KR20180080408A - Structured data and unstructured data extraction system and method - Google Patents

Abstract

Disclosed are a structured and unstructured data extraction system and a method thereof. According to an embodiment of the present invention, the structured and unstructured data extraction system includes: a data imaging module for generating image data to image financial data requiring a handwriting input; and a data extraction module for classifying a template after extracting full text data from the image data, extracting template data based on the classified template, and converting the extracted template data into verification data. Accordingly, the present invention can increase accuracy while maintaining work productivity.

Description

[0002] Structured data and unstructured data extraction system and method [

The present invention relates to a system for and a method for extracting structured and unstructured data.

In general, the asset management company performs the task of finding information corresponding to the base price in the financial data (for example, the operation instructions) periodically updated by utilizing the human resources and registering the data as the operating instruction data.

However, since various financial data have various unstructured templates, misinformation frequently occurs, and a large amount of market instructions are provided as PDF files, so that there is a high possibility of omission. In addition, the human error increases the compensation liability insurance premium and lowers the reliability of the customer.

In addition, in order to prevent such an accident, there is a structural limitation that can be done by the second or the third departmental eye checking. Therefore, there is a need for measures to compensate for such human errors.

Korean Patent Laid-Open No. 10-2008-0066729 (published on July 16, 2008) - Online data verification of list data

An object of the present invention is to provide a system and a method for extracting data of high accuracy from data for handwritten input, which is various types of regular / irregular contents, in order to verify consistency of handwritten input data.

Other objects of the present invention will become more apparent through the following preferred embodiments.

According to an aspect of the present invention, there is provided a data imaging module for generating image data in which financial data requiring handwriting input is imaged; And a data extraction module for extracting the professional data from the image data, classifying the template, and extracting template data based on the classified template and converting the template data into verification data.

Wherein the data extraction module comprises: a professional data extracting unit for extracting specialized data corresponding to text by using an optical character reading method on the image data; A template classifying unit classifying the expert data as corresponding to a template corresponding to the template classification keyword table when at least one template keyword registered in the predefined template classification keyword table is included in the expert data; A template data extracting unit for extracting template data from the specialized data based on a data extraction definition table set for the classified template; And a data parsing unit parsing and refining the template data to generate the verification data.

And a template registration module for registering the template for each financial data item, wherein the template registration module comprises: a template classification keyword collection unit for collecting and classifying keywords for classifying the templates to generate the template classification keyword table; And a data extraction definition unit for generating a data extraction definition table for setting a reference keyword and an n-th data extraction range (where n is a natural number equal to or greater than 1) for extracting the template data from the classified table.

The nth data extraction range skips the n1 line by using the reference keyword as a search position and extracts a word corresponding to a position shifted by n2 words with the template data for the specialized data classified by row and column units Lt; / RTI >

Wherein the template data extracting unit extracts the template data by a text processing method for extracting data by moving a position through a line skip and a word movement based on a reference keyword, and according to the text processing method, T () (N) indicates a line-by-line skip, and W (n) indicates a method of extracting data from a word (N) represents the extracted data attribute information, X (n) represents whether to extract single or multiple keys, and V (n) S (n) represents a processing option when there are one or more identical modifiers in the same file.

Or the template data extracting unit extracts the template data by an extension processing method for extracting data by moving a position through a reference line skip and a word movement when a keyword value is satisfied from the second or the second to the nth after the first keyword is satisfied, According to the extension processing method, E () indicates a method of moving to the next set position and extracting data when the second keyword is satisfied after the first keyword is satisfied. The start position keyword value, the end position keyword value, and the skip keyword are set and L N (n), N (n) .F (n) may represent extracted data attribute information.

The financial data may include at least one of a handwritten operation instruction data received via e-mail or a messenger, a Bloomberg terminal screen, and a market instruction data of a large capacity PDF file type, and the data imaging module transmits the handwriting instruction data and the market instruction data The image of the Bloomberg terminal screen can be converted into an image by using the S2I engine of the screen capture method.

According to another aspect of the present invention, there is provided a method of generating image data, the method comprising: generating image data in which financial data requiring handwriting input is imaged; Extracting specialized data from the image data; Classifying a template corresponding to the specialized data; Extracting template data based on the classified template; And converting the template data into data for verification.

The step of classifying the template may include the steps of: when the at least one template keyword registered in the predefined template classification keyword table is included in the specialized data, the specialized data corresponds to the template corresponding to the template classification keyword table Can be classified.

