WO2016009636A1

WO2016009636A1 - Account processing device, account processing method, and account processing program

Info

Publication number: WO2016009636A1
Application number: PCT/JP2015/003526
Authority: WO
Inventors: 登史文坂本
Original assignee: フリー株式会社
Priority date: 2014-07-14
Filing date: 2015-07-13
Publication date: 2016-01-21
Also published as: JP2016021147A; KR20170032327A; JP5936284B2

Abstract

Provided are a simple and inexpensive account processing device, account processing method, and account processing program for business operators for whom the timing restrictions to be followed by the accrual basis of accounting are lenient, such as small and medium-sized businesses and sole proprietors. An account processing device (100) is provided with a Web server (110), a DB (120) connected to the Web server (110) by VPN technology, and a scraping server (130) connected to the DB (120) by VPN technology. The Web server (110) identifies Web statement data loaded by the scraping server (130) on an individual transaction basis, and automatically journalizes each transaction under a specific account title on the basis of the description of the transaction content of each transaction. At this time, the Web server (110) breaks the description of the transaction content into morphemes, and estimates the account title indicated by the description of the transaction content with reference to the frequency of appearance of one or a plurality of account titles associated with each morpheme. Created journalization data include at least the date, transaction content, amount, and account title, for example.

Description

Accounting processing apparatus, accounting processing method, and accounting processing program

The present invention relates to an accounting apparatus, accounting method, and accounting program, and more particularly to an accounting device, accounting method, and accounting program for performing accounting processing by cloud computing.

Business accounting is based on the practice of corporate accounting principles that have been developed as a custom, and one of the principles is the principle of “accrual” for the income statement (P / L). Accrual basis is the principle that an accrual basis should be recorded when the fact of income or expense occurs, regardless of the actual income and expenditure of cash. This is a concept opposite to cash basis, which recognizes revenue and expenses at the time of cash delivery.

Each company calculates income and expenditure for a certain period, performs a settlement to calculate profit or loss, and financial statements such as balance sheet (B / S) and income statement (P / L) prepared in the settlement Is stipulated to be announced based on the provisions of the Companies Act. In particular, in Japan, financial instruments trading law application companies have strong time constraints on the settlement of accounts, and quarterly settlement is required for listed companies, and quarterly settlement or semiannual interim settlement is required even for unlisted companies. The need for settlement is the same in other countries.

In this way, if periodic settlement of accounts is necessary and the creation of a profit and loss statement is essential under the accrual principle, Instead of creating a profit and loss statement in retrospect, it is necessary to always process the income and expenses generated in daily operations on a daily basis in accordance with the accrual principle. That is how it is processed.

For example, in a large company, there is an in-house system for accounting processing, and the person in charge of each department conducts daily work as a journal data by registering the transactions that have occurred into account items. Accumulating financial journals (financial statements) in each fiscal period is possible by accumulating the journal data thus created and creating journal data and general ledger data. Commercial accounting software is often used to create financial statements such as balance sheets and profit and loss statements from the generated general ledger data.

Although commercially available accounting software has the functions necessary to cope with such corporate accounting practices, it has the disadvantage of being general-purpose software that can meet the demands of various companies.

In other words, not all companies that require accounting processing are required to have a sense of speed in daily accounting based on the accrual principle. Applicants have found that, when focusing on small and medium-sized enterprises, individual business owners, etc., rather than large enterprises, commercially available accounting software cannot provide a function that suits the actual situation. Although it is common for small and medium-sized enterprises and individual business owners, including those in Japan, to make a settlement only once a year, it is necessary to make public announcements without delay in accordance with the provisions of the law. Sometimes it is sufficient to obtain journal data and general ledger data in accordance with the accrual principle, and it may be said that the daily journal processing is unnecessary.

The present invention has been made in view of such problems, and the purpose of the present invention is simple and easy for businesses such as small and medium-sized enterprises, individual business owners, etc., who are loosely timed to follow accrual principles. An object is to provide an inexpensive accounting processing apparatus, accounting processing method, and accounting processing program.

In order to achieve such an object, according to a first aspect of the present invention, there is provided an accounting processing apparatus for performing accounting processing by cloud computing, comprising a web server that provides cloud computing to a user, The web server identifies the web detail data for each transaction, segments each transaction into keywords, and describes the frequency of appearance of one or more account items associated with each keyword. The journal data is automatically entered into a specific account, and journal data including at least the date, transaction contents, amount, and account is created. The created journal data is stored on a computer that the user accesses the web server. It is transmitted and displayed as a journal processing screen on the web browser of the computer.

