WO2022097189A1

WO2022097189A1 - Data processing device, data processing method, and program

Info

Publication number: WO2022097189A1
Application number: PCT/JP2020/041162
Authority: WO
Inventors: 鴻鵬葛; 顕松田; 貴亮佐藤; 智小俣; 啓太郎森
Original assignee: ファーストアカウンティング株式会社
Priority date: 2020-11-04
Filing date: 2020-11-04
Publication date: 2022-05-12
Also published as: JP2022075467A; JPWO2022097189A1; JP6870159B1

Abstract

A data processing device 1 comprises: a storage unit 12 which stores master data in which a plurality of character strings to be written on vouchers are associated as a plurality of registered character strings; a character recognition unit 132 which recognizes character strings included in voucher image data and thereby outputs a plurality of recognized character strings; and a correction unit 133 which, if a first character string of the plurality of recognized character strings is not included in the master data, but a second character string of the plurality of recognized character string that is different from the first character string is included in the master data, corrects the first character string to a similar character string that is most similar to the first character string among the one or more registered character strings that are associated, in the master data, with the second character string.

Description

Data processing equipment, data processing methods and programs

The present invention relates to a data processing apparatus, a data processing method and a program.

Conventionally, there is known a technique of recognizing characters based on an image obtained by reading an invoice and correcting the recognized characters when there is an error in the result of character recognition (see, for example, Patent Document 1).

Japanese Unexamined Patent Publication No. 2012-517637

In the conventional technique, the error is corrected by comparing the extracted characters with the contents of a predetermined database. However, there are cases where the character string described in the voucher is not registered in the database and does not match the contents of the database even though the character recognition result is correct. In such a case, if you correct it so that it matches the contents of other character strings contained in the database, a character string different from the character string described in the voucher will be displayed even though the character is correctly recognized. There was a problem that it was output.

Therefore, the present invention has been made in view of these points, and an object thereof is to improve the probability that the character string included in the image data of the voucher is correctly output.

The data processing device according to the first aspect of the present invention has a data acquisition unit for acquiring voucher image data and a character recognition unit for outputting a plurality of recognition character strings by recognizing character strings included in the voucher image data. And, the first character string among the plurality of recognition character strings is not included in the master data associated with the plurality of registered character strings, and is different from the first character string among the plurality of recognition character strings. When two character strings are included in the master data, a similar character string most similar to the first character string among one or more registered character strings associated with the second character string in the master data. It has a correction unit for correcting the first character string, and an output unit for outputting the corrected first character string after the correction of the first character string and the second character string in association with each other.

The correction unit corrects the first character string to the similar character string on condition that two or more of the second character strings among the plurality of recognition character strings are included in the master data. May be good.

When the correction unit identifies that there are a plurality of candidates for the similar character string that are most similar to the first character string, the correction unit determines the similar character string before correcting the first character string to the similar character string. A plurality of candidates may be output to the output unit, and the first character string may be corrected to the similar character string corresponding to the candidate selected from the plurality of candidates.

The master data includes the company name and account information as the plurality of registered character strings, and the correction unit includes the recognition company name which is the first character string recognized by the character recognition unit in the master data. If the master data includes the recognition account information which is the second character string recognized by the character recognition unit, the recognition is given to the company name associated with the recognition account information in the master data. You may correct the company name.

The master data includes a company name and an item name as the plurality of registered character strings, and the correction unit includes a recognition company name which is the first character string recognized by the character recognition unit in the master data. However, when the master data includes the recognized item name which is the second character string recognized by the character recognition unit, a plurality of company names associated with the recognized item name in the master data. Of these, the recognized company name may be corrected to the company name most similar to the recognized company name.

The master data includes the item name and the product unit price as the plurality of registered character strings, and the correction unit includes the recognized item name which is the first character string recognized by the character recognition unit in the master data. However, when the recognized product unit price, which is the second character string recognized by the character recognition unit, is included in the master data, a plurality of item names associated with the recognized product unit price in the master data. Of these, the recognized item name may be corrected to the item name most similar to the recognized item name.

Before correcting the first character string to the similar character string, the correction unit executes a search on the Internet using at least one of the first character string or the similar character string as a keyword, and performs a search on the Internet, and the similar character. If it is determined that the column is more likely to be correct than the first character string, the first character string may be corrected to the similar character string.

