Expense inspection device, expense inspection method, and program

Abstract

To address such a problem that it is difficult to properly check input of expenses by conventional techniques.SOLUTION: An expense inspection device 1 includes: a receiving unit 12 for receiving input expense information of expenses input by a user and a receipt image corresponding to the expenses; a character string acquisition unit 131 for acquiring one or more OCR character strings as a result of OCR for the receipt image; a score acquisition unit 132 for comparing the input expense information with the one or more OCR character strings and acquiring a score relating to certainty of the input expense information; and a related information output unit 141 for outputting related information related to the score, thereby properly checking input of expenses.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] The present invention relates to an expense inspection device that inspects information on expenses input by a user, etc.

Background Art

[0002] Conventionally, there has been an accounting system with a journal automatic creation function that can automatically create a journal for input expense information (see, for example, Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the prior art, it has been difficult to appropriately confirm the input of expenses.

Means for Solving the Problems

[0005] The expense inspection device of the first invention of the present invention includes a reception unit that receives input expense information regarding expenses input by a user and a receipt image corresponding to the expenses, a character string acquisition unit that acquires one or more OCR character strings that are the results of OCR for the receipt image, a score acquisition unit that compares the input expense information with the one or more OCR character strings and acquires a score regarding the plausibility of the input expense information, and a related information output unit that outputs related information regarding the score.

[0006] With such a configuration, it becomes possible to appropriately confirm the information on the input expenses.

[0007] Further, the expense inspection device of the second invention of the present invention is an expense inspection device in which, with respect to the first invention, two or more receipts are included in the receipt image.

[0008] This configuration allows for proper verification of expense information entered using receipt images that include two or more receipts.

[0009] Furthermore, the expense inspection device of this third invention differs from the second invention in that the character acquisition unit acquires one or more OCR character strings from each image of two or more receipts without determining the boundaries between the images of two or more receipts.

[0010] This configuration allows for easy processing and proper verification of expense information entered using receipt images that include two or more receipts.

[0011] Furthermore, the expense inspection device of this fourth invention, compared to the second or third invention, is an expense inspection device in which the receiving unit receives two or more input expense information and a receipt image containing two or more receipts, the string acquisition unit acquires one or more OCR strings which are the results of OCR on the receipt image containing two or more receipts, and the score acquisition unit compares each of the two or more input expense information with the one or more OCR strings and acquires a score regarding the accuracy of the input expense information.

[0012] This configuration allows for proper verification even when there is a mix of entries for two or more expenses and receipt images containing two or more receipts.

[0013] Furthermore, the expense inspection device of this fifth invention is an expense inspection device in which, with respect to any one of the first to fourth inventions, the score acquisition unit acquires two or more parts of information that constitute the input expense information, compares each of the two or more parts of information with a string in one or more OCR strings, and acquires a score using the comparison result for each of the two or more parts of information.

[0014] This configuration allows for proper verification of the entered expense information.

[0015] Furthermore, the expense inspection device of this sixth invention, compared to the fifth invention, is an expense inspection device in which the score acquisition unit acquires partial information which is a string of n-grams of element information constituting the input expense information, detects matching strings which are strings that match the partial information from one or more OCR strings, acquires the number of characters of the preceding string located before the matching string in the element information and the number of characters of the succeeding string located after the matching string in the element information, acquires a comparison string from the OCR string containing the matching string which consists of an OCR preceding string which is the number of characters of the preceding string, the matching string and an OCR succeeding string which is the number of characters of the succeeding string which is the number of characters of the succeeding string, calculates the distance between the element information and the comparison string, and acquires a score using the distance.

[0016] This configuration allows for proper verification of the entered expense information.

[0017] Furthermore, the expense inspection device of this seventh invention differs from the sixth invention in that the score acquisition unit generates second element information, which is a different notation with the same meaning as the element information, using element information that constitutes the input expense information, acquires a comparison string for the second element information, calculates the distance between the second element information and the comparison string, and uses the distance to acquire a score.

[0018] This configuration allows for proper verification of the entered expense information.

[0019] Furthermore, the expense inspection device of the eighth invention is an expense inspection device in which, with respect to any one of the first to seventh inventions, the related information output unit notifies the user or administrator corresponding to the input expense information of related information regarding errors in the input expense information if the score indicates information that shows a low probability of satisfying predetermined conditions.

[0020] This configuration allows for appropriate notifications to be sent when information about expenses that are likely to be inappropriate is entered.

[0021] In addition, for the expense inspection device of the ninth invention, with respect to any one of the first to seventh inventions, the related information output unit is an expense inspection device that outputs related information regarding the score in association with the input expense information.

[0022] With such a configuration, it becomes possible to appropriately confirm the input expense information.

[0023] In addition, for the expense inspection device of the tenth invention, with respect to any one of the first to ninth inventions, the input expense information includes two or more pieces of information among date, amount, and payee, the score acquisition unit compares each of the two or more pieces of information with one or more OCR character strings, acquires a score for each of the two or more pieces of information, acquires a representative score that is the representative value of the two or more scores, and the related information output unit outputs related information regarding the representative score.

[0024] With such a configuration, it becomes possible to appropriately confirm the input expense information.

Effect of the Invention

[0025] According to the expense inspection device of the present invention, it becomes possible to appropriately confirm the input of expenses.

Brief Description of the Drawings

[0026] [Figure 1] Conceptual diagram of the information system A in Embodiment 1 [Figure 2] Block diagram of the same information system A [Figure 3] Flowchart for explaining an operation example of the expense inspection device 1 [Figure 4] Flowchart for explaining an example of the element score acquisition process [Figure 5] Flowchart for explaining an example of the second element information generation process [Figure 6] Flowchart for explaining an example of the partial score acquisition process [Figure 7] Flowchart for explaining an example of the related information acquisition process [Figure 8] A diagram showing an example of the user information management table. [Figure 9] A diagram showing an example of the expense information management sheet. [Figure 10] A diagram showing an example of the same pattern management table. [Figure 11] A diagram showing an example of the receipt image. [Figure 12] Figure showing an example of character recognition results. [Figure 13] This diagram shows an example of a management table for sets of partial information and comparison strings used in the score acquisition process. [Figure 14] A diagram showing an example of the receipt image. [Figure 15] Figure showing an example of character recognition results. [Figure 16] Overview of the computer system [Figure 17] Block diagram of the computer system [Modes for carrying out the invention]

[0027] The following describes embodiments of the expense inspection device and the like with reference to the drawings. Note that components denoted by the same reference numerals in the embodiments perform similar operations, and therefore, further explanation may be omitted.

[0028] (Embodiment 1) This embodiment describes an expense verification device that compares user-inputted expense information with the OCR results of receipt images corresponding to that input expense information, obtains a score regarding the accuracy of the input expense information, and outputs related information regarding the score. In this embodiment, the receipt image may include images of two or more receipts. In this embodiment, for example, the score is obtained using the distance between some of the information in the input expense information and some of the information in the OCR results.

[0029] Furthermore, this embodiment describes an expense verification device that can acquire a score regarding the accuracy of input expense information and output related information regarding the score, even when the input expense information includes two or more input expense information items and two or more receipt images.

[0030] Figure 1 is a conceptual diagram of information system A in this embodiment. Information system A comprises an expense inspection device 1 and one or more terminal devices 2.

[0031] Expense inspection device 1 is a device that inspects input expense information. Expense inspection device 1 is typically a server, such as a cloud server or ASP server. The type of expense inspection device 1 is not specified.

