JPH11219394A - Automatic various financial chart input device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a device as a personal computer system for realizing the automatic check of account titles, the automatic correction of erroneously read characters and automatic representative account title summarization by a unified standard, etc., by utilizing the basic program of type OCR, substantially improving read accuracy, substantially shortening input time by a simple operation, coping with the nationwide network system (distributed processing) of a client/server type and being flexibly adopted to mass processings as well. SOLUTION: This device is constituted of a means for inputting type-printed various financial charts by utilizing a scanner 2 connected to a personal computer 1, digitizing image information and automatically checking and correcting the account titles of a digital result by utilizing an account title dictionary 8 and an erroneous read dictionary 9, etc., the means for automatically checking the balance of a money amount and the means for performing summarization to a representative account title.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] The present invention relates to a financial statement (balance sheet, profit and loss statement, profit appropriation plan,
After inputting image information using a scanner without manually inputting selling, general and administrative expenses, manufacturing cost statements, etc.), the characters in the image information are automatically recognized and digital characters are highly accurate. The present invention relates to an automatic financial statement input device for substituting with a financial statement.

[0002]

2. Description of the Related Art Conventional methods for inputting financial statements include a method of manually inputting all the contents from a computer terminal or the like, and a method of displaying a predicted account on a computer terminal and inputting only a numeral part. It was mainstream. In recent years, with the development of computers, some companies have also introduced a handwritten OCR method in which the amounts of financial statements are transcribed and input into a dedicated input sheet prepared while considering the aggregation. The type-printing character OCR input method is a general-purpose reader for general documents, newspapers, etc., though it is in the stage of spread.
Furthermore, when used as an automatic financial statement reader, it was necessary to check all the characters of the input result as the original form, and there were many misread characters and the operation was complicated. There are various accounts in the financial statements prepared by each company, and when each financial institution diagnoses and analyzes the financial contents of each company, it is necessary to consolidate them into a unified representative account. The actual situation of the aggregation was that humans had to judge and determine the subject of the aggregation each time. Aggregation took time. In many cases, the aggregation was different for each person, and because the unified aggregation standard could not be realized, the diagnosis results of the financial contents were often different even for the same company. In addition, because it is a single-function OCR machine, it is necessary to prepare many units for large-scale processing, and networking requires a special-order communication device, and the processing speed is slow and expensive. Was not.

[0003]

From these facts, it is possible to input financial statements more accurately, in a shorter time and more easily.
There is an increasing need to solve such problems at lower cost. In particular, financial institutions have required thorough examination standards, including early corrective measures in the notification of the Ministry of Finance since the bursting of the bubble, and appropriate loans in line with the actual conditions of the company. As a result, the number of financial statements entered by financial institutions has increased sharply more than before. Accordingly, the need for innovative input devices is increasing, but has not yet been realized. To solve these problems, type O
Using the basic program of CR, automatic check of accounts, automatic correction of misread characters, automatic integration of representative accounts based on unified standards, etc. are realized, greatly improving reading accuracy and easy operation. In order to realize that the input time can be greatly reduced. Furthermore, this device can be provided as a personal computer system which is compatible with a client-server type nationwide network system (distributed processing) and can be flexibly adapted to mass processing. Needless to say, the object of the present invention is to fundamentally solve the problems so far by providing an epoch-making method and a device at a significant cost reduction.

[0004]

The present invention provides such an apparatus as a means for solving these problems. This device is a means for inputting the printed financial statements (balance sheet, profit and loss statement, profit appropriation plan, selling, general and administrative expenses, manufacturing cost statement, etc.) with a scanner as an image and storing the input results. There is. The input image is managed on a page-by-page basis and digitized through a program that uses the characteristics of financial statements to replace it with digital characters. The digitized result is a character string and numerical value (money amount) representing an account.
There is a means for checking the matching with an account dictionary for automatically checking whether the account has been converted correctly by a predetermined method. Further, when there is no matching character string in the account dictionary in the matching check, it is determined as a character string containing misread characters. Further, as a means for automatically replacing a misread character with a correct character, there is a misread dictionary replacement function as one of the means, and a program correspondence function for complementing the misread dictionary replacement function as a function for measuring high precision replacement is provided. is there.