Before correcting the first character string to the similar character string, the correction unit executes a search on the Internet using at least one of the first character string or the similar character string as a keyword, and performs the first search. When it is determined that the character string is more likely to be correct than the similar character string, the similar character string in the master data may be corrected to the first character string without correcting the first character string. ..

The output unit can distinguish between the character string determined by the correction unit to be necessary for correction and the character string determined by the correction unit that correction is not necessary among the plurality of recognition character strings. You may output it.

When the master data does not include a character string having a similarity with the first character string equal to or higher than the threshold value, the output unit outputs the first character string as a character string to be registered in the master data. You may.

The output unit is said to have many character strings not included in the master data when the ratio of the recognition character strings not included in the master data is equal to or more than a predetermined value among the plurality of recognition character strings. Information indicating that may be output.

The data processing method of the second aspect of the present invention is a step of acquiring voucher image data executed by a computer and a step of outputting a plurality of recognition character strings by recognizing a character string included in the voucher image data. And, the first character string among the plurality of recognition character strings is not included in the master data associated with the plurality of registered character strings, and is different from the first character string among the plurality of recognition character strings. When two character strings are included in the master data, a similar character string most similar to the first character string among one or more registered character strings associated with the second character string in the master data. It has a step of correcting the first character string and a step of associating the corrected first character string with the corrected first character string after the correction of the first character string and outputting the second character string.

The program of the third aspect of the present invention includes a step of acquiring voucher image data, a step of outputting a plurality of recognition character strings by recognizing a character string included in the voucher image data, and the plurality of steps. The first character string among the recognition character strings of is not included in the master data associated with a plurality of registered character strings, and the second character string different from the first character string among the plurality of recognition character strings is When included in the master data, the first of the registered character strings associated with the second character string in the master data is the most similar to the first character string. The step of correcting the character string and the step of associating the corrected first character string with the corrected first character string after the correction of the first character string and outputting the second character string are executed.

According to the present invention, there is an effect that the probability that the character string included in the image data of the voucher is correctly output can be improved.

It is a figure for demonstrating the outline of a data processing apparatus. This is an example of invoice image data, which is a type of voucher image data. This is an example of invoice data. It is a figure which shows the structure of the data processing apparatus. It is a figure which shows an example of company master data. It is a figure which shows an example of product master data. It is a figure which shows an example of the display screen of the voucher data output by an output part. It is a flowchart which shows the processing flow of the data processing apparatus 1.

[Overview of data processing device 1]
FIG. 1 is a diagram for explaining an outline of the data processing device 1. The data processing device 1 is a device for specifying a character string included in the voucher image data by acquiring voucher image data and performing character recognition processing on the voucher image data, for example, a computer. The data processing device 1 creates voucher data including the specified character string, and outputs the created voucher data to the external device 3.

Voucher image data is image data of voucher documents such as invoices, purchase orders, invoices, quotations or inspection slips. If the voucher image data is the image data of the invoice, the voucher image data is an image of the voucher containing the company name, contact information, the name (item name) of the product or service to be billed, the billing amount, the tax amount, etc. It is the converted data. The voucher image data is, for example, image data generated by the image reading device 2 (for example, a scanner or a digital camera) reading the voucher, but may be image data or text data created by a computer.

The external device 3 is, for example, a computer used by a user (for example, an accounting person) who uses the data processing device 1 in a company that has received a voucher, or an enterprise resource planning (ERP). The data processing device 1 displays, for example, the specified character string on the user's computer, or transmits the specified character string to the core system. The core system is a system that stores various data used for accounting, for example.

FIG. 2 is an example of invoice image data, which is a kind of voucher image data. FIG. 3 is an example of invoice data which is voucher data including a character string specified based on the invoice image data shown in FIG. In the invoice data shown in FIG. 3, the character string included in the invoice image data shown in FIG. 2 is correctly described.

However, it is not always possible for the data processing device 1 to correctly identify all the character strings included in the invoice image data by performing character recognition. When the data processing device 1 erroneously recognizes a character string, the invoice data contains an erroneous character string, invoice data including the erroneous character string is created, and the erroneous character string is registered in the external device 3. Will be done.

