system for form information in non-fixed format

Through the automatic retrieval and automatic form filling system of non-fixed format form information, text recognition and vector database technology are used to solve the problems of redundancy and high error rates in rural reporting reports, improve work efficiency and digitalization level, and improve farmers' lives.

CN118966170APending Publication Date: 2024-11-15INST OF URBAN AGRI CHINESE ACADEMY OF AGRI SCI
View PDF 0 Cites 3 Cited by

Patent Information

Application Number
CN202410957896.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-17
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

When filling out various reports, rural village committee managers have a large amount of redundant data and cumbersome manpower work, resulting in high error rates, low work efficiency, and lack of information tools to simplify work processes.

Method used

Provide a system for automatic retrieval and automatic filling of table information in non-fixed format, including information preprocessing units, data units and storage units. The table information is extracted through text recognition model, generate json data, cut and convert it into vector data, and match the vector database. , fill in the vacant content, generate editable forms, and reduce the workload and error rate of filling out forms.

Benefits of technology

It improves the efficiency of filling out forms, reduces the error rate, improves the digitalization level of agriculture and rural areas, improves farmers' lives, and has broad application prospects.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118966170A_ABST
    Figure CN118966170A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of rural informatization, in particular to an automatic retrieval and automatic form filling system for non-fixed format form information. Comprising an information preprocessing unit, a first data unit, a second data unit and an information storage unit, the information preprocessing unit is used for obtaining to-be-processed table information, extracting characters and table formats in a table, generating json data, visually editing the obtained json data and then converting the json data into md format data; the first data unit is used for cutting data of an md format table into small tables according to a fixed size, reserving a table header, converting the cut small tables into vector data, and storing the vector data in the information storage unit; according to the automatic retrieval and automatic form filling system for the non-fixed format form information, the workload of form filling workers is reduced, the working efficiency is improved, the error rate is reduced, the agricultural rural digitization level is improved, and the life of farmers is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of rural information technology, in particular to an automatic retrieval and automatic form filling system for non-fixed format table information. Background Art

[0002] Rural village committee managers usually spend more than 80% of their time filling out various reports. A lot of data is redundant, the filling process is cumbersome, and many new reports are a recombination of data from historical reports. Some tedious manual work urgently needs information technology tools to assist in simplifying the workflow.

[0003] Currently, most form-filling technologies involve automatic recording of fixed-format forms, or the linkage of multiple forms with fixed processes. Information is sorted and filled in according to the process, which results in problems such as high error rates and low work efficiency.

[0004] Based on this, the present invention provides a non-fixed format table information automatic retrieval and automatic form filling system to solve the above-mentioned technical problems! Summary of the invention

[0005] The purpose of the present invention is to provide a system for automatic retrieval and automatic filling of non-fixed format table information. The system provided by the present invention reduces the workload of form-filling staff, improves work efficiency, reduces the error rate, improves the level of digitalization of agriculture and rural areas, and improves farmers' lives.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] The present invention provides a non-fixed format table information automatic retrieval and automatic form filling system, comprising an information preprocessing unit, a first data unit, a second data unit and an information storage unit, wherein:

[0008] The information preprocessing unit is used to obtain the table information to be processed, extract the text and table format in the table, generate json data, and convert the obtained json data into md format data after visual editing;

[0009] The first data unit is used to cut the data of the md format table into small tables according to a fixed size and retain the table header, and convert the cut small tables into vector data and store them in the information storage unit. The first data unit is connected to the information preprocessing unit;

[0010] The second data unit is used to convert the header and directory information in the empty table into vector data, and then match the vector data in the vector database, screen the vector data with the highest similarity, and convert the vector data format into a corresponding text statement, and to locate and screen the content of the table, generate a table in md format, and convert the md table into json data, and then visually generate an editable table. The second data unit is connected to both the information preprocessing unit and the first data unit;

[0011] The information storage unit is used to store the received information, and the information storage unit is connected to both the first data unit and the second data unit.