[0032] Terminal device 2 is a terminal used by the user. The user is, for example, a user submitting expense claims. The user is, for example, an accounting staff member, an administrator, etc. Terminal device 2 can be, for example, a so-called personal computer, a tablet device, a smartphone, etc., and the type is not limited.

[0033] Figure 2 is a block diagram of information system A in this embodiment.

[0034] The expense inspection device 1 comprises a storage unit 11, a reception unit 12, a processing unit 13, and an output unit 14. The storage unit 11 comprises a user information storage unit 111 and an expense information storage unit 112. The processing unit 13 comprises a string acquisition unit 131, a score acquisition unit 132, and a related information acquisition unit 133. The output unit 14 comprises a related information output unit 141.

[0035] The terminal device 2 includes a terminal storage unit 21, a terminal receiving unit 22, a terminal processing unit 23, a terminal transmission unit 24, a terminal receiving unit 25, and a terminal output unit 26.

[0036] The storage unit 11, which constitutes the expense inspection device 1, stores various types of information. These types of information include, for example, user information (described later), expense information (described later), calculation formulas for calculating scores (described later), patterns (described later), and normal conditions. It is preferable that one or more patterns correspond to type identifiers (described later). The normal conditions are the conditions for determining whether the score is normal or not. For example, the normal conditions are that the score is equal to or greater than a threshold, or that the score is greater than a threshold. Normal conditions may exist for each type identifier.

[0037] The user information storage unit 111 stores one or more user information entries. User information is information about a user. User information usually has a user identifier. User information usually has one or more user attribute values. The user identifier is information that identifies a user. Examples of user identifiers include user ID, employee number, email address, and telephone number. Examples of one or more user attribute values ​​include name, company name, department name, email address, and supervisor's email address.

[0038] The expense information storage unit 112 stores one or more expense information entries. Expense information is information that identifies the details of the expense claimed. Expense information may include, for example, the date, amount, and payee. Expense information may also include, for example, the account, description, etc. The payee is the name of the company or store to which the expense was paid. Each of the one or more expense information entries is associated with, for example, a receipt image. Two or more expense information entries may be associated with one receipt image. Two or more expense information entries may be associated with two or more receipt images. A receipt image is an image that includes a receipt. For example, one receipt image may contain only the image of one receipt. However, a receipt image may contain images of two or more receipts. A receipt image may be, for example, a jpeg or gif, but the data format is not restricted. A receipt image may be, for example, a file, but it may also be information in a database. A receipt image is information that is subject to character recognition processing.

[0039] The reception unit 12 receives various instructions and information. These instructions and information include, for example, input expense information and receipt images. Input expense information is information about expenses entered by the user. Input expense information has one or more elemental information. Elemental information is input expense information or information that constitutes expense information. Elemental information includes, for example, date, amount, and payee. It is preferable that input expense information includes two or more elemental information from date, amount, and payee. All or some of the elemental information that constitutes input expense information becomes information that constitutes expense information. The method of separating each of the two or more elemental information within the input expense information is not specified. For example, each of the two or more elemental information within the input expense information may be associated with a tag. For example, each of the two or more elemental information within the input expense information may be separated by a delimiter. The delimiter may be, for example, ",", "|", a space, etc., but is not specified.

[0040] The reception unit 12 accepts the associated input expense information and receipt image. The reception unit 12 may also accept two or more input expense information items and one receipt image, with the items associated with each other.

[0041] The reception unit 12 may accept receipt images containing two or more input expense information items and two or more receipts, matching them together.

[0042] The reception unit 12, for example, receives output instructions. An output instruction is an instruction to output expense information, etc. An output instruction includes information indicating the conditions for the expense information, etc. to be output.

[0043] In this context, "reception" typically refers to receiving data from terminal device 2. However, "reception" simply means that information can be obtained. Furthermore, the reception unit 12 does not need to receive the input expense information and the receipt image together. The reception unit 12 only needs to obtain the input expense information and the receipt image in a manner that allows them to be associated.

[0044] The processing unit 13 performs various processes. These processes are carried out by the string acquisition unit 131, the score acquisition unit 132, and the related information acquisition unit 133.

[0045] The string acquisition unit 131 acquires one or more OCR strings, which are the OCR results for the receipt image received by the reception unit 12. One OCR string is, for example, a single line of text. However, an OCR string may be a string of two or more lines.

[0046] The string acquisition unit 131 can be implemented using known OCR (optical character recognition) technology. The OCR string is the string obtained from the OCR process.

[0047] The string acquisition unit 131 may receive one or more OCR strings, which are the results of character recognition on the receipt image, from an external device (not shown). In this case, the string acquisition unit 131 does not perform character recognition processing. In this case, the string acquisition unit 131 transmits the receipt image to an external device that performs character recognition processing, and receives one or more OCR strings from said external device.

[0048] If a single receipt image contains images of two or more receipts, it is preferable for the string acquisition unit 131 to acquire one or more OCR strings as a result of character recognition processing on the single receipt image without determining the boundaries between the two or more receipt images. In other words, the string acquisition unit 131 acquires one or more OCR strings as a result of character recognition processing performed on the received receipt image file.

[0049] Furthermore, if a single receipt image contains images of two or more receipts, the string acquisition unit 131 may acquire one or more OCR strings in which the strings acquired from each of the two or more receipt images are mixed.

[0050] The score acquisition unit 132 compares the input expense information with one or more OCR strings and acquires a score. The score can be said to be information about the accuracy of the input expense information. The score may also be information indicating the degree of accuracy of the input expense information. The score may also be said to be probability information. Usually, a larger score indicates a greater degree of accuracy of the input expense information, but a smaller score may also indicate a greater degree of accuracy. The score is information obtained by comparing the input expense information with one or more OCR strings. For example, the more information in the input expense information is included in one or more OCR strings, the higher the score will be.

[0051] The score acquisition unit 132 preferably compares each of the two or more elemental pieces of information constituting the input expense information with one or more OCR strings to acquire a score. The method of acquiring the score is not limited. The score is, for example, the proportion of the characters constituting the input expense information that are present in one or more OCR strings.

[0052] The score acquisition unit 132 may compare two or more input expense information items with one or more OCR strings and acquire a score regarding the accuracy of the input expense information.

[0053] The score acquisition unit 132, for example, compares one or more elemental information components of the input expense information with one or more strings in the OCR string, and acquires a score using the comparison result with one or more elemental information components.

[0054] The score acquisition unit 132, for example, acquires two or more parts of the input expense information, compares each of the two or more parts of the information with a string in one or more OCR strings, and acquires a score using the comparison results for each of the two or more parts of the information.

[0055] The score acquisition unit 132, for example, acquires two or more sub-informations, which are parts of the element information, for each of the one or more element information constituting the input expense information, compares each of the two or more sub-informations with the string in the one or more OCR strings, and acquires a score using the comparison results for each of the two or more sub-informations.

[0056] The score acquisition unit 132 calculates the distance between, for example, the element information constituting the input expense information and one or more comparison strings, and uses this distance to acquire a score. The score acquisition unit 132 acquires a higher score the smaller the distance. The score acquisition unit 132 acquires a score using, for example, a decreasing function with distance as a parameter. The comparison strings will be described later.