Therefore, the data processing device 1 according to the present embodiment has the company name, branch name, telephone number, account information, contact information, department in charge, person in charge name, item name, and product unit price, which are character strings described in the voucher. Of these, the master data associated with a plurality of character strings is referred to, and it is determined whether or not the character-recognized character string is correct based on the plurality of character strings associated with the master data. When the data processing device 1 determines that there is an error in the character string to be determined by using another character string associated with the master data, the misrecognized character string is used as the above-mentioned other character in the master data. Correct to the most similar string among the multiple strings associated with the column. Since the data processing device 1 is configured in this way, the probability that the data processing device 1 outputs a correct character string is increased.

Hereinafter, the configuration and operation of the data processing device 1 will be described in detail. In the following explanation, the case where the voucher is an invoice will be mainly described, but the voucher may be an order form, a delivery note, a quotation, an acceptance slip, etc. other than the invoice.

[Configuration of data processing device 1]
FIG. 4 is a diagram showing the configuration of the data processing device 1. The data processing device 1 includes a communication unit 11, a storage unit 12, and a control unit 13. The control unit 13 includes a data acquisition unit 131, a character recognition unit 132, a correction unit 133, and an output unit 134.

The communication unit 11 includes a communication interface for transmitting and receiving various data to and from the image reading device 2 or the external device 3 via a network such as the Internet or an intranet. The communication unit 11 inputs the invoice image data received from the image reading device 2 to the data acquisition unit 131. Further, the communication unit 11 outputs the invoice data output by the output unit 134 to the external device 3.

The storage unit 12 has a storage medium such as a ROM (ReadOnlyMemory), a RAM (RandomAccessMemory), and an SSD (SolidStateDrive). The storage unit 12 stores a program executed by the control unit 13.

In addition, the storage unit 12 has a company name (registered company name), a branch name, a telephone number, account information (registered account information), contact information, a department in charge, a person in charge, and an item name, which are character strings written on the voucher. And, the master data in which a plurality of character strings among the product unit prices are associated as a plurality of registered character strings is stored. The storage unit 12 stores, for example, company master data 121 including data on the company and product master data 122 including data on the product. These master data are used when the user creates a voucher, but are also used by the control unit 13 for character recognition and correction of the specified character string. The company master data 121 and the product master data 122 may be stored in a storage medium external to the storage unit 12.

FIG. 5 is a diagram showing an example of company master data 121. In the company master data 121 shown in FIG. 5, a company name, a branch name, a company address, a telephone number, a department in charge, a person in charge, and a transfer account number are associated with each other.

FIG. 6 is a diagram showing an example of product master data 122. In the product master data 122 shown in FIG. 6, the product name, the price, the manufacturer name, and the business partner name are associated with each other. The product name includes the product name and the model name or model number, but may include only the product name or may include only the model name or model number.

The control unit 13 shown in FIG. 4 has, for example, a CPU (Central Processing Unit). The control unit 13 functions as a data acquisition unit 131, a character recognition unit 132, a correction unit 133, and an output unit 134 by executing the program stored in the storage unit 12.

The data acquisition unit 131 acquires voucher image data via the communication unit 11. The data acquisition unit 131 acquires, for example, the voucher image data output from the image reading device 2 that has read the voucher, and inputs the acquired voucher image data to the character recognition unit 132.

The character recognition unit 132 outputs a plurality of recognition character strings by recognizing the character strings included in the voucher image data. The character recognition unit 132 recognizes the characters included in the voucher image data by executing OCR (Optical Character Recognition) processing executed by, for example, AI (Artificial Intelligence), and sets a plurality of consecutive characters as character strings. recognize. The character recognition unit 132 inputs a plurality of recognized character strings to the correction unit 133.

In the correction unit 133, the first character string among the plurality of recognition character strings recognized by the character recognition unit 132 in one voucher image data is not included in the master data associated with the plurality of registered character strings, and a plurality of them. When the master data contains a second character string different from the first character string among the recognition character strings of, the first character of one or more registered character strings associated with the second character string in the master data. Correct the first string to a similar string that most closely resembles the column. When the first character string and the second character string are not included in the master data, the correction unit 133 does not correct the first character string, and the first character string and the second character string are included in the master data. Notify the output unit 134 that it has not been done. The master data may be company master data 121 or product master data 122. The correction unit 133 may use the company master data 121 and the product master data 122 in combination. The correction unit 133 inputs the corrected first character string to the output unit 134.