The step of extracting the template data may extract the template data from the specialized data based on the data extraction definition table set for the classified template.

Collecting a keyword for classifying the template and creating a table to generate the template classification keyword table; A step of generating a data extraction definition table for setting a reference keyword and an n-th data extraction range (where n is a natural number equal to or greater than 1) for extracting the template data from the classified table may be preceded.

The step of extracting the template data may include a text processing method for extracting data by moving a position through a line skip and a word movement based on a reference keyword or a text processing method for extracting data from a reference line The template data can be extracted by an extension processing method for extracting data by moving a position through skipping and word movement.

Other aspects, features, and advantages will become apparent from the following drawings, claims, and detailed description of the invention.

According to the embodiment of the present invention, in order to verify consistency of handwritten input data, data is detected with high accuracy in handwritten input data, which is various regular / irregular contents, and the consistency between detected data and handwritten input data is verified, Thereby maintaining business productivity and improving accuracy.

In addition, by strengthening the data verification function, it is possible to prevent errors in the operation instruction input, greatly reduce the error in the reference price, reduce the professional liability insurance premium due to the reduction in the reference price error, and simply check the error data in the verification report It reduces workload and increases the concentration of other tasks as work productivity increases.

1 is a schematic block diagram of a handwriting input data consistency verification system according to an embodiment of the present invention;
2 is a block diagram of a configuration of a data extraction module,
3 is a block diagram of a configuration verification module,
4 is a block diagram showing the configuration of the template registration module,
5 shows an example of generation of a template classification keyword table,
6 shows an example of generation of a template data extraction definition table,
7 is a flowchart of a handwritten input data consistency verification method,
8 is a flowchart showing a process of extracting template data from the data extraction module,
9 is a table showing the types of data extraction schemes,
Fig. 10 shows a definition table of a text processing method,
11 shows an example of a text processing method,
12 shows a definition table of the extension processing method,
13 is an example of an extended processing method,
Figs. 14 to 16 are application examples of template data extraction; Fig.

While the invention is susceptible to various modifications and alternative forms, specific embodiments thereof are shown by way of example in the drawings and will herein be described in detail. It is to be understood, however, that the invention is not to be limited to the specific embodiments, but includes all modifications, equivalents, and alternatives falling within the spirit and scope of the invention.

It is to be understood that when an element is referred to as being "connected" or "connected" to another element, it may be directly connected or connected to the other element, . On the other hand, when an element is referred to as being "directly connected" or "directly connected" to another element, it should be understood that there are no other elements in between.

The terms first, second, etc. may be used to describe various components, but the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from another.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The singular expressions include plural expressions unless the context clearly dictates otherwise. In this specification, the terms "comprises" or "having" and the like refer to the presence of stated features, integers, steps, operations, elements, components, or combinations thereof, But do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, or combinations thereof.

It is to be understood that the components of the embodiments described with reference to the drawings are not limited to the embodiments and may be embodied in other embodiments without departing from the spirit of the invention. It is to be understood that although the description is omitted, multiple embodiments may be implemented again in one integrated embodiment.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Reference will now be made in detail to the preferred embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like reference numerals refer to the like elements throughout. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the following description, well-known functions or constructions are not described in detail since they would obscure the invention in unnecessary detail.

FIG. 1 is a schematic block diagram of a handwriting input data consistency verification system according to an embodiment of the present invention, FIG. 2 is a block diagram of a data extraction module, FIG. 3 is a block diagram of a consistency verification module, 4 is a block diagram of a template registration module, FIG. 5 is an example of generating a template classification keyword table, FIG. 6 is an example of generating a template data extraction definition table, and FIG. 7 is a flowchart of a handwriting input data consistency verification method.

The handwritten input data consistency verification system 1 according to an embodiment of the present invention compares human handwritten data with handwritten input data to verify the consistency by automatically detecting and refining data from various types of handwritten input data, Can be prevented.

1, a handwritten input data consistency verification system 1 according to the present embodiment includes a structured and unstructured data extraction system including a data imaging module 10 and a data extraction module 20. [ The stereotyped and unstructured data extraction system may further include a template registration module 40. In addition, the handwriting input data consistency verification system 1 may include a consistency verification module 30 according to an embodiment.

The data imaging module 10 implements the financial data required for handwriting input.