[0057] The score acquisition unit 132 calculates the distance between the element information constituting the input expense information and one or more comparison strings, and obtains a score using the shortest distance. The score acquisition unit 132 obtains a higher score the smaller the shortest distance. The score acquisition unit 132 calculates the score using a decreasing function that takes the shortest distance as a parameter.

[0058] If there are two or more elemental information components that make up the input expense information, the score acquisition unit 132 acquires a score for each of the two or more elemental information components. Then, the score acquisition unit 132 acquires a score for the input expense information using the two or more scores. The score acquisition unit 132 acquires a representative value for the two or more scores. The representative value for the scores is, for example, the mean, median, minimum, and maximum of the scores. The score acquisition unit 132 calculates a score for the input expense information using an increasing function that takes the scores of each of the two or more elemental information components as parameters.

[0059] Note that the distance only needs to be information about the difference between two strings, such as edit distance, Levenshtein distance, or minimum edit distance.

[0060] The score acquisition unit 132 acquires the comparison string, for example, as follows: The score acquisition unit 132 detects a matching string from one or more OCR strings. Then, the score acquisition unit 132 acquires one or more OCR characters from the OCR string containing the matching string. The matching string is a string that matches the partial information.

[0061] Next, the score acquisition unit 132 obtains a comparison string consisting of, for example, an OCR prefix string, a match string, and an OCR suffix string. The OCR prefix string is the string within the OCR string that contains the match string, is the string immediately preceding the match string, and has a number of prefix characters equal to the number of prefix characters. The number of prefix characters is the number of characters in the prefix string that is located before the match string in the element information. The OCR suffix string is the string within the OCR string that contains the match string, is the string immediately following the match string, and has a number of suffix characters equal to the number of suffix characters. The number of suffix characters is the number of characters in the prefix string that is located after the match string in the element information.

[0062] Note that there may be cases where the OCR preceding string does not exist. In such cases, the score acquisition unit 132 obtains, for example, a comparison string consisting of the matching string and the OCR succeeding string. Also, there may be cases where the OCR succeeding string does not exist. In such cases, the score acquisition unit 132 obtains, for example, a comparison string consisting of the OCR preceding string and the matching string.

[0063] In other words, the score acquisition unit 132 acquires, for example, partial information (for example, an n-gram string of ) that constitutes the element information constituting the input expense information, and detects a matching string that matches the partial information from one or more OCR strings. Next, the score acquisition unit 132 acquires, for example, the number of characters in the preceding string that is located before the matching string in the element information, and the number of characters in the succeeding string that is located after the matching string in the element information, and obtains a comparison string consisting of the OCR preceding string, the matching string, and the OCR succeeding string from the OCR string containing the matching string. Next, the score acquisition unit 132 calculates, for example, the distance between the element information and the comparison string, and uses this distance to obtain a score for the element information. Furthermore, the score acquisition unit 132 uses, for example, the scores for two or more elements to obtain a score for the input expense information.

[0064] The score acquisition unit 132 calculates the distance between the element information constituting the input expense information and two or more comparison strings, and obtains a score using the smallest of these two or more distances. The score acquisition unit 132 obtains a score using a decrease function that uses the smallest distance as a parameter, for example.

[0065] The score acquisition unit 132 compares, for example, two or more elements of the input expense information, such as date, amount, and payee, with one or more OCR strings, acquires a score for each of the two or more elements, and obtains a representative score, which is a representative value of those two or more scores. The representative score is, for example, the average, median, highest, and lowest values ​​of the two or more scores.

[0066] Furthermore, the input expense information may include strings that the applicant writes after the receipt is issued, such as the number of participants, the applicant's signature, and the name of the department that used the expenses.The score acquisition unit 132 may then acquire a score using the accuracy of the strings that the applicant writes after the receipt is issued, such as the number of participants, the applicant's signature, and the name of the department that used the expenses.

[0067] The element information could be, for example, the date, amount, or payee from the input expense information. While the n-gram is preferably a 3-gram, it can also be a string of other lengths, such as a 2-gram or 4-gram. If the element information is the date "2021 / 3 / 11", the 3-gram sub-information could be, for example, "202", "021", "21 / ", "1 / 3 / ", " / 3 / ", "3 / 1", " / 11".

[0068] The score acquisition unit 132 preferably uses element information constituting the input expense information to acquire one or more second element information, which are different notations with the same meaning as the element information, acquires one or more comparison strings for each of the one or more second element information, calculates the distance between the second element information and each of the one or more comparison strings, and uses this distance to acquire a score. Note that the second element information may also be called element information. Furthermore, the algorithm for acquiring comparison strings using second element information is the same as the algorithm for acquiring comparison strings using element information.

[0069] The score acquisition unit 132 acquires one or more second element information using, for example, the element information that constitutes the input expense information, as follows: The score acquisition unit 132 acquires one or more second element information using, for example, one or more patterns stored in the storage unit 11. In other words, the score acquisition unit 132 transforms the element information using patterns and acquires the second element information.

[0070] If the element information type identifier is "Date", and the storage unit 11 contains information with the pattern "???? / ?? / ??" and "???? year ?? month ?? day" (where ? is a number from 0 to 9), and the element information contained in the input expense information is "2020 / 8 / 27", then the score acquisition unit 132 refers to the pattern in the storage unit 11 and obtains the second element information "2020 August 27". Note that the technique of generating the notation of one date from the information of one date, taking into account variations in date notation, is publicly known, so a detailed explanation is omitted. Note that the type identifier is information that identifies the type of element information, such as "Date", "Amount", or "Payee".

[0071] The score acquisition unit 132 obtains one or more strings from each of the one or more OCR strings, for example, which are the same number of characters as the number of characters in each element information constituting the input expense information, and calculates the distance between each of these one or more strings (which can be called comparison strings) and each element information. The score acquisition unit 132 may then acquire a score using the distance of one or more. It is preferable for the score acquisition unit 132 to acquire a score using the smallest of the one or more distances.

[0072] The score acquisition unit 132, for example, obtains the number of characters in each element information constituting the input expense information, and obtains one or more strings of the same number of characters as the input expense information, shifting each of the one or more OCR strings by one character from the beginning of each OCR string, and calculates the distance between each of these one or more strings (which can be called comparison strings) and each element information. The score acquisition unit 132 may then obtain a score using the minimum distance.

[0073] The related information acquisition unit 133 acquires related information regarding the score. The related information acquisition unit 133 acquires related information using the score acquired by the score acquisition unit 132.

[0074] Related information includes, for example, scores, information regarding the inspection results of input expense information, input expense information deemed correct, and input expense information deemed correct along with receipt images.

[0075] The related information acquisition unit 133 acquires related information indicating that there is an error in the input expense information, for example, if the score is smaller than or below a threshold.

[0076] When the score acquisition unit 132 acquires a score for each element of information, the related information acquisition unit 133 may, for example, use different thresholds (different criteria) for combinations of two or more different elements to determine whether the input expense information is correct or not.

[0077] When the related information acquisition unit 133 determines whether the input expense information is correct using two scores, the element information "date" and the element information "amount," the related information acquisition unit 133 determines whether the input expense information is correct by, for example, whether the sum of the two scores acquired by the score acquisition unit 132 (for example, the sum of the two edit distances) is equal to or greater than a first threshold (for example, 1.85). This occurs, for example, when there is no input for the payee.