[0012] The present invention is further configured as follows: the information preprocessing unit includes an information acquisition module, an information extraction module and an information editing module, wherein:

[0013] The information acquisition module is used to acquire a table to be processed;

[0014] The information extraction module extracts corresponding text and table format from the table to be processed based on the obtained table to be processed and uses the text recognition model, and then generates json data. The information extraction module is connected to the information acquisition module;

[0015] The information editing module is used to visualize the obtained JSON data to obtain an editable table, and the information editing module is connected to the information extraction module.

[0016] The present invention is further configured as follows: the information preprocessing unit further includes an information conversion module and a first communication module, wherein:

[0017] The information conversion module is used to convert the json format table data into md format data, and the information conversion module is connected to the information editing module;

[0018] The first communication module is used to realize information interaction between the information preprocessing unit and the first data unit and the second data unit. The first communication module is connected to both the information acquisition module and the information conversion module.

[0019] The present invention is further configured as follows: the first data unit includes a second communication module, an information segmentation module and an information conversion module, wherein:

[0020] The second communication module is used to realize information interaction between the first data unit and the information preprocessing unit, the second data unit and the information storage unit;

[0021] The information cutting module is used to cut the md format table containing data into small tables according to a fixed size and retain the table header, and the information cutting module is connected to the second communication module;

[0022] The information conversion module is used to convert the cut small tables into vector data and store them in the information storage unit. The information conversion module is connected to both the second communication module and the information cutting module.

[0023] The present invention is further configured as follows: the second data unit includes a third communication module, a screening and conversion module and an information filling module, wherein:

[0024] The third communication module is used to realize the connection between the second data unit and the information preprocessing unit, the first data unit and the information storage unit;

[0025] The screening and conversion module is used to convert the header and directory information in the empty table without data into vector data, and then match the vector data in the vector database, screen the vector data with the highest similarity, and convert the vector data format into a corresponding text statement, and the screening and conversion module is connected to the third communication module;

[0026] The information filling module is used to fill in the blank contents of the table, and the information filling module is connected with the screening and conversion module.

[0027] The present invention is further configured as follows: the process of filling in the blank content of the table is as follows:

[0028] The large language model will generate prompt words for filling in the form based on the content of the empty form, and then generate text with an md table based on the prompt words, the content of the form to be filled in, and the search content.

[0029] The present invention is further configured as follows: the second data unit further includes a format generation module and a format adjustment module;

[0030] The format generation module is used to locate and filter out the content of the table and generate a table in md format. The format generation module is connected to the information filling module;

[0031] The format adjustment module is used to convert the md table into json data and visually generate an editable table. The format adjustment module is connected to both the third communication module and the format generation module.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] The automatic retrieval and automatic filling system for non-fixed format table information provided by the present invention reduces the workload of form-filling staff, improves work efficiency, and reduces the error rate, thereby improving the level of digitalization in agriculture and rural areas, and improving farmers' lives. It has broad application prospects and is worthy of promotion.

[0034] Finished Product BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 The present invention is a system diagram of a system for automatically retrieving information in a non-fixed format table and automatically filling in a form.

[0036] Figure 2 It is a system diagram of an information preprocessing unit in a non-fixed format table information automatic retrieval and automatic form filling system of the present invention;

[0037] Figure 3 A system diagram of a first data unit in a system for automatically retrieving information in a non-fixed format table and automatically filling in a form according to the present invention;

[0038] Figure 4 This is a system diagram of the second data unit in a system for automatic retrieval and automatic form filling of non-fixed format table information of the present invention.

[0039] Legend: 100, information preprocessing unit; 110, information acquisition module; 120, information extraction module; 130, information editing module; 140, information conversion module; 150, first communication module; 200, first data unit; 210, second communication module; 220, information segmentation module; 230, information conversion module; 300, second data unit; 310, third communication module; 320, screening and conversion module; 330, information filling module; 340, format generation module; 350, format adjustment module; 400, information storage unit. DETAILED DESCRIPTION

[0040] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention; it is obvious that the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments, and all other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making creative work are within the scope of protection of the present invention.

[0041] In the description of the present invention, it should be noted that the terms "upper", "lower", "inner", "outer", "top / bottom" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific position, be constructed and operated in a specific position, and therefore cannot be understood as limiting the present invention. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance.