The second aspect of the present invention is characterized in that, in the first aspect, the frequency of appearance of one or more account items associated with each keyword is normalized for each keyword.

According to a third aspect of the present invention, in the first or second aspect, the journal sums up the appearance frequency of each keyword as a score for each account item and automatically sets the highest score account item. It is characterized by journalizing.

The fourth aspect of the present invention is characterized in that in any of the first to third aspects, a keyword consisting of only one symbol or number is not considered in the total score.

The fifth aspect of the present invention is characterized in that, in any of the first to fourth aspects, the journal processing screen has a menu for changing the account item.

According to a sixth aspect of the present invention, in the fifth aspect, when a user selects and corrects an automatically journalized account item from the menu, the description of the transaction content and the corrected account item The association is stored as a user rule for each user in the storage unit of the transaction processing apparatus.

The seventh aspect of the present invention is characterized in that, in the sixth aspect, the appearance frequency is determined based on user rules of a plurality of users stored in the storage unit.

According to an eighth aspect of the present invention, in the seventh aspect, the appearance frequency is obtained by dividing the description of the transaction content included in each user rule into keywords, and counting the number of appearances of the corrected account item for each keyword. It is characterized by being determined by incrementing.

A ninth aspect of the present invention is characterized in that, in any of the first to eighth aspects, the keyword is a morpheme.

The tenth aspect of the present invention is characterized in that, in the ninth aspect, segmentation into morphemes is performed without requiring domain processing.

An eleventh aspect of the present invention is an accounting processing method for performing accounting processing by cloud computing provided by a web server, wherein the web server identifies web detailed data for each transaction, and the web The server automatically divides each transaction into a specific account by segmenting the transaction description of each transaction into keywords and referring to the frequency of occurrence of one or more account items associated with each keyword. Journalizing step, and the web server creating journal data including at least a date, transaction content, amount, and account item, and the created journal data is sent to a computer from which a user accesses the web server. It is transmitted and displayed as a journal processing screen on the web browser of the computer.

A twelfth aspect of the present invention is an accounting processing program for performing accounting processing by cloud computing provided by a web server, the step of identifying web detailed data for each transaction in the web server, Categorizing the description of transaction details of each transaction into keywords, referring to the frequency of appearance of one or more account items associated with each keyword, and automatically journaling to specific account items; Creating journal data including at least a date, transaction content, amount, and account, and the created journal data is transmitted to a computer where a user accesses the web server, and the web browser of the computer Accounting processing program for executing a method characterized by being displayed as a journal processing screen A.

According to the present invention, each transaction identified from the Web detailed data is segmented into keywords by describing the transaction contents of each transaction, and the frequency of appearance of one or more account items associated with each keyword is referred to. By automatically translating into specific account items, a simple and inexpensive accounting processing device, accounting processing method and accounting processing program are provided to businesses whose time constraints to follow the accrual principle are moderate. can do.

It is a figure which shows the accounting processing apparatus which concerns on one Embodiment of this invention. It is a figure which shows the journalizing process screen which concerns on one Embodiment of this invention. It is a figure which shows the journalizing process screen at the time of correction of the account item classified automatically. It is a figure which shows the list of the created journal data.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

(Outline of the present invention)
In the present invention, journal data is created by analyzing journal details (online / web statement data) and performing journal processing after the fact, not on a daily basis.

FIG. 1 shows an accounting processing apparatus according to an embodiment of the present invention. The accounting processing apparatus 100 includes a web server 110 that provides users with cloud computing, a database 120 connected to the web server 110 using virtual private network (VPN) technology, and a scraping server 130 connected to the database 120 using VPN technology. With.

The user accesses the URL of the web server 110 using the web browser of the computer 140 such as a PC, a smartphone, or a tablet. The web server 110 identifies the captured web detail data for each transaction, and automatically translates each transaction into a specific account item based on a description of the transaction contents of each transaction. Journal entries such as income when the amount is positive and expenditure when the amount is negative are possible. At this time, the web server 110 divides the description of the transaction contents into morphemes and refers to the occurrence rate of one or a plurality of account items associated with each morpheme. Guess the account represented by. The created journal data includes, for example, at least a date, transaction contents, an amount, and an account item, and accumulation of such journal data becomes journal data.

Here, “morpheme” means the smallest linguistic unit having meaning, and will be described below using the term. However, in the present invention, the term is not necessarily the smallest linguistic unit, but is a keyword in a wider linguistic unit. The segmentation may be performed.

The appearance frequency and journal data may be stored in the web server 110 or may be stored in the database 120.