[0005] As a result of re-matching to the account dictionary through the above-described processes, if there is no character string that matches the account dictionary, the character string containing misread characters is displayed in red. Next, as a means to more easily correct misread characters, a combination of characters is used from the account character string of the recognized result, and similar account candidates are automatically extracted from the account dictionary. There is a simple method that allows the user to select and input the corresponding account item described in the original form. By these means, there is no need to check all the input results and check them against the original, and it is a means for eliminating the time-consuming correction inputting work while looking at the original.

More specifically, the features of the financial statements (the method of describing the accounting process) include an account item on the left and an amount on the right of the same line. Furthermore, in the case of the balance sheet only, the financial content is one column vertically on the left as assets and one column vertically on the right as liabilities.
Each row in each column has an account on the left and an amount on the right. Major and medium-class items (current assets, fixed assets, etc. belong to the large category, tangible fixed assets, intangible fixed assets, etc.
It belongs to the middle category, and vehicles, tools and equipment, etc. are classified into small categories (detailed accounts). ) Indicates the case where the amount is described on the same line as the amount and the amount is not on the same line, and the total subject name (such as total assets) is described on the last line of the applicable subject group and the amount is described There is. When the same line is viewed sideways, the pattern is
If the term is stated, it is described as the subject / amount / amount. It goes without saying that both the asset section and the debt section have the same pattern. In any case, since the description method is based on the accounting process, this feature is used. The income statement and other tables are similarly categorized into several patterns. The digitization of each table is recognized using such features.

[0007] In particular, the profit and loss statement has one column, regardless of the department, but has the same description. However, in the case of the profit and loss statement, even when only one period is recorded, there are patterns of the item, the amount A, and the amount B. At this time, the amount to the right of the last line item line amount A of the same applicable item group is the same as the case where the amounts of the large class items and the middle class items are described in the amount B of the same line as the large and middle class items. B. In such a case, the program determines which major classification group or middle classification group the program belongs to, and the program automatically recognizes the total value of each subject of the corresponding major classification and middle classification, and the program automatically recognizes the total value of each subject in the predetermined column (the corresponding It moves to the amount A) of the section subject line. Further, when the total amount is not described anywhere in the large / medium classification group in the recognition result of each table, the total amount can be automatically calculated. Similarly, the pattern of item / money A / money B / money C can be processed in the same manner. In this pattern, the amount A is the amount of the small section (detailed account), the amount B is the total amount of the medium section group, and the amount C is automatically determined as the total amount of the large section group. Execute the processing of These are just a few examples, but there are many other ideas based on each pattern based on analysis for each industry and company.

[0008] The financial statements must have mandatory accounts based on accounting. For example, the balance sheet has an asset section and a liability section. The first item of the asset section is a current asset, and there must be a fixed asset in the account section of the asset section. The income statement also has certain principles, all of which follow the accounting method.
From these facts, if an event that does not match the accounting method in the account of the recognition result occurs, the large category and the middle category can be automatically complemented. For example, if there is no fixed asset subject, the program automatically adds a “fixed asset” subject to a predetermined line, and if “fixed XXX”, “fixed XXX”, etc., exist at the head of the detailed subject of the fixed asset, ,
Forcibly modify the "Fixed Assets" subject. Further capital,
When the capital is two consecutive banks, the first capital is used as the capital total. In this way, the automatic balance check of the amount can be performed. And so on.

[0009] The recognition of each of the above tables is processed for each page, and the range of the account item and the range of the amount of money are enclosed by a line as the range of the reading recognition frame. Recognizing whether the field is a subject field or an amount field, a predetermined program is executed and the image data is digitized. The reading recognition frame is not all specified. Several reading frame patterns have been prepared from the original analysis so far. By selecting the pattern that is the most similar from the reading frame patterns displayed on the screen, a frame that automatically covers the account and the amount of money can be created. In each pattern, information for performing digital conversion is stored as a character string in the subject column and a numerical value in the amount column (the recognition accuracy can be further improved by giving information such as characters and numerical values to the program). It is devised that there is no need to indicate whether it is a letter or a number after the frame specification. Since the reading frame pattern is slightly shifted from company to company, fine adjustment is required. The reading frame pattern once registered is stored for each company. Since the financial statements of each company often have the same pattern every year, a convenient mechanism is provided for automatically setting the previous year's frame when entering the next year. Therefore, the operation has been devised so that the operation can be completed in the simplest, more accurate, and shorter time.