[0042] In the description of the present invention, it should be noted that, unless otherwise clearly stipulated and limited, the terms "installed", "provided with", "mounted / connected", "connected", etc. should be understood in a broad sense. For example, "connection" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, and it can be the internal connection of two components; for ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0043] Example:

[0044] like Figure 1-Figure 4 As shown, the present embodiment provides a non-fixed format table information automatic retrieval and automatic form filling system, including an information preprocessing unit 100, a first data unit 200, a second data unit 300 and an information storage unit 400, wherein: the information preprocessing unit 100 is used to obtain the table information to be processed, and extract the text and table format in the table, generate json data, and convert the obtained json data into md format data after visual editing; the first data unit 200 is used to cut the data of the md format table into small tables according to a fixed size and retain the table header, and convert the cut small tables into vector data, and store them in the information storage unit 400, the first data unit 200 and the information storage unit 400 The preprocessing unit 100 is connected; the second data unit 300 is used to convert the header and directory information in the empty table into vector data, and then match the vector data in the vector database, filter the vector data with the highest similarity, and convert the vector data format into the corresponding text statement, and is used to locate and filter the content of the table, generate a table in md format, and convert the md table into json data, and then visually generate an editable table. The second data unit 300 is connected to both the information preprocessing unit 100 and the first data unit 200; the information storage unit 400 is used to store the received information, and the information storage unit 400 is connected to both the first data unit 200 and the second data unit 300.

[0045] In this embodiment, it should be noted that the information preprocessing unit 100 will obtain the table information to be processed, as an implementation method, including but not limited to uploading a paper form by taking a photo or uploading an electronic form by taking a screenshot, etc., through the table, based on the machine learning text recognition model, extract the text and table format in the table, and generate json data, wherein the machine learning-based text recognition model includes but is not limited to a trained deep neural network model, etc., can realize the visualization of json data on a computer or mobile phone, generate an editable table, so as to manually modify and calibrate the data and table format, after calibration, convert the json format data into md (markdown) format data, and then transmit it to the first data unit 200, the first data unit 200 will store the table with data in the information storage unit 400, including but not limited to directly storing the table photo in the information storage unit 400, and cutting the table data into small tables of fixed size and retaining the table header, the small table is converted into vector data based on the embedding technology of natural language processing, and stored in the information storage unit 400;

[0046] In addition, when the table is empty, the second data unit 300 matches the header and directory information in the empty table with the vector data in the vector database, selects the vector data with the highest similarity, and converts the vector data format into a corresponding text sentence. As an implementation method, it can be converted into vector data through embedding technology, and according to the Yuxuan similarity algorithm, the vector data in the vector database is matched, the vector data with the highest similarity is selected, and the vector data format is converted into a corresponding text sentence, and then the double large language model reasoning is used to fill in the blank content of the table, that is, one language model is used to generate prompt words for filling in the table based on the content of the empty table, and the other language model is used to generate prompt words for filling in the table based on the content of the empty table. A language model generates text with an md table based on prompt words, table contents to be filled in, and search contents, and constructs a text language structure combing method based on regular expressions. In the text feedback given by the large language model, the main contents of the table are correctly located and screened out, and a table in md format is generated. The md table is converted into json data for visualization of json data on a computer or mobile phone, and an editable table is generated to facilitate manual modification and calibration of data and table formats. After manual verification, modification, and supplementation, the data is saved and printed, and then the table content and information repeat the above process and are stored in the information storage unit 400.

[0047] In the present invention, the information preprocessing unit 100 includes an information acquisition module 110, an information extraction module 120 and an information editing module 130, wherein: the information acquisition module 110 is used to acquire the table to be processed; the information extraction module 120 extracts corresponding text and table format from the table to be processed according to the obtained table to be processed using a text recognition model, and then generates json data, and the information extraction module 120 is connected to the information acquisition module 110; the information editing module 130 is used to visualize the obtained json data to obtain an editable table, and the information editing module 130 is connected to the information extraction module 120.

[0048] In addition, the information preprocessing unit 100 also includes an information conversion module 140 and a first communication module 150, wherein: the information conversion module 140 is used to convert json format table data into md format data, and the information conversion module 140 is connected to the information editing module 130; the first communication module 150 is used to realize information interaction between the information preprocessing unit 100 and the first data unit 200 and the second data unit 300, and the first communication module 150 is connected to both the information acquisition module 110 and the information conversion module 140.