The web server 110 transmits an HTML program including the created journal data to the computer 140, which is displayed as a journal processing screen by the web browser of the computer 140 and presented to the user. Each function of the present invention can be implemented by Ruby and MySQL, and a part thereof can be implemented by JavaScript (registered trademark) in HTML (for example, a combo box 301 in FIG. 3 described later). That is, the web server 110 operates as a means for executing each function by executing a program described in a language such as Ruby, and may be configured by a single server or a plurality of servers. .

FIG. 2 shows a journalizing processing screen according to an embodiment of the present invention. The captured Web detailed data is identified for each transaction, and for each transaction, a date column 201 (“2014/6/22”), an amount column 202 (“-1,530”), and a summary column 203 (“GU ( Registered trademark) ”), account item column 204 (“ consumables expenses ”), a summary tag column (not shown) that provides a separate category from the account item, and the like.

Although it is possible for the user to upload the web specification data by himself / herself, a file in a format such as CSV can be uploaded by himself / herself, but in one embodiment of the present invention, a web specification data acquisition unit that automatically acquires from the Internet is further provided. be able to. Specifically, a scraping server 130 connected to the database 120 by the VPN technology is arranged, a user registers, and an account number and a password stored in the web server 110 or the database 120 are read out. 151, various types of web detail data can be acquired from a credit card company 152, a settlement company (eg, electronic money Suica (registered trademark)) (not shown), or the like. The acquired web detail data is stored in the web server 110 or the database 120 and is read out during the journalizing process.

Further, instead of web scraping, the web detailed data may be acquired by photographing a receipt or the like with a camera and transmitting it to the web server 110. At this time, if it is implemented as an application for a mobile terminal such as a smartphone or a tablet, when a receipt or the like is read into the application, it can be automatically taken into the user's account. In order to promote application development, an API for capturing web detailed data can also be provided. In addition, electronic data such as receipts may be received by e-mail. In this case, if the received address includes an account associated with the user's account, the electronic data such as the receipt is analyzed. It is also possible to import as web detailed data. Electronic data includes PDF, image files, text files, and the like. Further, the web server 110 may receive the data and analyze it with a mail server (not shown) instead of receiving it, and transmit the obtained web detail data to the web server 110. In addition to receipts, it is also possible to record that a debt has occurred by taking a bill with a camera and sending it to the web server 110. For example, when an invoice is received, it is converted into a liability by taking it digitalized by camera shooting and OCR, journalizing it as a liability, and registering it as a reminder. When the payment is made, it is possible to manage which payment schedule is paid by matching between the bank statement and the payment schedule. There is a need to easily cancel such payment schedules (accounts payable, accounts payable, etc.) and to deal with accrualism, and image capture of receipts is also a useful function. In addition, such image capture such as receipts and invoices allows image data to be linked to each journal and facilitates voucher management. In addition to general accounting documents such as receipts, we analyze accounting data associated with users on the Internet, such as accounting data described in e-mails and accounting data described in SNS posts. May be included in the user's account. By doing in this way, management of expenditure becomes easy when paying with cash.

In many cases, a plurality of types of web specification data provided by a plurality of financial institutions, credit card companies, settlement companies, etc. may be captured by the web server 110 and displayed on the transaction processing screen 200 for each type. You may display in order of a date, as shown in FIG.

In FIG. 2, for each transaction, the user confirms the displayed contents and presses the “Register” button to create and store journal data, but “Register” The journal data automatically created by the web server 110 may be stored as it is without pressing the button, and only necessary corrections may be performed on the transaction processing screen 200.

Fig. 3 shows the journal processing screen when correcting automatically classified account items. When the account item column 204 is clicked, the journal entry processing screen 200 displays a combo box 301 such as a pull-down menu showing other options. By adding help 302 to each option, it is possible to easily select an appropriate account item without knowledge of accounting. If the combo box 301 is nested interactively as shown, ie, the lower choices are shown according to the user's choice, the user can be sensibly navigated to the appropriate account. Can be put on.

The corrected result is stored in the web server 110 or the database 120 as a user rule for each user, and the account item corrected for the transaction content can be displayed from the next time. The correction results accumulated in this manner are used to generate a learned database (learned DB) by machine learning of the appearance frequency of one or more account items associated with each morpheme, as will be described later. This is possible because it is a cloud technology, not a standalone application.

The description tag field, etc. can be automatically assigned and displayed as well as the account item. In that case, the combo box such as a pull-down menu is displayed and corrected in the same way as the account item field 204. Results can be saved and used.

The journal data created as described above can be displayed as a list on the web browser of the computer 140 as shown in FIG.

(Details of journal processing)
In the present invention, the inference / inference of the account item represented by the description of the transaction refers to the appearance frequency of one or more account items associated with each morpheme by segmenting the description of the transaction content into morphemes. The account item represented by the transaction description is estimated.