The financial data may be one of the handwriting instruction data received via e-mail or IM, data on the Bloomberg terminal screen, and market instruction data for extracting a plurality of templates and data from a large capacity PDF file. The handwritten operation instruction data is composed of office document type such as PDF, EXCEL file, or is composed of reporting tool, XML format, e-mail contents, or having various templates such as bonds, derivatives, ELS / ELW, cash assets, beneficiary certificates, And may be data in a file format. The data on the Bloomberg terminal screen can be various data such as stock beneficiary certificates, bonds, futures, etc., which are output on the screen from the Bloomberg terminal. The market instruction data has a large PDF file format posted on the stock exchange, and a plurality of templates can be extracted from one file, and various forms such as KOSPI, KOSDAQ, and KONEX can be registered.

The data imaging module 10 may include a P2I engine that prints and manages the hand operation instruction data (P2I, Print to Image). Alternatively, the data imaging module 10 may include an S2I engine for capturing a Bloomberg terminal screen (S2I). The scan of the printed matter by the P2I engine and the screen capture by the S2I engine are obvious to those skilled in the art, and a detailed description thereof will be omitted.

The data extraction module 20 extracts only characters (text) from the image data imaged by the data imaging module 10 through professional recognition and stores them in a database. Then, the data is parsed and refined according to a predetermined data processing algorithm and converted into verification data for verification of consistency.

Referring to FIG. 2, the data extraction module 20 includes a specialized data extraction unit 22, a template classification unit 24, a template data extraction unit 26, and a data parsing unit 28.

The professional data extracting unit 22 extracts professional data from the data (image data) imaged by the data imaging module 10. The specialized data is data generated by recognizing all the portions corresponding to characters (text) in the image data by the optical character recognition (OCR) method.

The extracted professional data can be stored in a separate database. The specialized data may be stored in a spreadsheet form comprising a plurality of rows and columns.

The template classifying unit 24 analyzes the extracted professional data and classifies the template corresponding to the corresponding data. The template classification keyword table 43 previously defined by the template registration module 40 is used for the template classification. When at least one template keyword registered in the template classification keyword table 43 for a specific template is included in all of the specialized data extracted by the specialized data extraction unit 22, the corresponding specialized data is classified as corresponding to a specific template .

The template data extracting unit 26 extracts template data that needs parsing and refinement from the specialized data based on the data extraction definition table 45 set corresponding to the classified template. The template data to be extracted can be defined by the template registration module 40 and can be defined to be processable by a text processing method or an extended processing method described later.

The data parsing unit 28 generates verification data in which the template data extracted by the template data extracting unit 26 is parsed. The verification data can be refined so that it can be used for verification of compatibility with handwritten input data. The verification data may be tabulated in the form of a spreadsheet having rows and columns.

The consistency verification module 30 verifies the data consistency by comparing the handwritten input data and the verification data to the financial data for which handwriting input is required. The data consistency means the correspondence between the automatically extracted data for verification and the data of handwritten input data.

The conformance verification module 30 described in the present embodiment is one embodiment. In addition, in a technical field to which the present invention belongs, a common descriptor compares handwritten input data with verification data through a database matching technique, You can also verify.

The results of the data consistency verification can be provided as a report, and it is possible to prevent errors in handwriting registration and improve the reliability of the data based on the result of the data consistency verification.

Referring to FIG. 3, the consistency verification module 30 according to the present embodiment includes a data receiving unit 32, a consistency verification unit 34, and a reporting unit 36.

The data receiving unit 32 receives the verification data generated by the data extraction module 20. The data receiving unit 32 receives the handwriting input data to be verified.

The consistency verification unit 34 compares verification data and handwritten input data with each other and determines the correspondence.

In this case, it is a matter of course that the data imaging process and the data extraction process can be performed on the handwritten input data as described above. In this case, the template registration module 40 may separately register a template for the handwritten input data, and notify the position of the data for verifying the handwritten input data so that only necessary character data is parsed and refined. As a result, it can be tabulated in the form of a spreadsheet having rows and columns as handwritten input data, and can be compared with the verification table in rows and columns.

The verification result by the consistency verification unit 34 can be output by the reporting unit 36 so that the user can confirm it. The reporting unit 36 may output the verification result to the screen or may be printed using the printer.

When the verification result is reported, data corresponding to inconsistent data is separately displayed, thereby enabling the user to easily confirm unmatched data. To do this, one or more of font, color, and size may be set differently for inconsistent data, or highlighted or underlined. Alternatively, only data that do not match can be extracted and displayed collectively.