[0078] On the other hand, when the related information acquisition unit 133 determines whether the input expense information is correct using three scores: element information "date," element information "amount," and element information "payee," the related information acquisition unit 133 determines whether the input expense information is correct by, for example, whether the sum of the three scores acquired by the score acquisition unit 132 (for example, the sum of the three edit distances) is equal to or greater than the second threshold (for example, 2.7). This occurs, for example, when there is no input for the payee.

[0079] Furthermore, for example, the score is calculated for each of two or more types of elemental information, such as "date," "amount," and "payee." The more types of elemental information for which a score is obtained, the more preferable it is for the related information acquisition unit 133 to increase the threshold to determine whether the input expense information is correct or not.

[0080] If the score acquisition unit 132 acquires two or more scores, the related information acquisition unit 133 acquires related information indicating that there is an error in the input expense information, for example, if one or more scores are smaller than or equal to a threshold. Note that the two or more scores here usually refer to the scores of two or more individual element information.

[0081] The output unit 14 outputs various types of information. These types of information include, for example, related information, expense information, and receipt images.

[0082] The related information output unit 141 outputs the related information acquired by the related information acquisition unit 133.

[0083] The related information output unit 141 notifies the user or administrator corresponding to the input expense information of related information regarding errors in the input expense information if the score is low enough to meet predetermined conditions.

[0084] The related information output unit 141 outputs, for example, input expense information and receipt images in association. The related information output unit 141 also outputs, for example, a score in association with input expense information. Output here usually refers to storage. However, output may also be a concept that includes display on a screen, projection using a projector, printing with a printer, sound output, transmission to an external device, or delivery of processing results to other processing devices or other programs.

[0085] The related information output unit 141 outputs, for example, related information regarding representative scores.

[0086] The terminal storage unit 21, which constitutes the terminal device 2, stores various types of information. These types of information include, for example, a user identifier, input expense information, and receipt images.

[0087] The terminal reception unit 22 receives various instructions and information. These instructions and information include, for example, input expense information, receipt images, and output instructions.

[0088] The terminal processing unit 23 performs various processes. These processes include, for example, converting instructions and information received by the terminal receiving unit 22 into instructions and information for a data structure to be transmitted. Other processes include, for example, converting information received by the terminal receiving unit 25 into information for a data structure to be output.

[0089] For example, when the terminal receiving unit 22 receives input expense information and a receipt image, the terminal processing unit 23 obtains the user identifier from the terminal storage unit 21, associates the user identifier with the input expense information and the receipt image, and configures the information to be transmitted.

[0090] The terminal transmission unit 24 transmits various instructions and information. These instructions and information include, for example, input expense information, receipt images, user identifiers, and output instructions.

[0091] The terminal receiving unit 25 receives various types of information. These types of information include, for example, related information, expense information, and receipt images.

[0092] The terminal output unit 26 outputs various types of information. These types of information include, for example, related information, expense information, and receipt images.

[0093] The storage unit 11, user information storage unit 111, expense information storage unit 112, and terminal storage unit 21 are preferably made of non-volatile recording media, but can also be made of volatile recording media.

[0094] The process by which information is stored in the storage unit 11, etc. is not relevant. For example, information may be stored in the storage unit 11, etc. via a recording medium, information transmitted via a communication line, etc. may be stored in the storage unit 11, etc., or information input via an input device may be stored in the storage unit 11, etc.

[0095] The reception unit 12 and the terminal receiving unit 25 are typically implemented using wireless or wired communication means.

[0096] The processing unit 13, string acquisition unit 131, score acquisition unit 132, related information acquisition unit 133, and terminal processing unit 23 can typically be implemented using a processor, memory, etc. The processing procedures of the processing unit 13, etc., are usually implemented in software, and this software is recorded on a recording medium such as ROM. However, it may also be implemented in hardware (dedicated circuitry). The processor can be an MPU, CPU, GPU, etc., and the type is not limited.

[0097] The output unit 14, the related information output unit 141, and the terminal transmission unit 24 are typically implemented by wireless or wired communication means.

[0098] The terminal reception unit 22 can be implemented using device drivers for input means such as touch panels and keyboards, or control software for menu screens, etc.

[0099] The terminal output unit 26 may or may not be considered to include output devices such as a display or speakers. The terminal output unit 26 can be implemented using driver software for an output device, or driver software for an output device and an output device.

[0100] Next, we will explain an example of the operation of information system A. First, we will explain an example of the operation of expense inspection device 1 using the flowchart in Figure 3.

[0101] (Step S301) The reception unit 12 determines whether or not it has received input expense information and a receipt image from the terminal device 2. If it has received input expense information, it proceeds to step S302; if it has not received input expense information, it proceeds to step S309.

[0102] (Step S302) The character acquisition unit 131 performs character recognition processing on the receipt image received in step S301 and acquires one or more OCR strings. Note that the character recognition processing may be performed by an external device (not shown). Here, one OCR string is, for example, a line of text or a string in a text box.

[0103] (Step S303) The score acquisition unit 132 assigns 1 to counter i.

[0104] (Step S304) The score acquisition unit 132 determines whether or not the i-th element information exists in the input expense information received in step S301. If the i-th element information exists, the unit proceeds to step S305; otherwise, the unit proceeds to step S307.

[0105] (Step S305) The score acquisition unit 132 acquires the score of the i-th element information. An example of this element score acquisition process will be explained using the flowchart in Figure 4.

[0106] (Step S306) The score acquisition unit 132 increments counter i by 1. Return to step S304.

[0107] (Step S307) The related information acquisition unit 133 acquires related information using the score acquired in step S305. An example of this related information acquisition process will be explained using the flowchart in Figure 7.

[0108] (Step S308) The related information output unit 141 outputs the related information acquired in step S307. Return to step S301.

[0109] (Step S309) The reception unit 12 determines whether or not it has received an output instruction from the terminal device 2. If an output instruction is received, the process proceeds to step S310; otherwise, the process returns to step S301.

[0110] (Step S310) The processing unit 13 obtains one or more expense information items corresponding to the output instruction received in step S309 from the expense information storage unit 112.

[0111] (Step S311) The output unit 14 transmits one or more expense information items obtained in step S310 to the terminal device 2 that sent the output instruction. Return to step S301.

[0112] In the flowchart of Figure 3, if two or more input expense information entries are received in step S301, it is preferable to perform the processing in steps S303 to S308 for each of the two or more input expense information entries.

[0113] Furthermore, in the flowchart of Figure 3, processing is terminated by power off or processing termination interrupts.

[0114] Next, an example of the element score acquisition process in step S305 will be explained using the flowchart in Figure 4.

[0115] (Step S401) The score acquisition unit 132 performs a second element information generation process. An example of the second element information generation process will be explained using the flowchart in Figure 5. The second element information generation process is a process that generates one or more second element information using element information. It is also possible that it is not possible to generate second element information at this stage. Furthermore, the second element information can also be called element information.

[0116] (Step S402) The score acquisition unit 132 assigns 1 to counter i.

[0117] (Step S403) The score acquisition unit 132 determines whether or not the i-th element information exists among the i-th element information from step S304 and one or more second element information generated in step S401.

[0118] (Step S404) The score acquisition unit 132 assigns 1 to counter j.

[0119] (Step S405) The score acquisition unit 132 determines whether the j-th sub-information exists within the i-th element information. If the j-th sub-information exists, the process proceeds to step S406; otherwise, the process proceeds to step S409. The sub-information is, for example, an n-gram within the element information.

[0120] (Step S406) The score acquisition unit 132 acquires the j-th sub-information from the i-th element information.