The output unit 134 outputs the corrected first character string and the second character string in association with each other. The output unit 134 outputs voucher data in a state in which a plurality of character strings corresponding to one voucher image data are associated, for example, like the invoice data shown in FIG. The output unit 134 outputs the voucher data to the external device 3 via the communication unit 11, or outputs the voucher data to the display, for example.

The output unit 134 outputs the first character string in association with the similar character string before the correction unit 133 corrects the first character string to a similar character string, and the correction unit 133 outputs the first character string by the output unit 134. After associating the column with the similar character string and outputting it, the first character string may be corrected to the similar character string on condition that the instruction for permitting the correction is received. Further, when the output unit 134 receives a notification from the correction unit 133 that the first character string and the second character string are not included in the master data, the first character string and the second character string are included in the master data. You may output the information indicating that it has not been done.

[Details of correction processing]
Hereinafter, the details of the correction process by the correction unit 133 will be described.
As an example, in the correction unit 133, the recognition company name which is the first character string recognized by the character recognition unit 132 is not included in the master data, and the recognition account information which is the second character string recognized by the character recognition unit is included. If it is included in the master data, the recognized company name is corrected to the company name associated with the recognized account information in the master data.

As an example, it is assumed that a plurality of recognition character strings are "Taguchi Shoji" and "AAA Bank Tokyo Branch Ordinary 12233334". When the first character string is "Taguchi Shoji", the company master data 121 shown in FIG. 5 does not include the first character string. However, the second character string "AAA Bank Tokyo Branch Ordinary 12233334" is included in the company master data 121 shown in FIG. In such a case, the correction unit 133 may use a plurality of character strings such as "Tanaka Shoji", "Tokyo Branch", and "001-12, Chiyoda-ku, Tokyo" associated with "AAA Bank Tokyo Branch Ordinary 12233334". Of these, "Tanaka Shoji" is judged to be the most similar to "Taguchi Shoji", and "Tanaka Shoji" in the first character string is corrected to "Tanaka Shoji".

The correction unit 133 identifies the most similar similar character string, for example, based on the similarity calculated based on the number of matching characters among a plurality of characters included in each of the plurality of character strings. The correction unit 133 specifies the character string having the largest number of character strings that match the first character string as a similar character string. When the correction unit 133 specifies that the first character string is a company name, the correction unit 133 may specify the character string registered as the company name in the master data as a similar character string.

As another example, it is assumed that a plurality of recognition character strings are "ink (AK-123)" and "¥ 1,800". When the first character string is "¥ 1,800", the product master data 122 shown in FIG. 6 does not include the first character string. However, the second character string "ink (AK-123)" is included in the product master data 122 shown in FIG. In such a case, the correction unit 133 uses the character strings of "¥ 1,000", "ABC", "Tanaka Shoji Co., Ltd.", etc. associated with "ink (AK-123)" to be "¥ 1,000". It is determined that "1,000" is the most similar to "¥ 1,800", and "¥ 1,800" in the first character string is corrected to "¥ 1,000".

The correction unit 133 does not include the recognition item name, which is the first character string recognized by the character recognition unit 132, in the master data, and the recognition product unit price, which is the second character string recognized by the character recognition unit 132, is the master data. When included in, the recognized item name may be corrected to the item name most similar to the recognized item name among the plurality of item names associated with the recognized product unit price in the master data. For example, when the character recognition unit 132 recognizes the ink (AK-723) as the first character string and the character recognition unit 132 recognizes "¥ 1,000" as the second character string, the correction unit 133 is shown in FIG. In the indicated product master data 122, the "ink (AK-0123)" and the "copy paper (A1)" associated with "¥ 1,000" are specified. Then, the correction unit 133 puts the first character string "ink (AK-723)" in "ink (AK-0123)" and "copy paper (A1)" rather than "copy paper (A1)". The first character string is corrected to a similar "ink (AK-0123)".