[0049] In this embodiment, it should be noted that the forms to be processed acquired by the information acquisition module 110 include but are not limited to uploading paper forms by taking photos or uploading electronic forms by taking screenshots, and then the acquired forms to be processed are transmitted to the information extraction module 120. The information extraction module 120 extracts the corresponding text and table format from the forms to be processed and generates json data. The generated json data is transmitted to the information editing module 130. The information editing module 130 performs data visualization and generates an editable table. The information conversion module 140 converts the json format table data into md format data and uploads it through the first communication module 150.

[0050] In the present invention, the first data unit 200 includes a second communication module 210, an information cutting module 220 and an information conversion module 230, wherein: the second communication module 210 is used to realize information interaction between the first data unit 200 and the information preprocessing unit 100, the second data unit 300 and the information storage unit 400; the information cutting module 220 is used to cut the md format table containing data into small tables according to a fixed size and retain the table header, and the information cutting module 220 is connected to the second communication module 210; the information conversion module 230 is used to convert the cut small tables into vector data and store them in the information storage unit 400, and the information conversion module 230 is connected to both the second communication module 210 and the information cutting module 220.

[0051] In this embodiment, it should be noted that the information of the information preprocessing unit 100 can be received through the second communication module 210, and the information segmentation module 220 cuts the data in the md format table containing data into small tables of a fixed size and retains the table header. The information conversion module 230 can convert the cut small tables into vector data and store them in the information storage unit 400.

[0052] In the present invention, the second data unit 300 includes a third communication module 310, a screening and conversion module 320 and an information filling module 330, wherein: the third communication module 310 is used to realize the connection between the second data unit 300 and the information preprocessing unit 100, the first data unit 200 and the information storage unit 400; the screening and conversion module 320 is used to convert the header and directory information in the empty table containing no data into vector data, and then match the vector data in the vector database, screen the vector data with the highest similarity, and convert the vector data format into a corresponding text statement, and the screening and conversion module 320 is connected to the third communication module 310; the information filling module 330 is used to fill in the blank content of the table, and the information filling module 330 is connected to the screening and conversion module 320.

[0053] The process of filling in the blanks in the table is as follows:

[0054] The large language model will generate prompt words for filling in the form based on the content of the empty form, and then generate text with an md table based on the prompt words, the content of the form to be filled in, and the search content.

[0055] In addition, the second data unit 300 also includes a format generation module 340 and a format adjustment module 350; the format generation module 340 is used to locate and filter out the content of the table, generate a table in md format, and the format generation module 340 is connected to the information filling module 330; the format adjustment module 350 is used to convert the md table into json data, and visually generate an editable table, and the format adjustment module 350 is connected to both the third communication module 310 and the format generation module 340.

[0056] In this embodiment, it should be noted that the header and directory information in the empty table are converted into vector data through embedding technology, and the vector data in the vector database is matched according to the Yuxuan similarity algorithm, the vector data with the highest similarity is screened, and the vector data format is converted into a corresponding text sentence; then, the double large language model reasoning is used to fill in the blanks in the table, and then a text language structure combing method based on regular expressions is constructed. In the text feedback given by the large language model, the main content of the table is correctly located and screened out, and a table in md format is generated, wherein the md table is converted into json data for realizing the visualization of json data on a computer or mobile phone, and an editable table is generated to facilitate manual modification and calibration of data and table formats, and finally, after manual verification, modification, and supplementation, the data is saved and printed, and then the table content and information repeat the above process and are stored in the information storage unit 400.

[0057] The automatic retrieval and automatic form filling system for non-fixed format table information provided by the present invention reduces the workload of form filling staff, improves work efficiency, reduces the error rate, improves the level of digitalization of agriculture and rural areas, and improves farmers' lives.