It is sufficient for the verifier to cross-check only the error data in the verification report, which reduces the amount of simple verification work.

The template registration module 40 extracts a character from the imaged data in the data extraction module 20 and then selects only necessary characters and converts it into verification data. In the process of converting the data into verification data, In order to distinguish the bar and select only the correct data, it is necessary to register templates for each financial data.

The template to be registered has a template classification keyword for distinguishing it from other templates, and data to be extracted for the template can be defined according to a specific rule.

The template registration module 40 can register various unstructured templates, and can also manage the version of each template. It is also possible to prevent duplicate registration at the time of template registration.

It is possible to define details of data extraction in various formats, and code and character substitution for each item to be extracted is also possible. Details can be set for applying the formulas for each item to be extracted, and detailed definitions can be made for the data base system and the data consistency check.

Referring to FIG. 4, the template registration module 40 includes a template classification keyword collection unit 42 and a data extraction definition unit 44.

The template classification keyword collecting unit 42 analyzes the professional data extracted by the professional data extracting unit 22 and collects keywords for classifying the templates corresponding to the data to generate a template classification keyword table 43. [

For the handwritten operating instructions 50 as illustrated in FIG. 5, keywords for distinguishing from other financial data may be tabulated with template classification keywords and stored in a database. For example, in the handwritten operation instruction sheet 50 shown in FIG. 5, "① stock price-linked securities / stock warrants evaluation price notification, ② Samsung securities stock management team, ③ name of securities, ④ base asset price, A word or phrase used only in the corresponding financial data may be defined as a template classification keyword and registered in the template classification keyword table 43 for the corresponding template.

The data extraction defining unit 44 may set a reference keyword for extracting data from the classified template and an n-th data extracting range (n is a natural number equal to or greater than 1), and may store the data in a database in the form of data extraction definition rules. For example, the 'deposit deal code' for the handwritten operating instructions 50 illustrated in FIG. 6 is set as a reference keyword for data extraction, and the first extracted data range is set to 1 Character data up to the last line after the line skipping, and the second extracted data range is character data from the reference keyword 'deposit deal code' to the last line after skipping one line downward and shifting by three words to the right.

FIG. 8 is a flowchart showing a process of extracting template data from the data extraction module, FIG. 9 is a table showing the types of data extraction methods, FIG. 10 is a definition table of a text processing method, Fig. 12 is a definition table of the extension processing method, and Fig. 13 is an example of the extension processing method.

After the template classification by the template classifying unit 24 is completed, the template data extracting unit 26 extracts template data from the specialized data. To do this, follow the procedure below.

First, based on the data extraction definition table 45 defined in the template registered by the template registration module 40, it is checked whether the reference keyword is in the specialized data (step S105).

If the reference keyword is found, the position of the reference keyword is set as the basic position of the search position (step S110).

(Basic position) is skipped by n1 lines based on the first extracted data range defined in the data extraction definition table 45 (step S115). Skip means moving in the downward direction by n1 lines. n1 is an integer of 0 or more.

If necessary, the position can be moved to n2 words (step S120). Position movement means moving by n2 words in the right direction. n2 is an integer of 0 or more.

When the line skip and the word movement are completed, a word corresponding to the moved position is extracted (step S125). The extracted word is the first extracted data among the template data for the corresponding template.

If a second or more extracted data range is defined in the data extraction definition table 45, steps S115 to S125 are repeated.

The extracted word (s) are parsed and refined as template data according to predetermined parsing / refining rules to function as verification data.

In this embodiment, the data extraction method such as line skip and position shift in steps S115 and S120 corresponds to the text processing method among the data processing methods shown in Fig.

T (s_kw ^ e_kw, skip_kw1 ^ skip_kw2 ^ ..., pick_kw1 ^ pick_kw2 ^ ...), L (s_kw1 ^ pick_kw2 ^ ...) is a method of extracting data by moving a position through line skip and word (n), W (n), C (n) or N (n) or F (n) And the like.

Referring to FIG. 10, explanations and notation examples of each notation format for the text processing method are listed in a table.

T () denotes a method (text processing method) for extracting data by moving by a keyword reference line and a word position, and a start position keyword value, an end position keyword value, a skip keyword, an extraction reference keyword and the like are set. C (n), N (n), N (n) .F (n) represent extracted data attribute information, and L (n) X (n) represents whether or not to extract single or multiple keys, V (n) represents a valid data extraction process and the number of extraction processes, and S (n) represents a processing option when there are one or more modifiers in the same file.