In the correction unit 133, the recognition company name, which is the first character string recognized by the character recognition unit 132, is not included in the master data, and the recognition item name, which is the second character string recognized by the character recognition unit, is included in the master data. If it is included, the recognized company name may be corrected to the company name most similar to the recognized company name among the plurality of company names associated with the recognized item name in the master data. For example, when the character recognition unit 132 recognizes "Taguchi Shoji Co., Ltd." as the first character string and the character recognition unit 132 recognizes "ink (AK-0123)" as the second character string, the correction unit 133 is shown in the figure. In the product master data 122 shown in 6, "Tanaka Shoji Co., Ltd." and "MM Electric Co., Ltd." associated with "ink (AK-0123)" are specified. Then, the correction unit 133 is a "Tanaka Shoji Co., Ltd." that is more similar to the first character string "Taguchi Shoji Co., Ltd." than "MM Electric Co., Ltd." among "Tanaka Shoji Co., Ltd." and "MM Electric Co., Ltd." Correct the first character string to "company".

By the way, even if the second character string is included in the master data, it may be included in the master data as a result of erroneous recognition of the second character string. In such a case, if the correction unit 133 corrects the first character string to the character string associated with the second character string on the premise that the second character string is correct, the first character is added to the wrong character string. The columns may be corrected. Therefore, the correction unit 133 corrects the first character string to a similar character string on condition that the correction unit 133 includes two or more second character strings among the plurality of recognition character strings in the master data. You may.

In the case of the previous example, the correction unit 133 is associated with, for example, "AAA Bank Tokyo Branch Ordinary 12233334" and "03-1234-5678" by being included in the master data. Correct "Taguchi Shoji" in the first character string to "Tanaka Shoji". When the probability that two or more character strings are erroneously recognized is sufficiently low, the correction unit 133 operates in this way to reduce the probability that the first character string is corrected to the wrong character string. Therefore, it is possible to improve the probability that the character string described in the voucher is output correctly.

When the correction unit 133 identifies that there are a plurality of candidates for a similar character string most similar to the first character string, the correction unit 133 selects a plurality of candidates for the similar character string before correcting the first character string to the similar character string. The first character string may be corrected to a similar character string corresponding to a candidate selected from a plurality of candidates by causing the output unit 134 to output. For example, if the company master data 121 includes the telephone number "03-1234-5678" and the fax number "03-1234-5679" and the first character string is "03-1234-5670", "03-" Both "1234-5678" and "03-1234-5679" are similar to the first string.

In such a case, since it is necessary for the user to determine which character string should be corrected, the correction unit 133 uses "03-1234-5678" and "03-1234-5679" as correction candidates. , Display on the display of the computer used by the user. The correction unit 133 corrects the first character string to the character string selected by the user. By operating the correction unit 133 in this way, even when a plurality of similar character strings are registered in the master data, the probability that the first character string is corrected to the wrong character string is reduced. be able to.

By the way, if the character string recognized by the character recognition unit 132 does not match the character string included in the master data, it is assumed that the character string included in the master data is incorrect. For example, if the company relocates, the address contained in the master data may not be the latest address. Therefore, the correction unit 133 executes a search on the Internet using at least one of the first character string or the similar character string as a keyword before correcting the first character string to the similar character string, and the similar character string is the first. If it is determined that the probability of being correct is higher than that of one character string, the first character string may be corrected to a similar character string.

As an example, the first character string recognized by the character recognition unit 132 is "○ 2-15, Chiyoda-ku, Tokyo", and the correction unit 133 is "001, Chiyoda-ku, Tokyo" in the company master data shown in FIG. When "-12" is specified as a similar character string, the correction unit 133 searches on the Internet using the company name or telephone number associated with the similar character string as a keyword. The correction unit 133 does not correct the first character string when the address described in the searched and displayed website matches the first character string, and the address described in the website is similar. If it matches the character string and it is determined that the similar character string is more correct than the first character string, the first character string "○ 2-15, Chiyoda-ku, Tokyo" is changed to "001, Chiyoda-ku, Tokyo". -12 ". By operating the correction unit 133 in this way, it is possible to prevent erroneous correction when the character string registered in the master data is not the latest.