Suppose here that the transaction description is “Sompo Japan Insurance (registered trademark) premium (Internet)”. When this description is divided into morphemes, it is divided into five categories: “damage”, “insurance”, “Japan”, “insurance premium”, and “Internet”. For each morpheme, the web server 110 reads the appearance frequency of one or more account items associated with the morpheme from the learned database. For example, the appearance frequency of each morpheme is as follows. For simplicity, those with an appearance probability of less than 0.01 are omitted.

In the above example, the appearance frequency of each morpheme is the appearance probability normalized so that the sum is 1.0. If normalization is performed, the accuracy can be improved. For example, if the description “○ × Pharmacy Kanda Station East Exit Store” appears in the summary column 203, adding the number of appearances without normalization, the “station” is better than “○ ×” or “pharmacy”. Since the number of times included in the Web detailed data is overwhelmingly large, “travel expense transportation cost”, which is an account item estimated from the morpheme “station”, is assigned. The estimation result is pulled by the morpheme that is included in the web specification data a large number of times. On the other hand, when normalization is performed, the appearance probability of the morpheme “pharmacy” is higher in the account item “consumables expense” than the account item “travel expense” by the morpheme “station”, and “pharmacy” is a web detail It is possible to correctly infer “consumables costs” at least when the number of times included in the data is “station”.

In one aspect, the web server 110 sums up the appearance probabilities of each morpheme as a score for each account item. As a result of the calculation, the account item with the highest score is assigned as the guess result. In the above example, “Japan” is the account of “Purchase”, and “Internet” is the account of “Insurance”, which is the highest score from the total score. This will be assigned.

In the present invention, automatic journals can be made for descriptions of unknown transaction details. Even when processing is performed for the first time using the appearance frequency generated in advance for each morpheme included in the description of the transaction content, it can be classified into the most probable account items. For example, when a new non-life insurance company called “XYZ Insurance” is established and this appears for the first time in the description of transaction details, the web server 110 has not processed it. However, the web server 110 divides into “XYZ” and “Non-life insurance”. For example, when “XYZ” has no registration of appearance frequency in the learned database and “Non-life” has the following registration, the estimation result As “insurance premium” can be assigned. The left side is the account item, and the right side is the normalized appearance frequency, that is, the appearance probability.

In the present invention, it is sufficient to extract information as much as necessary to infer the account item represented by the description of the transaction, rather than specifying the transaction partner from the description of the transaction and performing a journal on it. Is focused on. In response to the establishment of a new company or a change in the name of an existing company, there is a risk that periodic database maintenance will be required for automatic journal entries. However, in the present invention, morphemes that can explain account items are used. Is included in the description of transaction details, it is possible to classify it into a probable account item.

This is the same for the problem of notation that the description of the transaction contents fluctuates depending on the card company, settlement company, etc. For example, a certain company such as “Seven-Eleven”, “Seven-Eleven”, “Seven-Eleven”, etc., does not necessarily use a certain notation on the Web detailed data. In such a case, according to the present invention, the account item of “miscellaneous expenses” can be similarly estimated for any notation. For example, even if “Seven-Eleven” appears in the transaction description, it is segmented into “Seven” and “Eleven”. For each morpheme, the following occurrence frequency is obtained from the learned database. The Note that the appearance probability is less than 0.05.

Summing up the appearance probabilities of each morpheme as a score, the score of 0.63509 is the highest value for the account item “Miscellaneous expenses”, and this is assigned as a guess result. This result is the same whether the notation is “Seven-Eleven” or “Seven-Eleven”, and the name processing for correcting the notation and domain processing to deal with various notations were performed. A special morphological analysis dictionary (for example, other than an IPA dictionary) is not required. This indicates that it is extremely efficient to analyze the explanatory power of each morpheme rather than specifying the trading partner itself in order to infer the account item represented by the description of the transaction content.

In addition, “Seven-Eleven” hyphens, “Water charges for April”, etc. have a low appearance probability in multiple account items, and there is no spike in any account item. You may remove without. Thereby, calculation efficiency can be improved.

(Learn database details)
Taking the morpheme “Internet” as an example, generation of a learned database will be described.

When the user corrects the guess result using the menu 301 for changing the account item on the journal processing screen on the web browser, the correspondence between the description of the transaction content and the corrected account item is changed according to the user rule for each user. Is stored in the storage unit of the accounting processing apparatus 100.

In one aspect, the description of the transaction content included in each user rule is segmented into keywords, and for each keyword, the count of the number of occurrences of the corrected account item is incremented, whereby the frequency of appearance is changed to a plurality of users. Can be determined based on user rules.