[0121] (Step S407) The score acquisition unit 132 performs partial score acquisition processing. An example of partial score acquisition processing will be explained using the flowchart in Figure 6. Partial score acquisition processing is the process of acquiring the score corresponding to the j-th partial information.

[0122] (Step S408) The score acquisition unit 132 increments counter j by 1. Return to step S405.

[0123] (Step S409) The score acquisition unit 132 increments counter i by 1. Return to step S403.

[0124] (Step S410) The score acquisition unit 132 uses one or more scores acquired in step S407 to acquire the score of the i-th element information, and temporarily stores the score paired with the i-th element information in a buffer (not shown). It then returns to the higher-level processing.

[0125] The score acquisition unit 132, for example, acquires the maximum value of the 1 or greater scores acquired in step S407 as the score of the i-th element information.

[0126] Next, an example of the second element information generation process in step S401 will be explained using the flowchart in Figure 5.

[0127] (Step S501) The score acquisition unit 132 acquires a type identifier that identifies the type of the i-th element information (element information of interest) in step S304.

[0128] (Step S502) The score acquisition unit 132 assigns 1 to counter i.

[0129] (Step S503) The score acquisition unit 132 determines whether or not the i-th pattern corresponding to the type identifier obtained in step S501 exists. If the i-th pattern exists, the unit proceeds to step S504; otherwise, it returns to the higher-level processing.

[0130] (Step S504) The score acquisition unit 132 determines whether the element information of interest is a string corresponding to the i-th pattern. If it is a string corresponding to the i-th pattern, the unit proceeds to step S506; otherwise, the unit proceeds to step S505.

[0131] (Step S505) The score acquisition unit 132 uses the element information of interest to obtain a string corresponding to the i-th pattern corresponding to the type identifier obtained in step S501. This string is the second element information.

[0132] (Step S506) The score acquisition unit 132 increments counter i by 1. Return to step S503.

[0133] Next, an example of the partial score acquisition process in step S407 will be explained using the flowchart in Figure 6.

[0134] (Step S601) The score acquisition unit 132 assigns 1 to counter i.

[0135] (Step S602) The score acquisition unit 132 determines whether the i-th OCR string exists in the OCR string acquired in step S302. If the i-th OCR string exists, the process proceeds to step S603; otherwise, the process returns to the higher level.

[0136] (Step S603) The score acquisition unit 132 assigns 1 to counter j.

[0137] (Step S604) The score acquisition unit 132 determines whether it is possible to obtain a string that starts from the j-th string within the i-th OCR string and has the number of characters of the j-th partial information from step S406. If such a string is obtainable, the unit proceeds to step S605; otherwise, the unit proceeds to step S612.

[0138] (Step S605) The score acquisition unit 132 determines whether the j-th partial information from step S406 matches the string that starts from the j-th string in the i-th OCR string. If there is a match, proceed to step S606; otherwise, proceed to step S611.

[0139] (Step S606) The score acquisition unit 132 obtains the number of preceding characters. In other words, the score acquisition unit 132 obtains the number of preceding characters, which is the string obtained by removing the j-th partial information from step S406 (which can be said to be a matching string) from the element information of interest. Note that the number of preceding characters may be "0".

[0140] (Step S607) The score acquisition unit 132 obtains the OCR prefix string. That is, the score acquisition unit 132 obtains the OCR prefix string from the i-th OCR string, which is the string immediately preceding the matching string, and is the number of characters obtained in step S606. However, there are cases where the OCR prefix string cannot be obtained.

[0141] (Step S608) The score acquisition unit 132 obtains the number of characters at the end. In other words, the score acquisition unit 132 obtains the number of characters at the end of the string obtained by removing the j-th part of the information from step S406 (which can be said to be a matching string) from the element information of interest. Note that the number of characters at the end of the end may be "0".

[0142] (Step S609) The score acquisition unit 132 obtains the OCR suffix string. That is, the score acquisition unit 132 obtains the OCR suffix string from the i-th OCR string, which is the string immediately following the matched string and consists of the number of characters obtained in step S606.

[0143] (Step S610) The score acquisition unit 132 sequentially concatenates the OCR forward string, the matching string, and the OCR backward string acquired in step S607, obtains a comparison string, and temporarily stores the comparison string in a buffer (not shown). Note that there may be cases where the OCR backward string cannot be obtained at this stage.

[0144] (Step S611) The score acquisition unit 132 increments counter j by 1. Return to step S604.

[0145] (Step S612) The score acquisition unit 132 assigns 1 to counter k.

[0146] (Step S613) The score acquisition unit 132 determines whether the k-th comparison string exists in a buffer (not shown). If the k-th comparison string exists, the unit proceeds to step S614; otherwise, the unit proceeds to step S616.

[0147] (Step S614) The score acquisition unit 132 calculates the distance between the i-th element information from step S403 and the k-th comparison string, and temporarily stores this distance in a buffer (not shown) in pairs with the k-th comparison string.

[0148] (Step S615) The score acquisition unit 132 increments counter k by 1. Return to step S613.

[0149] (Step S616) The score acquisition unit 132 increments counter i by 1. Return to step S602.

[0150] In step S610 of Figure 6, it is preferable for the score acquisition unit 132 to acquire the same comparison string at the same location only once.

[0151] Furthermore, in step S614 of Figure 6, it is preferable for the score acquisition unit 132 to acquire the distance between the same comparison string and element information at the same location only once.

[0152] Next, an example of the related information acquisition process in step S307 will be explained using the flowchart in Figure 7.

[0153] (Step S701) The related information acquisition unit 133 determines whether the score of each element information accumulated in step S410 satisfies the normal conditions. If the normal conditions are met, the unit proceeds to step S702; otherwise, the unit proceeds to step S703.

[0154] (Step S702) The related information acquisition unit 133 generates related information that includes the input expense information and receipt image received in step S301. It returns to the higher-level processing. It is preferable that the related information is associated with a user identifier.

[0155] (Step S703) The related information acquisition unit 133 acquires abnormal element information.

[0156] (Step S704) The related information acquisition unit 133 generates related information that includes the abnormal element information acquired in step S703, or the type identifier of said element information. It returns to the higher-level processing. This related information indicates that the received input expense information is abnormal.

[0157] Next, an example of the operation of terminal device 2 will be described. The terminal reception unit 22 of terminal device 2 receives input expense information and a receipt image. Next, the terminal processing unit 23 associates the input expense information, receipt image, and user identifier of terminal storage unit 21 received by the terminal reception unit 22 and configures the information to be transmitted. The terminal transmission unit 24 transmits this information to the expense inspection device 1.

[0158] Next, the terminal receiving unit 25 of the terminal device 2 receives relevant information from the expense inspection device 1 indicating whether the input expense information is valid or not. Then, the terminal output unit 26 outputs this relevant information.

[0159] Furthermore, the terminal receiving unit 22 of the terminal device 2 receives an output instruction. Then, the terminal transmitting unit 24 transmits the output instruction to the expense inspection device 1. Next, the terminal receiving unit 25 receives one or more expense information items corresponding to the output instruction from the expense inspection device 1. Then, the terminal output unit 26 outputs one or more expense information items.

[0160] Furthermore, for example, the terminal receiving unit 25 of the terminal device 2 used by an administrator or accounting staff member receives relevant information indicating that incorrect expense information has been entered by another user. The terminal output unit 26 then outputs this relevant information.

[0161] The following describes a specific example of the operation of information system A in this embodiment.