For example, referring to FIG. 11, when the data extraction definition for the handwritten operation instruction is defined as T ("deposit deal code"), (2) L (1) The process is as follows.

① In case of "Deposit Deal Code" word, set the search position ② Skip the search line by one line ③ Move the position from the search position to the third word and extract the word, "Data of the item can be extracted.

In addition to the text processing method, there may be an extension processing method. In the extended processing method, when the keyword value is satisfied from the second or the second to the n-th after the first keyword is satisfied, the position is moved through the reference line skip and word movement, . L (n), W (n), C (n) or N (n) or N (n) .F (n) And the like.

Referring to FIG. 12, the case of the marketplace parsing process is exemplified among the extended processing methods.

If the second keyword is satisfied after the first keyword is satisfied, E () indicates a method of moving to the next set position and extracting data. A start position keyword value, an end position keyword value, a skip keyword, and the like are set. C (n), N (n), and N (n) .F (n) represent extracted data attribute information, and L (n) represents a line-by-line skip.

Referring to FIG. 13, an example of an extended processing method is shown.

If the data extraction definition is defined as ① E ("A011930", "supply contract", "contract period", "start date"), ② L (0), ③ W (1) Respectively.

① In case of continuous words satisfying the conditions such as "A011930", "Supply contract", "Contract period", "Start date", etc., set the search position. ② Skips 0 line at search position. ③ Move the position from the search position to the first word to extract the word. In this example, "2013-09-23" will be extracted.

Figs. 14 to 16 are application examples of template data extraction. The red square represents the data extraction item.

FIG. 14A is an example of extracting the entire data of the file format specific item and extracting only the first item data, FIG. 14B is an example of extracting data by line skipping processing after extracting the file format first item and extracting only the last data (C) shows various examples of data extraction by data extraction by a file format item and data extraction by a subtotal partial skip process, and (d) is an example of data extraction by item at various positions of a file format.

FIG. 15 (a) is a data extraction range for each Bloomberg stock-beneficiary screen item, (b) is a data extraction range for each Bloomberg bond screen item, (c) is a data extraction range for each Bloomberg gift screen item, Shows the data extraction range for each Bloomberg end screen item.

FIG. 16A is a table format disclosure range extraction within a market instruction, FIG. 16B is a range extraction format of a number format disclosure within a market instruction, FIG. 16C is a table format data extraction range in an email body, d) represents the tabular data extraction range that is repeated in the body of the e-mail.

The handwriting input data consistency verification method and / or data extraction method according to the present invention can be implemented as a computer-readable code on a computer-readable recording medium. The computer-readable recording medium includes all kinds of recording media storing data that can be decoded by a computer system. For example, it may be a ROM (Read Only Memory), a RAM (Random Access Memory), a magnetic tape, a magnetic disk, a flash memory, an optical data storage device, or the like. In addition, the computer-readable recording medium may be distributed and executed in a computer system connected to a computer network, and may be stored and executed as a code readable in a distributed manner.

According to the method and system for verifying the handwritten input data consistency according to the present embodiment, it is possible to reduce the error of the reference value, reduce the professional liability insurance premium, and reduce the amount of simple verification work, thereby improving work productivity.

In addition, it provides formal / irregular format registration functions in various formats such as EXCEL, PDF, E-MAIL, Bloomberg screen, text document, and reporting tool printout. Form registration can be classified into one registration per template, Function.

Detailed function definition for extracting data desired by the user is provided, and continuous data extraction is possible for a file composed of a plurality of pages without any additional processing. It is also possible to skip unnecessary data or lines when extracting consecutive data.

The extracted data can be transformed or mapped to various formats, and detailed function definitions for data type conversion are provided. If the data is numeric, it also provides a detailed setting function that allows the application of a separate formula.

It is also possible to extract data from the market instruction format (large-capacity PDF file) posted on the stock exchange and to provide simultaneous extraction of multiple templates from the same file.

It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit or scope of the invention as defined in the appended claims. It will be understood that the invention may be varied and varied without departing from the scope of the invention.