As for the amount of money, there is a balance checking means utilizing a description method of accounting processing for checking each financial statement, checking each major classification group and medium classification group, and checking the logic between financial statements. In the balance check, a registered pattern is used to sum up the amounts of each of the large and medium classifications for each group and the amounts of the detailed items in the group to which the group belongs, and check the sum according to the accounting method. In the case of imbalance, there is provided a means for repeating the correction until the balance is achieved for the group displayed in red because there is an erroneous reading of the numerical value, and terminating the input when the balance is achieved.

The specific means of the above-mentioned method for creating a misread dictionary is as follows:
The device of the present invention has the same habit as a character which is apt to be erroneous when stored and read by a human. Therefore, such cases are collected in advance from the literature, and, for convenience, the apparatus according to the present invention is used in combination to input a large number of financial statements of different companies and analyze the resulting large number of misread characters. From the analysis result, it is found that a character tends to be misread as a specific character that is always similar, that is, a character that frequently misreads as the same character. For example, characters such as “in” and “in”, “gold” and “all”, “wood” and “book”, “ko” and “d”, “1” and “1” and “1”, “pressure” and “public”, and “listening and opening” are collected. An example of a specific subject is borrowing ⇒ lessee borrowing ⇒ all borrowing. Therefore, a dictionary is created in the misreading dictionary by incorporating a pattern in which the borrower is replaced with borrowed money and the entire borrowing is replaced with borrowed money. Specifically, the misregistered character may be registered in the misread character field and the character to be replaced may be incorporated in the replacement field. Furthermore, if you want to delete the characters of the recognition result, you can set the target character in the misread field and leave the replacement field blank (for example, I. ⇒ “I.
If you want to replace any character by force, set the symbol to be replaced unconditionally in that place. For example, when the symbol is “*”, if “borrowed * debt ⇒ borrowed money”, the borrowed money, debt, etc. are all borrowed money. The dictionary is provided with a means for constructing a more accurate misreading dictionary by adding such cases as patterns as much as possible.

Further, there is a means for replacing a character string containing a misread character with a correct character on a program (a function corresponding to a program for complementing a misread dictionary as a function for ensuring high precision recognition). A specific example of the correspondence by the program is a case where the misread dictionary cannot cope, or is determined to be more effective than the misread dictionary. For example, "1.
“Current assets” and “(Current assets)” are: This is when only the account items are excluded except for () and (). In addition, the account item generally does not have a number, alphabet, or special character at the beginning, but may partially exist. In this case, for example, “JCB”, “VISA”, “two-year XXX”, etc., the first character of the account dictionary or the like is an alphabet,
As a result of extracting under the condition of numbers, it is an existing account, but if there are some characters, alphabets and special characters at the beginning except for such exceptions, the leading numbers, alphabets and special characters are automatically For the deletion, etc., means corresponding to the program are taken.

With regard to the account dictionary, more than about 80,000 words (character strings), in which the account items input / extracted from various industries and a large amount of corporate / financial statements, etc. are combined into one character string, are registered. ing. In addition, account classification code information etc. was added to each account to supplement the contents of the aggregation destination code and recognition result for aggregation into the representative account based on accounting processing, and to automatically check the character string etc. . Therefore, by using this account item dictionary to determine whether or not the account item character string of the recognition result exists in the dictionary, the conventional manual calibration work can be eliminated. In addition, a program check is performed on the recognition result contents based on the accounting description method, and if a defect is found, a means for automatically complementing the defect is provided.

Means for retrieving a similar account for correcting misread characters more easily is, for example, when the recognition result is “Thunder talk subscription right” (operator compares with original form (telephone subscription right)). If it is determined that the program is not in the account dictionary, it is judged to be misread, so it is displayed in red.), And what characters other than the misread of the corresponding character string in the screen displayed in red are displayed. Specify (drag) the character or characters and use the corresponding character (in this example, "subscribe" or "subscribe")
Search and create a candidate list of accounts including the account from the account dictionary and display it on the screen below the dragged account. As a simpler method, the program automatically searches for a similar account from the account dictionary by using the combination of characters of the recognition result, so the button in the same column as the misread character string displayed in red is used. When pressed, search result candidates are displayed at the bottom. In any case, when a correct character string (character string of an account) exists (in this example, telephone subscription right), there is a function that can be corrected to a correct character string with one touch.