[0058] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0059] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A system for automatically retrieving and filling in information in non-fixed formats. It is characterized in that The invention comprises an information preprocessing unit (100), a first data unit (200), a second data unit (300) and an information storage unit (400), wherein: The information preprocessing unit (100) is used to obtain table information to be processed, extract text and table format in the table, generate json data, and convert the obtained json data into md format data after visual editing; The first data unit (200) is used to cut the data of the md format table into small tables according to a fixed size and retain the table header, and convert the cut small tables into vector data, and store them in the information storage unit (400). The first data unit (200) is connected to the information preprocessing unit (100); The second data unit (300) is used to convert the header and directory information in the empty table into vector data, then match the vector data in the vector database, screen the vector data with the highest similarity, and convert the vector data format into a corresponding text sentence, and to locate and screen the content of the table, generate a table in md format, and convert the md table into json data, and then visually generate an editable table. The second data unit (300) is connected to both the information preprocessing unit (100) and the first data unit (200); The information storage unit (400) is used to store the received information, and the information storage unit (400) is connected to both the first data unit (200) and the second data unit (300).

2. A system for automatically retrieving and filling in information in a non-fixed format table according to claim 1, It is characterized in that The information preprocessing unit (100) comprises an information acquisition module (110), an information extraction module (120) and an information editing module (130), wherein: The information acquisition module (110) is used to acquire a table to be processed; The information extraction module (120) extracts corresponding text and table format from the table to be processed based on the obtained table to be processed and uses a text recognition model, and then generates JSON data. The information extraction module (120) is connected to the information acquisition module (110); The information editing module (130) is used to visualize the obtained JSON data to obtain an editable table, and the information editing module (130) is connected to the information extraction module (120).

3. A system for automatically retrieving and filling in information in a non-fixed format table according to claim 2, It is characterized in that The information preprocessing unit (100) further comprises an information conversion module (140) and a first communication module (150), wherein: The information conversion module (140) is used to convert the json format table data into md format data, and the information conversion module (140) is connected to the information editing module (130); The first communication module (150) is used to realize information interaction between the information preprocessing unit (100) and the first data unit (200) and the second data unit (300), and the first communication module (150) is connected to both the information acquisition module (110) and the information conversion module (140).

4. A system for automatically retrieving and filling in information in a non-fixed format table according to claim 1, It is characterized in that The first data unit (200) comprises a second communication module (210), an information segmentation module (220) and an information conversion module (230), wherein: The second communication module (210) is used to realize information interaction between the first data unit (200) and the information preprocessing unit (100), the second data unit (300) and the information storage unit (400); The information cutting module (220) is used to cut the md format table containing data into small tables according to a fixed size and retain the table header, and the information cutting module (220) is connected to the second communication module (210); The information conversion module (230) is used to convert the cut small table into vector data and store it in the information storage unit (400). The information conversion module (230) is connected to both the second communication module (210) and the information cutting module (220).

5. A system for automatically retrieving and filling in information in a non-fixed format table according to claim 1, It is characterized in that The second data unit (300) comprises a third communication module (310), a screening and conversion module (320) and an information filling module (330), wherein: The third communication module (310) is used to realize the connection between the second data unit (300) and the information preprocessing unit (100), the first data unit (200) and the information storage unit (400); The screening and conversion module (320) is used to convert the header and directory information in the empty table without data into vector data, and then match the vector data in the vector database, screen the vector data with the highest similarity, and convert the vector data format into a corresponding text sentence. The screening and conversion module (320) is connected to the third communication module (310); The information filling module (330) is used to fill in the blank contents of the table, and the information filling module (330) is connected to the screening conversion module (320).

6. A system for automatically retrieving and filling in information in a non-fixed format table according to claim 5, It is characterized in that The process of filling the blanks in the table is as follows: The large language model will generate prompt words for filling in the form based on the content of the empty form, and then generate text with an md table based on the prompt words, the content of the form to be filled in, and the search content.

7. A system for automatically retrieving and filling in information in a non-fixed format table according to claim 5, It is characterized in that The second data unit (300) further includes a format generation module (340) and a format adjustment module (350); The format generation module (340) is used to locate and filter out the content of the table and generate a table in the md format. The format generation module (340) is connected to the information filling module (330); The format adjustment module (350) is used to convert the md table into json data and visually generate an editable table. The format adjustment module (350) is connected to both the third communication module (310) and the format generation module (340).

Citation Information

Cited By

  • Automatic report filling method based on large language model

    CN119578383A

  • Automatic filling method and system for local table

    CN120106026A

  • A method and system for automatic filling of local forms

    CN120106026B