Further, the correction unit 133 executes a search on the Internet using at least one of the first character string or the similar character string as a keyword before correcting the first character string to a similar character string, and the first character string is used. When it is determined that the probability of being correct is higher than that of the similar character string, the similar character string in the master data may be corrected to the first character string without correcting the first character string. In the case of the above example, the correction unit 133 has a high probability that the address described in the website searched and displayed matches the first character string, and the first character string is more correct than the similar character string. If it is determined that, the character string "001-12, Chiyoda-ku, Tokyo" in the company master data is corrected to the first character string "XX2-15, Chiyoda-ku, Tokyo". By operating the correction unit 133 in this way, the master data is updated. As a result, in the future, the accuracy of correction when the character recognition unit 132 misrecognizes and the correction unit 133 corrects the character string is improved.

[Operation of output unit 134]
The output unit 134 outputs the corrected character string to the core system in order to improve the probability that an appropriate character string is registered in the core system based on the character string described in the voucher. It may be possible for the user to confirm. As an example, the output unit 134 can distinguish between a character string determined by the correction unit 133 that correction is necessary and a character string determined by the correction unit 133 that correction is not necessary among the plurality of recognition character strings. Output in the mode.

FIG. 7 is a diagram showing an example of a voucher data display screen output by the output unit 134. In FIG. 7 (a), "Taguchi Shoji" and "ink (AK-723)" are not included in the company master data or the product master data, so that they are different from other character strings (italicized characters in a thick frame). ) Is displayed. By outputting such data by the output unit 134, for example, the user can easily grasp which character string needs to be corrected.

The output unit 134 may display a character string that is a candidate for correction when a predetermined operation is performed on the screen of FIG. 7A. The predetermined operation is, for example, an operation (for example, a click operation or a touch operation) in which the user selects a character string that needs to be corrected. FIG. 7B is a diagram showing an example in which a character string that is a candidate for correction is displayed. When the user selects the displayed candidate, the correction unit 133 corrects the character string displayed in FIG. 7A to the displayed candidate character string. As a result, the voucher data is corrected to the state shown in FIG.

By the way, as a result of recognizing the character strings included in the voucher image data, there are cases where many character strings are not included in the master data. It is assumed that this happens when the image quality of the voucher image data is too poor, when it is a voucher issued by mistake, or when it is a voucher from a business partner who is not registered in the master data. To. In order for the user to grasp these situations, the output unit 134 determines that the ratio of the recognition character strings not included in the master data among the plurality of recognition character strings is equal to or more than a predetermined value. Warning information indicating that there are many character strings not included in the master data may be output.

The output unit 134 may output warning information together with a plurality of recognition character strings. As shown in FIG. 7, the output unit 134 may output warning information together with a plurality of recognition character strings in a state where the character string requiring correction can be identified. The output unit 134 may display a screen for inputting a process to be executed by the user together with the warning information or as the warning information. The output unit 134 displays, for example, a screen for performing an operation for associating a plurality of recognition character strings and registering them in master data. The output unit 134 registers a plurality of recognition character strings in the master data when the operation for registration is performed.

When the master data does not include a character string similar to the first character string (for example, a character string having a similarity equal to or higher than the threshold value), the output unit 134 sets the first character string as a character string to be registered in the master data. You may output it. The output unit 134, for example, when the first character string is "Sato Shoji" and the master data does not include "Sato Shoji", such as "Do you want to register Sato Shoji in the master data?" , Display a message containing the first character string on the user's computer.

The output unit 134 registers in the master data among a plurality of other character strings included in the voucher image data including the character string determined to be registered (for example, the above-mentioned "Sato Shoji"). A plurality of character strings corresponding to the items to be registered may be displayed as character strings of candidates for registration. The output unit 134 displays the operation image that accepts the operation for registration together with the character string of the registration target candidate, and when the operation image is operated, the first character string and the character string of the registration target candidate are displayed. It may be registered in the master data. By operating the output unit 134 in this way, for example, when a new customer who has no past transaction occurs, the user does not need to input the character string to be registered in the master data, so that the user's work efficiency is improved. improves.

[Processing flow of data processing device 1]
FIG. 8 is a flowchart showing a processing flow of the data processing device 1. The flowchart shown in FIG. 8 starts from the time when the image reading device 2 outputs the voucher image data.