The reading range frame pattern of the financial statements is based on the results of analyzing corporate and financial statements of various types of business, as follows: subject, amount A subject, amount A, other subject, other, amount, other subject, amount A, amount B Subjects,
Amount A, Others, Amount B, Other Subjects, Others, Amount A, Amount B Subject A, Amount A, Subject B, Amount B Subjects, Amount A, Amount B, Amount C Subjects / Others / Amount
Etc. are registered. Further subjects are
The information that the amount, which is a character, is a numerical value is incorporated. With these nine patterns at the center, a pattern selection screen is prepared so that patterns in accordance with the description method of the input original form can be identified at a glance in a short time. There is a means to automatically set numerical value information and give it to the program.

Further, if the digitized account does not exist in the account dictionary and there is an unregistered account determined not to be misread, an account information for easily adding account information to the account dictionary is provided. There are means. One specific example is to search for similar accounts currently registered,
Display similar account information (large division code, middle division code, small division code (subject code), aggregation destination code) on the screen, and refer to it to see the group (large division, middle division) This is a method of registering an account item at a predetermined place in (1) and registering information of the account item. For example, when “cash and deposit” is not registered, a similar account is searched from the character string “cash and deposit” of the digital result. Here, the character string of "cash" which is most present in the account dictionary and is estimated to be in the same group is designated (dragged), and "cash" and "cash major division code, middle division code, small division" A code, an aggregation destination code "is displayed, and an account of" cash and deposit ", a large division code, a medium division code, a small division code (item code), and an aggregation destination code can be easily determined and input. The same account item that is generated after being registered in the account dictionary is correctly processed in a predetermined procedure. The provision of such means enables the use of means that can be accurately and easily input and learning effects, and is designed to be a device with higher recognition accuracy.

The device of the present invention, unlike the conventional OCR device, can easily build a network system using a communication line at low cost, and can transmit input information to the other party in a timely manner. Therefore, it is common sense so far,
This eliminates the need to send to the input center by mail, home delivery, etc., and enables on-site input. In addition, since image information can be transmitted, confirmation, correction, addition, recording of information, etc. can be easily performed at the headquarters side, achieving a paperless environment, thorough information management, improved security, and efficient operation in distributed processing. realizable.

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below in detail with reference to the drawings. FIG. 1 is a diagram showing the overall configuration of an embodiment of the financial statement automatic input device according to the present invention. This device is
It comprises a computer 1, a scanner 2, a display (screen) 3, a keyboard 4, a floppy disk 5, a hard disk 6, a printer 7, other necessary external devices 10, and programs connected thereto. A telephone line cable 12 for constructing a network can be connected to the computer. Of course, the communication device 1
Contains 1 In particular, the hard disk 6 stores an account dictionary 8 and a misread dictionary 9 which are the heart of the present invention.

The original form of the financial statements (balance sheet, profit and loss statement, profit appropriation plan, selling and general administrative expenses, manufacturing cost statement, etc.) to be recognized is set on the scanner 2 and displayed on the display 3 (screen). 4) When a command to start scanning is given from FIG. 4, the scanner starts reading and image input starts. When the original set in the scanner runs out, the scanning ends. After the scan is complete,
Since the image for each page is displayed on the screen in the order in which it was read, the financial statement pattern “FIG. 3” displayed on the screen is selected while comparing it with the image. When a pattern is selected, a frame indicating the reading range is automatically displayed on the image. At this time, since the frame encloses the approximate range, fine adjustment is required to manually adjust the range to the reading range. The frame designates where to go from the account item (Kanji) field and from to where to go from the amount (numerical) field. Thereafter, in order to digitize the stored image information, a digitizing L / C analysis button is pressed from the screen. When the L / C analysis button is pressed, a program for converting to digital characters is activated, the characters of the image information (within the range surrounded by a frame) are digitized, and the result is displayed instead of the image information display. . The digitized result is displayed in a number column from 2 vertically, based on the selected pattern, an account column from the left (character string left-justified is displayed), and an amount column on the right (displayed right-justified) ) Is displayed as a number sequence. In principle, the columns and rows are basically the same as the original financial statements.If the amount of the middle section is on the right of the amount of the small section and the amount of the large section is on the right, the program automatically Consideration is given to moving the amount to each of the large and medium category account lines to which the amount belongs.