[0162] Currently, the user information storage unit 111 of the expense inspection device 1 stores the user information management table shown in Figure 8. The user information management table is a table for managing user information. The user information management table manages one or more records that have an "ID," a "user identifier," and "user attribute values." In this case, the "user attribute values" include "email address," "name," and "supervisor email address." The "ID" is information that identifies the record. The "email address" is the email address of the user identified by the user identifier. The "supervisor email address" is the email address of the supervisor of the user identified by the user identifier.

[0163] In addition, an expense information management table having the structure shown in FIG. 9 is stored in the expense information storage unit 112. The expense information management table is a table for managing expense information. The expense information management table manages one or more records having "ID", "user identifier", "expense information", "receipt image", and "score". "ID" is information for identifying a record. "Expense information" here has "date", "amount", and "payee".

[0164] Furthermore, a pattern management table shown in FIG. 10 is stored in the storage unit 11. The pattern management table is a table for managing notation patterns (hereinafter, appropriately referred to as "patterns") for each type of element information constituting expense information. The pattern management table manages one or more records having "type identifier", "pattern identifier", and "notation pattern". "Type identifier" is information for identifying the type of element information constituting the expense information. "Pattern identifier" is identification information for the notation pattern within the same type. "Notation pattern" here is a regular expression, but the definition method is not limited.

[0165] The character string according to the notation pattern of pattern identifier "1" of type identifier "date" is, for example, "2020 / 8 / 27". Also, the character string according to the notation pattern of pattern identifier "2" of type identifier "date" is, for example, "August 27, 2020". Furthermore, the character string according to the notation pattern of pattern identifier "3" of type identifier "date" is, for example, "August 27, 2020".

[0166] The character string according to the notation pattern of pattern identifier "1" of type identifier "amount" is, for example, "940 yen". Also, the character string according to the notation pattern of pattern identifier "2" of type identifier "amount" is, for example, "￥940".

[0167] The string following the notation pattern of pattern identifier "1" for the type identifier "Payee" is, for example, "Tsubame Taxi". The string following the notation pattern of pattern identifier "2" for the type identifier "Payee" is, for example, "(Ltd.) Tsubame Taxi". The string following the notation pattern of pattern identifier "3" for the type identifier "Payee" is, for example, "Tsubame Taxi Limited Company". The string following the notation pattern of pattern identifier "4" for the type identifier "Payee" is, for example, "Tsubame Taxi (Co., Ltd.)". The string following the notation pattern of pattern identifier "3" for the type identifier "Payee" is, for example, "Tsubame Taxi Co., Ltd.".

[0168] In this situation, the following two specific examples will be explained. Specific example 1 is the case where one receipt image contains the image of one receipt. Specific example 2 is the case where one receipt image contains the images of two receipts.

[0169] (Specific example 1) Assume that user "Yamada Ao" entered the following expense information into his terminal device 2: "<Date> 2020 / 8 / 27 <Amount> 940 <Payee> Tsubame Taxi Co., Ltd." along with the receipt image shown in Figure 11. The input screen and input interface for this information are not specified.

[0170] The terminal receiving unit 22 of the terminal device 2 then receives the input expense information and the receipt image. Next, the terminal processing unit 23 associates the input expense information received by the terminal receiving unit 22, the receipt image, and the user identifier "U001" in the terminal storage unit 21 to compose the information to be transmitted. Next, the terminal transmission unit 24 transmits the input expense information "<Date> 2020 / 8 / 27 <Amount> 940 <Payee> Tsubame Taxi Co., Ltd.", the receipt image in Figure 11, and the user identifier "U001" to the expense inspection device 1.

[0171] Next, the reception unit 12 of the expense inspection device 1 receives the input expense information "<Date> 2020 / 8 / 27 <Amount> 940 <Payee> Tsubame Taxi Co., Ltd.", the receipt image shown in Figure 11, and the user identifier "U001".

[0172] In this specific example 1, the information from receipt images 1101, 1102, and 1103 in Figure 11 will be compared with the entered expense information.

[0173] Next, the string acquisition unit 131 performs character recognition processing on the receipt image in Figure 11. Then, the string acquisition unit 131 acquires one or more OCR strings as shown in Figure 12. Here, each OCR string is the string for each line.

[0174] Next, the score acquisition unit 132 acquires scores for each of the three elements of the received input expense information, "<Date> 2020 / 8 / 27", "<Amount> 940", and "<Payee> Tsubame Taxi Co., Ltd.", in the following order.

[0175] In other words, the score acquisition unit 132 first acquires the score of the element information "2020 / 8 / 27" of the type identifier "Date" as follows.

[0176] The score acquisition unit 132 performs a second element information generation process according to the notation pattern (see Figure 10) that is paired with the type identifier "date," and obtains "August 27, 2020," "August 27, 2020," etc. Here, the score acquisition unit 132 obtains the received element information "2020 / 8 / 27," the generated element information "August 27, 2020," "August 27, 2020," etc.

[0177] Next, the score acquisition unit 132 acquires one or more sub-informations for each element information. Here, the sub-information is assumed to be a string of 3-grams contained in the element information.

[0178] Then, the score acquisition unit 132 first obtains partial information of two or more characters while shifting one character at a time from the first element information "2020 / 8 / 27". That is, the score acquisition unit 132 sequentially obtains partial information "202", "020", "20 / ", "0 / 8", " / 8 / ", "8 / 2", " / 27" for the element information "2020 / 8 / 27".

[0179] Then, the score acquisition unit 132 checks whether any of the above partial information is included in each of the one or more OCR character strings shown in FIG. 12. And the score acquisition unit 132 determines that none of the above partial information is included in each of the one or more OCR character strings shown in FIG. 12. That is, the score acquisition unit 132 cannot obtain a comparison character string for the element information "2020 / 8 / 27".

[0180] Next, the score acquisition unit 132 sequentially obtains partial information "202", "020", "20 years", "0 years 0", "years 08", "0 years 0", "years 08", "08 months", "8 months 2", "months 27", "27 days" for the next element information "August 27, 2020".

[0181] Next, the score acquisition unit 132 sequentially checks whether any of the above partial information is included in each of the one or more OCR character strings shown in FIG. 12. And the score acquisition unit 132 detects that "years 08" is included in the OCR character string "Date 20 still? years 08 months 27 days" shown in FIG. 12.

[0182] Next, the score acquisition unit 132 obtains the number of characters in front of the character example "2020" in front of the partial information, which is the number of characters in front of the partial information "years 08" removed from the target element information "August 27, 2020", and the number of characters in front is "4".

[0183] Next, the score acquisition unit 132 obtains the OCR front character string "20 still?", which is the character string of the obtained number of characters in front "4", from the OCR character string "Date by? years 08 months 27 days".

[0184] Next, the score acquisition unit 132 obtains a character string obtained by removing the partial information "year 08" from the element information "August 27, 2020" to be focused on, and obtains the number of characters "4" after the character example "month 27th" behind the partial information.

[0185] Next, the score acquisition unit 132 obtains the OCR trailing character string "month 27th", which is the character string of the obtained number of trailing characters "4", from the OCR character string "date 20?? year 08 month 27th".

[0186] Next, the score acquisition unit 132 concatenates the OCR leading character "20??", the matching character string "year 08", and the OCR trailing character string "month 27th" to obtain a comparison character string "20?? year 08 month 27th".