When the data acquisition unit 131 acquires the voucher image data output by the image reading device 2 (S11), the character recognition unit 132 executes an OCR process for recognizing the characters included in the voucher image data (S12). The character recognition unit 132 recognizes a plurality of character strings based on the recognized characters.

The correction unit 133 first selects one recognition character string from the plurality of recognition character strings in order to determine whether or not the plurality of character strings recognized by the character recognition unit 132 are correctly recognized (S13). The correction unit 133 determines whether or not one of the selected recognition character strings matches any of the plurality of character strings included in the master data (S14). When the correction unit 133 determines that one selected recognition character string matches any of a plurality of character strings included in the master data (YES in S14), the correction unit 133 selects another recognition character string (S15). ), The process of S14 is executed again.

When the correction unit 133 determines that the first character string, which is one selected recognition character string, does not match all of the plurality of character strings included in the master data (NO in S14), the character recognition unit 132 It is determined whether or not the other recognition character string among the plurality of recognition character strings recognized by the user matches any of the plurality of character strings included in the master data (S16).

When the correction unit 133 determines that the other recognition character string matches any of the plurality of character strings included in the master data (YES in S16), the master data that matches the other recognition character string. Among the plurality of character strings associated with the character string in the above, the recognized first character string is corrected to the character string most similar to the first character string (S17). When the correction unit 133 determines that the other recognition character string does not match any of the plurality of character strings included in the master data (NO in S16), the correction unit 133 further refers to the other recognition character string in S16. Executes the processing of.

The correction unit 133 determines whether or not the processing from S14 to S17 has been completed for all the recognition character strings, and if not, returns to S14. When the processing for all the recognized character strings is completed (YES in S18), the correction unit 133 creates voucher data composed of the corrected character strings, and the output unit 134 outputs the voucher data (S19).

[Effect of data processing device 1]
As described above, the data processing device 1 includes the company name, branch name, telephone number, account information, contact information, department in charge, person in charge name, item name, and product unit price, which are character strings written on the voucher. Refer to the master data in which multiple strings are associated as multiple registered strings. Then, the correction unit 133 does not include the first character string in the master data among the plurality of recognition character strings specified by recognizing the character strings included in the voucher image data, and the correction unit 133 has a plurality of recognition character strings. When the master data contains a second character string different from the first character string, it is most similar to the first character string among one or more registered character strings associated with the second character string in the master data. The first character string is corrected to the similar character string.

Since the data processing device 1 is configured in this way, the output unit 134 can output a correct character string even if an error occurs in character recognition, so that the character string included in the image data of the voucher is correct. Improve the probability of being output. As a result, it is possible to improve the work efficiency and work quality of the user who performs the work using the data described in the voucher.

Although the present invention has been described above using the embodiments, the technical scope of the present invention is not limited to the scope described in the above embodiments, and various modifications and changes can be made within the scope of the gist. be. For example, all or part of the device can be functionally or physically distributed / integrated in any unit. Also included in the embodiments of the present invention are new embodiments resulting from any combination of the plurality of embodiments. The effect of the new embodiment produced by the combination has the effect of the original embodiment together.

For example, in the above description, the case where the company master data 121 and the product master data 122 are stored in the storage unit 12 of the data processing device 1 has been illustrated, but the data processing device 1 has the company master data 121 and the product master. Only one of the data 122 may be used. Further, the data processing device 1 does not have to be configured by one computer, and a plurality of computers may operate in cooperation with each other, or the computer and the storage medium in which the master data is stored may be physically separated. You may be doing it.

1 Data processing device 2 Image reader 3 External device 11 Communication unit 12 Storage unit 13 Control unit 121 Company master data 122 Product master data 131 Data acquisition unit 132 Character recognition unit 133 Correction unit 134 Output unit