Further, when the consistency check button on the screen is pressed, a predetermined account check program is activated, and the digitized (recognized) account character strings are matched line by line with the account dictionary, and the same account is Check if it exists in the subject dictionary. Only when it does not exist, the misread dictionary replacement function for automatically correcting misread characters and the misread dictionary replacement function complementing program are executed. The result is compared with the account dictionary again to check whether the same account exists. If it still does not exist, it is assumed that misread characters exist and the account character string is displayed in red. Regarding the accounts displayed in red, the characters read incorrectly are compared with the original input slips and corrected. When making corrections, pressing a button displayed in the right corner of each line of the column displaying the account item will immediately display an account candidate having characters similar to the account item in that column. Candidates are extracted by the program from the account dictionary by a predetermined method. afterwards,
The program returns to the point where the consistency check button is pressed again and is repeated. It is assumed that the account has been registered correctly when all the digitized red character strings of each account have disappeared. In some cases, when an unregistered account in the account dictionary occurs (when it is displayed in red but the intuitive item is as described in the input original form), it is necessary to register the account using a predetermined method. Become.

Next, the amount column is checked. When the total check button displayed on the screen is pressed, each amount of the digitized result in "FIG. 4" is activated by a predetermined program and checked based on the accounting process. If there is a misreading of the amount (in the case of a balance check error by the same method as the accounting method), the relevant numerical value in the large and medium group is displayed in red. Each amount in the group displayed in red is compared with the input original form, and the misread amount is found and corrected. When the correction has been made, press the total check button again and check in the same way. When the red display disappears, the amount check ends. Thereafter, a program to be aggregated in the representative account is started by a predetermined method, and the input result is aggregated and stored in the representative account in each accounting period of each company. Further, the stored results are stored as accumulated information in units of companies and settlement periods until they are deleted. This stored information is used in a company diagnosis system or the like. Although not particularly described in detail, since the means of aggregation into representative intellectual subjects differs for each company using the apparatus of the present invention, there is a device that can be easily set corresponding to a screen for each company.

[0022]

According to the financial statement automatic input device of the present invention, financial statements printed in type can be input accurately in the shortest time. Even those who are not good at computer operation and who are inexperienced can learn the input method in a short time. By using the aggregation destination code for the representative account registered in the account dictionary, anyone can always aggregate based on a certain standard, so that the diagnosis of each company can be standardized. Realization of a low-cost, client-server type system that does not depend on general-purpose computers that does not require specialized input, etc. By providing the device of the present invention, direct input at each business office nationwide is realized and distributed. Processing, paperless environment realization, and input based on unified standardization are planned. Since financial statements can be digitized easily in a short period of time, financial institutions, etc. eliminate ambiguous diagnosis, manual diagnosis, diagnosis by specialists, etc. It is further enhanced. Therefore, the apparatus of the present invention can solve the computerization of the most delayed business in the financial institution system by an epoch-making method, and can realize computerization.

[Brief description of the drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of an automatic financial statement input device according to the present invention.

[Explanation of symbols]

 1. Computer 2. Scanner 3. Display 4. Keyboard 5. Floppy disk (FD) Hard disk 7. Printer 8. Account dictionary 9. Misread dictionary 10. Other external storage device 11. Built-in communication device 12. telephone line

FIG. 2 is a network diagram using the device of the present invention.

FIG. 3 is a diagram of an embodiment in which a pattern of a method of describing items and amounts of money in financial statements is pictorially illustrated.

FIG. 4 is a diagram of an embodiment of an input screen of the device of the present invention.