1: Handwritten input data consistency verification system
10: Data imaging module 20: Data extraction module
22: professional data extracting unit 24: template classification unit
26: Template data extracting unit 28: Data parsing unit
30: consistency verification module 32: data reception unit
34: consistency verification unit 36:
40: template registration module 42: template classification keyword collection unit
44: Data extraction definition section

Claims (13)

A data imaging module for generating image data in which financial data requiring manual input is imaged; And
And a data extraction module for extracting specialized data from the image data, classifying the template, and extracting template data based on the classified template and converting the template data into verification data.
The method according to claim 1,
Wherein the data extraction module comprises:
A specialized data extracting unit for extracting specialized data corresponding to a text by using an optical character reading system for the image data;
A template classifying unit classifying the expert data as corresponding to a template corresponding to the template classification keyword table when at least one template keyword registered in the predefined template classification keyword table is included in the expert data;
A template data extracting unit for extracting template data from the specialized data based on a data extraction definition table set for the classified template;
And a data parsing unit parsing and refining the template data to generate the verification data.
3. The method of claim 2,
And a template registration module for registering the template for each financial data,
The template registration module may include a template classification keyword collection unit for collecting and classifying keywords for classifying a template to generate the template classification keyword table; And a data extraction definition unit for generating a data extraction definition table for setting a reference keyword and an nth data extraction range (where n is a natural number equal to or greater than 1) for extracting the template data from the classified table. And unstructured data extraction systems.
The method of claim 3,
The nth data extraction range skips the n1 line by using the reference keyword as a search position and extracts a word corresponding to a position shifted by n2 words with the template data for the specialized data classified by row and column units Wherein the data extracting rule is a data extracting rule for extracting the data from the extracted data.
3. The method of claim 2,
Wherein the template data extracting unit extracts the template data by a text processing method for extracting data by moving a position through line skipping and word movement based on a reference keyword,
The start position keyword value, the end position keyword value, the skip keyword, and the extraction criterion keyword are set, and L (n (n)) is set according to the text processing method. (N) represents extracted data attribute information, X (n) represents a line-by-line skip, W (n) (N) represents a processing option when there are one or more identical modifiers in the same file, and V (n) represents a valid data extraction process and the number of extraction processes, And unstructured data extraction systems.
3. The method of claim 2,
Wherein the template data extracting unit extracts the template data by an extension processing method for extracting data by moving a position through a reference line skip and a word movement when a keyword value is satisfied for a second,
According to the extension processing method, E () indicates a method of moving to the next set position and extracting data if the second keyword is satisfied after the first keyword is satisfied, and a start position keyword value, an end position keyword value and a skip keyword are set and L (n) represents a line-by-line skip, W (n) represents word-by-word movement, and C (n), N (n), and N (n) A systematic and unstructured data extraction system.
The method according to claim 1,
The financial data may include one or more of handwriting instruction data received by mail or messenger, Bloomberg terminal screen, market indication data of a large capacity PDF file type,
Wherein the data imaging module implements the handwriting instruction data and the market instruction data by using a P2I engine of a printing and scanning type and the Bloomberg terminal screen is imaged by using a S2I engine of a screen capture method. And unstructured data extraction systems.
Generating image data in which financial data requiring handwriting input is imaged;
Extracting specialized data from the image data;
Classifying a template corresponding to the specialized data;
Extracting template data based on the classified template;
Converting the template data into verification data
And extracting the input data.
9. The method of claim 8,
The step of classifying the template may include the steps of: when the at least one template keyword registered in the predefined template classification keyword table is included in the specialized data, the specialized data corresponds to the template corresponding to the template classification keyword table And extracting the extracted data.
10. The method of claim 9,
Wherein the extracting of the template data extracts template data from the specialized data based on a data extraction definition table set for the classified template.
11. The method of claim 10,
Collecting a keyword for classifying the template and creating a table to generate the template classification keyword table;
And generating a data extraction definition table for setting a reference keyword and an n-th data extraction range (where n is a natural number equal to or greater than 1) for extracting the template data from the classified table. Data extraction method.
11. The method of claim 10,
The step of extracting the template data may include a text processing method for extracting data by moving a position through a line skip and a word movement based on a reference keyword or a text processing method for extracting data from a reference line Wherein the template data is extracted by an extension processing method for extracting data by moving a position through a skip and a word movement.
9. The method of claim 8,
The financial data may include one or more of handwriting instruction data received by mail or messenger, Bloomberg terminal screen, market indication data of a large capacity PDF file type,
Wherein the image data generation step images the handwriting instruction data and the market instruction data using a print and scan type P2I engine and the Bloomberg terminal screen is imaged using a screen capture type S2I engine Method for extracting stereotyped and unstructured data.

Images