Claims

The data acquisition unit that acquires voucher image data,
A character recognition unit that outputs a plurality of recognition character strings by recognizing the character strings included in the voucher image data, and
The first character string among the plurality of recognition character strings is not included in the master data associated with the plurality of registered character strings, and the second character different from the first character string among the plurality of recognition character strings. When the column is included in the master data, the similar character string most similar to the first character string among the one or more registered character strings associated with the second character string in the master data. A correction unit that corrects the first character string,
An output unit that outputs the corrected first character string after the first character string is corrected in association with the second character string, and
Data processing device with.
The correction unit corrects the first character string to the similar character string on condition that two or more of the second character strings among the plurality of recognition character strings are included in the master data.
The data processing apparatus according to claim 1.
When the correction unit identifies that there are a plurality of candidates for the similar character string that are most similar to the first character string, the correction unit determines the similar character string before correcting the first character string to the similar character string. The first character string is corrected to the similar character string corresponding to the candidate selected from the plurality of candidates by outputting the plurality of candidates of the above to the output unit.
The data processing apparatus according to claim 1 or 2.
The master data includes the company name and account information as the plurality of registered character strings.
The correction unit does not include the recognition company name which is the first character string recognized by the character recognition unit in the master data, and the recognition account information which is the second character string recognized by the character recognition unit. Is included in the master data, the recognized company name is corrected to the company name associated with the recognized account information in the master data.
The data processing apparatus according to any one of claims 1 to 3.
The master data includes the company name and the item name as the plurality of registered character strings.
In the correction unit, the recognition company name which is the first character string recognized by the character recognition unit is not included in the master data, and the recognition item name which is the second character string recognized by the character recognition unit. Is included in the master data, the recognized company name is corrected to the company name most similar to the recognized company name among the plurality of company names associated with the recognized item name in the master data. ,
The data processing apparatus according to any one of claims 1 to 4.
The master data includes the item name and the product unit price as the plurality of registered character strings.
In the correction unit, the recognition item name which is the first character string recognized by the character recognition unit is not included in the master data, and the recognition product unit price which is the second character string recognized by the character recognition unit is not included in the master data. Is included in the master data, the recognized item name is corrected to the item name most similar to the recognized item name among the plurality of item names associated with the recognized product unit price in the master data. ,
The data processing apparatus according to any one of claims 1 to 5.
Before correcting the first character string to the similar character string, the correction unit executes a search on the Internet using at least one of the first character string or the similar character string as a keyword, and performs a search on the Internet, and the similar character. When it is determined that the column has a higher probability of being correct than the first character string, the first character string is corrected to the similar character string.
The data processing apparatus according to any one of claims 1 to 6.
Before correcting the first character string to the similar character string, the correction unit executes a search on the Internet using at least one of the first character string or the similar character string as a keyword, and performs the first search. When it is determined that the character string is more likely to be correct than the similar character string, the similar character string in the master data is corrected to the first character string without correcting the first character string.
The data processing apparatus according to any one of claims 1 to 6.
The output unit can distinguish between the character string determined by the correction unit to be necessary for correction and the character string determined by the correction unit that correction is not necessary among the plurality of recognition character strings. Output,
The data processing apparatus according to any one of claims 1 to 8.
When the master data does not include a character string having a similarity with the first character string equal to or higher than the threshold value, the output unit outputs the first character string as a character string to be registered in the master data. ,
The data processing apparatus according to any one of claims 1 to 9.
The output unit is said to have many character strings not included in the master data when the ratio of the recognition character strings not included in the master data is equal to or more than a predetermined value among the plurality of recognition character strings. Output information indicating that
The data processing apparatus according to any one of claims 1 to 10.
Computer runs,
Steps to get voucher image data and
A step of outputting a plurality of recognition character strings by recognizing the character strings included in the voucher image data, and
The first character string among the plurality of recognition character strings is not included in the master data associated with the plurality of registered character strings, and the second character different from the first character string among the plurality of recognition character strings. When the column is included in the master data, the similar character string most similar to the first character string among the one or more registered character strings associated with the second character string in the master data. The step to correct the first character string and
A step of associating and outputting the corrected first character string after the first character string is corrected and the second character string, and
Data processing method.
On the computer
Steps to get voucher image data and
A step of outputting a plurality of recognition character strings by recognizing the character strings included in the voucher image data, and
The first character string among the plurality of recognition character strings is not included in the master data associated with the plurality of registered character strings, and the second character different from the first character string among the plurality of recognition character strings. When the column is included in the master data, the similar character string most similar to the first character string among the one or more registered character strings associated with the second character string in the master data. The step to correct the first character string and
A step of associating and outputting the corrected first character string after the first character string is corrected and the second character string, and
A program to execute.