Claims (12)

[Claims] An apparatus according to the present invention uses a scanner to execute a program having a function of replacing the result of inputting type-printed financial statements as image information with digital characters, and converts the characters of the image information into digital characters. Means for digitizing and digitized (recognized) characters may cause misreading recognition because of the sharpness of image characters, similar characters that are easy to misread, etc. Utilizing the features of the statement method (accounting method), a means for dividing into account character strings and numerical values (money amounts) by a predetermined method, and a separate character string (account item) were separately prepared. Means for checking with the account dictionary in a predetermined manner and checking whether or not the character string exists in the account dictionary; and determining that the existing character string is correctly recognized, Means for assuming that the column contains misread characters. 2. In the same way as when there are characters that are easily erroneous when stored or written by humans, when converting image characters into digital characters, there is a tendency to misread the same character similar to an image. As a method of automatically correcting to correct characters using this habit, a large amount of events of characters that are easily misrecognized are collected in advance, and a number of similar character replacement patterns are added to the misread dictionary by a predetermined method, and correct characters are corrected. An automatic financial statement input device comprising means for creating a misread dictionary by incorporating a table to be replaced with a financial statement. 3. In the case of processing every day, events for recognizing misreading of similar similar characters that occur each time are collected, and a similar character replacement pattern is added to the misreading dictionary by a predetermined method every day to improve recognition accuracy. An automatic financial statement input device comprising learning means. 4. A method for correcting a misread character by using a pattern for replacing a character string containing a misread character in a recognition result with a correct character registered in the misread dictionary. Financial statement automatic input device. 5. In order to further improve the accuracy of the recognition function,
In addition to using the misread dictionary, instead of the method of registering the collection results described in claims 2 and 3 in the misread dictionary, the event is registered in a program, and the misread characters are replaced with correct characters on the program. An automatic financial statement input device, comprising a means for supporting a misread dictionary complementing program for ensuring more accurate recognition by providing a predetermined method. 6. It is assumed that the character string matches again with the account dictionary after the processing of the means described in claim 4 and claim 5, and if the character string exists, it is correctly recognized. If it does not exist, the character string is converted to a character string containing misread characters, and color display is performed to distinguish it from other correctly replaced character strings. A financial statement automatic input device, characterized in that the financial statement automatic input device is provided with means for performing the following. 7. In order to more easily correct a misread character, a combination of characters of a character string containing the misread character is used, and a similar account having the combination character is extracted from the account dictionary by a predetermined method. And displaying the extracted candidates on a screen, and, when a correct account item described in the input original form exists in the candidate, a similar guessed subject search means for easily selecting / correcting and inputting from the candidate. Automatic financial statement input device. 8. When an account identical to the account of the recognition result exists in the account dictionary, the account is input in a predetermined manner using the aggregation destination code set in the corresponding account of the account dictionary. An automatic financial statement input device comprising means for aggregating accounts into representative accounts. 9. A means for automatically correcting / adding a required account in a predetermined manner when a required account does not exist in the character string (account) of the recognition result due to misreading or the like. An automatic financial statement input device. 10. A method of describing a subject and a monetary amount (a subject / amount A or a subject / amount A / amount B, a subject / amount A / amount B / amount C, etc.) according to the characteristics of the financial statements (the description method of accounting treatment). Since it applies to several patterns, a typical multiple written pattern is displayed on the screen as a picture, simply compared with the input original, and the reading range frame is easily specified by pressing the corresponding picture pattern button An automatic input device for financial statements, wherein the information is automatically set by using information stored in the pattern to determine whether the item is a character string or a numerical value, and the image information is easily and accurately input. . 11. As a result of inputting financial statements, if the account does not exist in the account dictionary, and there is an account determined not to be misread, the account and its subdivision code are determined by a predetermined method. Use similar accounts in the account dictionary for the account code of the item), the middle category code (medium category account code), the major category code (major category account code), and the aggregation destination code for the representative account An automatic financial statement input device comprising: a method of easily additionally registering in an intuitive subject dictionary; and means for processing an input result. 12. A network system using a communication line is constructed by using the apparatus of the present invention, and a direct input on site can be realized without sending an original form for financial statements to an input center at another place. An automatic financial statement input device, characterized in that distributed processing is enabled by transmitting digital information and image information of an input result, and a means for exchanging information in a short time is provided.