The following table shows the correction results by multiple users.

In the above correction result, the description of each transaction content is morphologically analyzed and the appearance frequency of each morpheme is calculated as follows. The appearance probability is normalized assuming that only the account items listed in the table below are associated with each morpheme.

Therefore, the account items “communication fee”, “insurance fee”, and “newspaper book fee” are associated with the morpheme “Internet”, and the appearance frequencies thereof are 0.60, 0.35, and 0.05, respectively. .

The user rule used for generating the learned database may be a user rule that is currently used, may include a log, or may be a log only. By using collective intelligence, the burden of database construction / maintenance can be reduced. By configuring the learned database based only on the currently used user rules, it is possible to present inference results reflecting in real time what kind of matching the surrounding users are performing.

One advantage of reflecting in real time is that once a user submits a journal entry, the updated inference algorithm is applied to all users even for completely new transactions. Above, no maintenance is required, and as a user experience, automatic journaling accuracy can be improved.

Also, users can be segmented to improve the accuracy of the learned database. By segmenting according to the type of business, individual / corporation, etc., the same keyword can be appropriately sorted according to the segment. For example, when the description “Ox wholesaler” appears in the summary column 203, it may be a case where a corporation such as a retail business corresponds to “purchase” and a private business owner corresponds to “consumables”. In addition, segmentation according to the money amount column 202 can be considered. For example, “Yodobashi Camera” of less than 100,000 yen may be “consumables” and 100,000 yen or more may be “fixed assets”.

DESCRIPTION OF SYMBOLS 100 Accounting processing device 110 Web server 120 Database 130 Scraping server 140 Computer 151 Financial institution 152 Credit card company 200 Transaction processing screen 201 Date column 202 Amount column 203 Summary column 204 Account item column 301 Combo box 302 Help

Claims

An accounting processing apparatus for performing accounting processing by cloud computing,
A web server for providing cloud computing to users, the web server comprising:
Identify web statement data by transaction,
Each transaction is automatically journalized to a specific account by segmenting the transaction description of each transaction into keywords, referring to the frequency of occurrence of one or more accounts associated with each keyword,
Create journal data that includes at least the date, transaction details, amount and account,
The created journal data is transmitted to a computer where a user accesses the web server, and is displayed as a journal processing screen on a web browser of the computer.
The accounting processing apparatus according to claim 1, wherein the appearance frequency of one or more account items associated with each keyword is normalized for each keyword.
3. The accounting process according to claim 1 or 2, wherein the journal is automatically summed up as an account item having the highest score by summing up the appearance frequency of each keyword as a score for each account item. apparatus.
4. The accounting apparatus according to claim 1, wherein a keyword consisting of only one symbol or number is not considered in the total score.
5. The accounting processing apparatus according to claim 1, wherein the journal processing screen has a menu for changing an account item.
When the user selects and corrects an automatically journalized account item from the menu, the accounting processing device uses the correspondence between the description of the transaction contents and the corrected account item as a user rule for each user. 6. The accounting processing apparatus according to claim 5, wherein the accounting processing apparatus stores the information in a storage unit.
The accounting apparatus according to claim 6, wherein the appearance frequency is determined based on user rules of a plurality of users stored in the storage unit.
8. The appearance frequency is determined by dividing a description of transaction contents included in each user rule into keywords, and incrementing a count of the number of appearances of a modified account item for each keyword. The accounting processing device described.
9. The accounting processing apparatus according to claim 1, wherein the keyword is a morpheme.
The accounting processing apparatus according to claim 9, wherein segmentation into morphemes is performed without requiring domain processing.
An accounting processing method for performing accounting processing by cloud computing provided by a web server,
The web server identifying web detail data for each transaction;
The web server automatically divides each transaction into a specific account item by segmenting the description of the transaction contents of each transaction into keywords and referring to the frequency of appearance of one or more account items associated with each keyword. Journalizing steps,
The web server creating journal data including at least a date, a transaction content, an amount, and an account item,
The created journal data is transmitted to a computer where a user accesses the web server, and displayed as a journal processing screen on a web browser of the computer.
An accounting processing program for performing accounting processing by cloud computing provided by a web server, the web server,
Identifying web statement data for each transaction;
A step of automatically translating each transaction into a specific account item by segmenting the transaction description of each transaction into keywords and referring to the frequency of appearance of one or more account items associated with each keyword. When,
Creating journal data including at least a date, transaction content, amount and account,
An accounting processing program for executing a method, wherein the created journal data is transmitted to a computer where a user accesses the web server, and displayed on a web browser of the computer as a journal processing screen.