[0187] Next, the score acquisition unit 132 calculates the edit distance (d1) between the element information "August 27, 2020" and the comparison character string "20?? year 08 month 27th", and accumulates the edit distance (d1) in association with the element information ("2020 / 8 / 27" or "August 27, 2020").

[0188] The score acquisition unit 132 performs the above processing on other partial information as well. Also, the score acquisition unit 132 performs the above processing on other element information (such as "August 27, 2020 of Reiwa 2"). Then, during the above processing, the score acquisition unit 132 obtains the edit distance (d1) for the element information "2020 / 8 / 27" with respect to the type identifier "date" while obtaining a set of partial information and comparison character strings as shown in FIG. 13, and accumulates it.

[0189] Furthermore, the score acquisition unit 132 performs the same processing as above for the element information "940" corresponding to the type identifier "amount". In other words, the score acquisition unit 132 generates second element information such as "940 yen" and "¥940" for the element information "940". Then, the score acquisition unit 132 calculates the edit distance for each element information such as "940", "940 yen", and "¥940" using the algorithm described above. In other words, the score acquisition unit 132 obtains and stores the edit distance (d2=0) between the comparison string "940" obtained from the OCR string "Basic Ice?? Approximately 940 yen" and the element information "940".

[0190] Furthermore, the score acquisition unit 132 performs the same processing as above for the element information "(Ltd.) Tsubame Taxi" for the type identifier "Payee". Then, the score acquisition unit 132 obtains and stores the edit distance (d3) between the comparison string "(f) Tsubame Tact" obtained from the OCR string "(f) Tsubame Tact" and the element information "(Ltd.) Tsubame Taxi".

[0191] Through the above process, the score acquisition unit 132 was able to obtain the edit distance for each of the three element information.

[0192] Next, the score acquisition unit 132 obtained a score of "0.90" using a decrease function that takes the edit distance for each of the three element information as a parameter.

[0193] Next, the related information acquisition unit 133 determines that the score of the element information satisfies the normal condition because the score is above the threshold (for example, 0.8 or higher). Here, the normal condition is defined as "the score being above the threshold."

[0194] Next, the related information acquisition unit 133 constitutes related information having the following: "Input expense information "<Date> 2020 / 8 / 27 <Amount> 940 <Payee> Tsubame Taxi Co., Ltd." and receipt image (Figure 11)".

[0195] Next, the related information output unit 141 associates the related information "Input expense information "<Date> 2020 / 8 / 27 <Amount> 940 <Payee> Tsubame Taxi Co., Ltd." and receipt image (Figure 11)" and the score "0.9" with the user identifier "U001" and stores it in the expense information management table (see Figure 9). An example of such a record is the record with "ID=268" in Figure 9. The receipt image (Figure 11) is stored as a file with the generated filename "f268 / jpg".

[0196] (Specific example 2) User "Inoue B-ko" entered the following expense information into her terminal device 2: "<Date>2020 / 9 / 2 <Amount>660 <Payee>Aomori Prefectural Road Corporation", "<Date>2020 / 9 / 2 <Amount>160 <Payee>Aomori Prefectural Road Corporation", and the receipt image shown in Figure 14.

[0197] Then, the terminal reception unit 22 of Inoue B's terminal device 2 receives the two pieces of input expense information and the receipt image. Next, the terminal processing unit 23 associates the input expense information, the receipt image, and the user identifier "U002" in the terminal storage unit 21 that were received by the terminal reception unit 22, and configures the information to be transmitted. Next, the terminal transmission unit 24 transmits the input expense information, the receipt image in Figure 14, and the user identifier "U002" to the expense inspection device 1.

[0198] Next, the reception unit 12 of the expense inspection device 1 receives two entries of expense information, the receipt image shown in Figure 14, and the user identifier "U002".

[0199] In this specific example 2, the information from receipt images 1401, 1402, 1403, 1404, 1405, and 1406 in Figure 14 will be compared with the two sets of input expense information.

[0200] Next, the string acquisition unit 131 performs character recognition processing on the receipt image in Figure 14. Then, the string acquisition unit 131 acquires one or more OCR strings as shown in Figure 15. Here, each OCR string is the string for each line. Furthermore, the string acquisition unit 131 does not distinguish between the two receipt images, but performs character recognition processing on the image file in Figure 14.

[0201] Next, the score acquisition unit 132 first performs the processing described in Specific Example 1 on the first input expense information "<Date>2020 / 9 / 2 <Amount>660 <Payee>Aomori Prefectural Road Corporation" using one or more OCR strings shown in Figure 15, and acquires a score for the first input expense information.

[0202] Next, the related information acquisition unit 133 determines that the score of the first input expense information meets the normal conditions.

[0203] Next, the related information acquisition unit 133 constitutes related information that includes "<Date> 2020 / 9 / 2 <Amount> 660 <Payee> Aomori Prefectural Road Corporation" and an image of the receipt (Figure 14).

[0204] Next, the related information output unit 141 associates the related information with the score (in this case, "0.98") with the user identifier "U002" and stores it in the expense information management table (see Figure 9). An example of such a record is the record with "ID=351" in Figure 9.

[0205] Furthermore, the score acquisition unit 132 performs the same process on the second input expense information, "<Date>2020 / 9 / 2 <Amount>160 <Payee>Aomori Prefectural Road Corporation", and acquires a score for the second input expense information.

[0206] Next, the related information acquisition unit 133 determines that the score of the second input expense information meets the normal conditions.

[0207] Next, the related information acquisition unit 133 constitutes related information that includes "<Date> 2020 / 9 / 2 <Amount> 160 <Payee> Aomori Prefectural Road Corporation" and an image of the receipt (Figure 14).

[0208] Next, the related information output unit 141 associates the related information with the score (in this case, "0.95") with the user identifier "U002" and stores it in the expense information management table (see Figure 9). An example of such a record is the record with "ID=352" in Figure 9.

[0209] In this example, we were able to process cases where multiple receipt images were included. Furthermore, we were able to process multiple input expense information entries in this example.

[0210] As described above, according to this embodiment, it becomes possible to properly verify the entered expense information.

[0211] Furthermore, according to this embodiment, it becomes possible to properly verify the information on expenses entered using receipt images that include two or more receipts.

[0212] Furthermore, according to this embodiment, the string acquisition unit 131 acquires one or more OCR strings from each of the two or more receipt images without determining the boundaries between the images of the two or more receipts. This makes it possible to easily process and appropriately verify the expense information entered using receipt images that include two or more receipts.

[0213] Furthermore, according to this embodiment, even when there is a mix of inputs for two or more expenses and receipt images containing two or more receipts, proper verification becomes possible.

[0214] Furthermore, according to this embodiment, appropriate notifications can be made when information about expenses that are likely to be inappropriate is entered.

[0215] In this embodiment, the terminal device 2 may perform some of the processing that the expense inspection device 1 performs. For example, the terminal device 2 may perform processing for one or more of the components among the string acquisition unit 131, the score acquisition unit 132, and the related information acquisition unit 13. In other words, the method of dividing the processing between the expense inspection device 1 and the terminal device 2 is not limited. The expense inspection device 1 may also be a standalone device.

[0216] Furthermore, the processing in this embodiment may be implemented in software. This software may be distributed by software download or the like. Alternatively, this software may be recorded on a recording medium such as a CD-ROM and distributed. This also applies to other embodiments in this specification. The software that implements the expense inspection device 1 in this embodiment is the following program. In other words, this program is a program that causes a computer to function as a reception unit that receives input expense information and receipt images corresponding to expenses entered by the user, a string acquisition unit that acquires one or more OCR strings which are the results of OCR on the receipt images, a score acquisition unit that compares the input expense information with the one or more OCR strings and acquires a score regarding the certainty of the input expense information, and a related information output unit that outputs related information regarding the score.

[0217] Figure 16 also shows the external appearance of a computer that executes the program described herein to realize the various embodiments of the cost inspection device 1, etc. described above. The embodiments described above can be realized with computer hardware and computer programs executed thereon. Figure 16 is an overview of this computer system 300, and Figure 17 is a block diagram of the system 300.

[0218] In Figure 16, the computer system 300 includes a computer 301 with a CD-ROM drive, a keyboard 302, a mouse 303, and a monitor 304.

[0219] In Figure 17, the computer 301 includes, in addition to the CD-ROM drive 3012, an MPU 3013, a bus 3014 connected to the CD-ROM drive 3012, a ROM 3015 for storing programs such as boot-up programs, a RAM 3016 connected to the MPU 3013 for temporarily storing instructions for application programs and providing temporary storage space, and a hard disk 3017 for storing application programs, system programs, and data. Although not shown here, the computer 301 may further include a network card for providing connectivity to a LAN.

[0220] The program that causes the computer system 300 to execute the functions of the expense inspection device 1, etc., as described above, may be stored on CD-ROM 3101, inserted into CD-ROM drive 3012, and then transferred to hard disk 3017. Alternatively, the program may be transmitted to computer 301 via a network (not shown) and stored on hard disk 3017. The program is loaded into RAM 3016 during execution. The program may also be loaded directly from CD-ROM 3101 or the network.

[0221] The program does not necessarily have to include an operating system (OS) or third-party program that causes the computer 301 to execute functions such as the expense inspection device 1 of the above-described embodiment. The program only needs to include the instruction portion that calls appropriate functions (modules) in a controlled manner and obtains the desired result. How the computer system 300 operates is well known, so a detailed explanation is omitted.

[0222] In the above program, steps such as sending information and receiving information do not include hardware-based processing, such as processing performed by a modem or interface card in the transmission step (processing that can only be performed by hardware).

[0223] Furthermore, the computer running the above program may be a single computer or multiple computers. In other words, it may perform centralized processing or distributed processing.

[0224] Furthermore, it goes without saying that in each of the above embodiments, two or more communication means present in a single device may be physically implemented in a single medium.

[0225] Furthermore, in each of the above embodiments, each process may be implemented by centralized processing by a single device, or by distributed processing by multiple devices.

[0226] It goes without saying that the present invention is not limited to the embodiments described above, and various modifications are possible, all of which are also included within the scope of the present invention. [Industrial applicability]

[0227] As described above, the expense inspection device 1 according to the present invention has the effect of enabling appropriate verification of expense inputs and is useful as a server for inspecting expense information. [Explanation of Symbols]

[0228] 1. Cost Inspection Device 2 Terminal devices 11 Storage Unit 12 Reception Department 13 Processing Unit 14 Output section 21 Terminal storage section 22 Terminal Reception Section 23 Terminal Processing Unit 24 Terminal transmission unit 25 Receiving part of the terminal 26 Terminal output section 111 User Information Storage Unit 112 Expense Information Storage Unit 131 String acquisition part 132 Score Acquisition Section 133 Related Information Acquisition Department 141 Related Information Output Unit

Claims

1. A reception desk that receives expense information entered by the user and corresponding receipt images for those expenses, A string acquisition unit that acquires one or more OCR strings which are the result of OCR on the aforementioned receipt image, A score acquisition unit compares the input expense information with the one or more OCR strings and obtains a score regarding the accuracy of the input expense information. The system comprises a related information output unit that outputs related information regarding the score, The aforementioned receipt image contains two or more receipts. The aforementioned score acquisition unit, An expense inspection device that acquires two or more partial pieces of information, which are n-gram strings that constitute the input expense information, detects the string with the smallest distance from each of the two or more partial pieces of information from the one or more OCR strings, and obtains a score using the distance to each of the detected strings with the smallest distance.

2. A receiving unit that receives input expense information and receipt images corresponding to expenses entered by the user, A string acquisition unit that acquires one or more OCR strings which are the result of OCR on the aforementioned receipt image, A score acquisition unit compares the input expense information with the one or more OCR strings and obtains a score regarding the accuracy of the input expense information. The system comprises a related information output unit that outputs related information regarding the score, The aforementioned receipt image contains two or more receipts. The aforementioned input expense information includes two or more pieces of information from date, amount, and payee. The aforementioned score acquisition unit, The two or more pieces of information are compared with the one or more OCR strings, a score is obtained for each of the two or more pieces of information, and a representative score, which is a representative value of the two or more scores, is obtained. The aforementioned related information output unit is: An expense inspection device that outputs relevant information regarding the aforementioned representative score.

3. The string acquisition unit, The expense inspection device according to claim 1 or claim 2, which obtains one or more OCR strings from each image of the two or more receipts without determining the boundaries of the images of the two or more receipts.

4. The aforementioned related information output unit is: The expense inspection device according to any one of claims 1 to 3, wherein if the score indicates information with a low probability of satisfying predetermined conditions, it notifies the user or administrator corresponding to the input expense information of relevant information regarding errors in the input expense information.

5. The aforementioned related information output unit is: The expense inspection device according to any one of claims 1 to 4, which outputs related information regarding the score in association with the input expense information.

6. A cost inspection method realized by a reception unit, a string acquisition unit, a score acquisition unit, and a related information output unit, The aforementioned reception unit receives input expense information and corresponding receipt images for expenses entered by the user in a reception step, The string acquisition unit performs a string acquisition step of acquiring one or more OCR strings which are the results of OCR on the receipt image, The score acquisition step involves the score acquisition unit comparing the input expense information with the one or more OCR strings and acquiring a score regarding the accuracy of the input expense information. The related information output unit comprises a related information output step that outputs related information relating to the score, The aforementioned receipt image contains two or more receipts. In the score acquisition step, An expense inspection method comprising: obtaining two or more partial information pieces, which are n-gram strings that constitute the input expense information; detecting the string with the smallest distance from each of the two or more partial information pieces from the one or more OCR strings; and obtaining a score using the distance to each of the detected strings with the smallest distance.

7. An expense inspection method realized by a reception unit, a string acquisition unit, a score acquisition unit, and a related information output unit, The aforementioned reception unit receives input expense information and corresponding receipt images for expenses entered by the user in a reception step, The string acquisition unit performs a string acquisition step of acquiring one or more OCR strings which are the results of OCR on the receipt image, The score acquisition step involves the score acquisition unit comparing the input expense information with the one or more OCR strings and acquiring a score regarding the accuracy of the input expense information. The related information output unit comprises a related information output step that outputs related information relating to the score, The aforementioned receipt image contains two or more receipts. The aforementioned input expense information includes two or more pieces of information from date, amount, and payee. In the score acquisition step, The two or more pieces of information are compared with the one or more OCR strings, a score is obtained for each of the two or more pieces of information, and a representative score, which is a representative value of the two or more scores, is obtained. In the aforementioned related information output step, A cost inspection method that outputs related information regarding the aforementioned representative score.

8. Computers, A program for causing a cost inspection device to function as described in any one of claims 1 to 5.

Images