Document determination system, document determination device, and document determination method

Abstract

To provide a form determination system, a form determination apparatus, and a form determination method configured to properly identify the type of a form.SOLUTION: A form determination system SYS includes: a character recognition processing unit 102 which recognizes a character written on a form which is to be character recognition, and extracts character information and position information of the character written on the form; a character information analysis unit 103 which extracts a word from the character information extracted by the character recognition processing unit 102 and converts the extracted word into a vectorized word vector; and a form identifying unit 104 which identifies the type of the form to be subjected to character recognition, on the basis of form information in which information on word vectors included in each region of the form is defined for each type of the form and the word vector and the position information of the word extracted from the character information by the character information analysis unit.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a form determination system, a form determination device, and a form determination method.

Background Art

[0002] There is a technology for recognizing and extracting characters from a form by OCR (Optical Character Recognition). However, since the location of the characters to be extracted on the form varies depending on the type of form, it may not be possible to identify which characters are the extraction targets using only OCR technology. Also, when the characters to be extracted cannot be extracted, it may be impossible to distinguish whether they are not described on the form or whether they are described on the form but cannot be recognized by OCR. Therefore, it is necessary to identify the type of form to improve the accuracy of character recognition by OCR.

[0003] For example, Patent Document 1 discloses a technology for acquiring character information and position information of characters included in a form, checking which area of the form the acquired character information belongs to, and classifying the type of form based on the degree of match with the character information in the corresponding area of the pre-classified form.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, forms contain both items that are included as part of the format (such as the form name) and items that are written by the form creator (such as entry fields). A method of identifying the type of form based on the degree of matching of the character information contained in the form, as disclosed in Patent Document 1, is suitable for identifying the type of form in the case of slips and checklists, where a large proportion of the items are included as part of the format, because they contain many prescribed words and sentences. On the other hand, it is difficult to know in advance what words and sentences will be written in the items that are written by the form creator (such as entry fields), making it difficult to identify the type of form based on the degree of matching of the character information.

[0006] The present invention has been made in view of the above-mentioned problems, and one of its objectives is to provide a document determination system, a document determination device, and a document determination method that can appropriately identify the type of document. [Means for solving the problem]

[0007] To solve the above-mentioned problems, a document determination system according to one aspect of the present invention comprises: a character recognition processing unit that recognizes characters written in a document to be subject to character recognition and extracts character information and positional information of the characters written in the document; a character information analysis unit that extracts words from the character information extracted by the character recognition processing unit and converts the extracted words into vectorized word vectors; document information in which information on word vectors contained in each area of ​​the document is defined for each type of document; and a document identification unit that identifies the type of document to be subject to character recognition based on the word vectors and positional information of the words extracted from the character information by the character information analysis unit. The document information includes a definition of the probability of occurrence of the word vector for each region of the document, and the character information analysis unit calculates the probability of occurrence of the word vector for each region of the document based on the word vector and position information of the word extracted from the character information. The document identification unit identifies the type of document to be subject to character recognition based on the probability of occurrence of the word vector for each region calculated by the character information analysis unit and the document information. .

[0008] Furthermore, a document determination device according to one aspect of the present invention includes: a character recognition processing unit that recognizes characters written on a document to be subject to character recognition and extracts character information and positional information of the characters written on the document; a character information analysis unit that extracts words from the character information extracted by the character recognition processing unit and converts the extracted words into vectorized word vectors; document information in which information regarding word vectors contained in each area of ​​the document is defined for each type of document; and a document identification unit that identifies the type of document to be subject to character recognition based on the word vectors and positional information of the words extracted from the character information by the character information analysis unit. The document information includes a definition of the probability of occurrence of the word vector for each region of the document, and the character information analysis unit calculates the probability of occurrence of the word vector for each region of the document based on the word vector and position information of the word extracted from the character information. The document identification unit identifies the type of document to be subject to character recognition based on the probability of occurrence of the word vector for each region calculated by the character information analysis unit and the document information. .

[0009] Furthermore, a document determination method according to one aspect of the present invention includes the steps of: a character recognition processing unit recognizing characters written in a document to be recognized and extracting character information and positional information of the characters written in the document; and a character information analysis unit extracting words from the character information extracted by the character recognition processing unit and converting the extracted words into vectorized word vectors. The character information analysis unit calculates the probability of occurrence of the word vector for each area of ​​the form based on the word vector and position information of the word extracted from the character information. The form identification unit identifies the word vectors contained in each area of ​​the form. Probability of appearance This includes the document information defined for each type of document, The probability of occurrence of the word vector for each region calculated by the character information analysis unit. The method includes the step of identifying the type of document to be subject to character recognition based on the above. [Effects of the Invention]

[0010] According to the above embodiment of the present invention, the type of document can be appropriately identified. [Brief explanation of the drawing]

[0011] [Figure 1] A block diagram showing an example of the schematic configuration of the document determination system according to the embodiment. [Figure 2] A schematic diagram showing an example of the types of forms according to this embodiment. [Figure 3] A diagram showing an example of document information defined in the embodiment. [Figure 4] A block diagram showing an example of the hardware configuration of the document determination device according to the embodiment. [Figure 5] A flowchart showing an example of the document determination process according to the embodiment. [Modes for carrying out the invention]

[0012] Embodiments of the present invention will be described below with reference to the drawings. [System Configuration] First, the configuration of the document determination system according to the embodiment will be described. Figure 1 is a block diagram showing an example of a schematic configuration of the document determination system according to this embodiment. The document determination system SYS shown in the figure includes a document determination device 10. The document determination device 10 may consist of a single computer device, or it may be implemented in a configuration in which multiple computer devices work together.

[0013] The document determination device 10 receives documents that are subject to OCR character recognition. A document is a document in which various information is written in a prescribed format, such as a quotation, purchase order, delivery note, invoice, or application form. Specifically, it can be any document, not limited to a particular purpose, such as an insurance application form, a credit card application form, or an application form submitted to a government office. To obtain the character information of the contents written in this document, it is necessary to recognize and read the characters written on it using OCR. The document determination device 10 performs character recognition processing, which performs character recognition using OCR, and document determination processing, which identifies the type of document. By identifying the type of document, the document determination device 10 can improve the accuracy of OCR character recognition. The types of data in a document include, for example, image data, text data, and information that is a mixture of image data and text data.

[0014] For example, the form determination device 10 may be a personal computer, a scanner device, or a multifunction device having functions such as a copy function, a scanner function, a printer function, and a fax function. The form determination device 10 shown in FIG. 1 includes a form acquisition unit 101, a character recognition processing unit 102, a character information analysis unit 103, a form identification unit 104, an update unit 105, and a form information storage unit 110.

[0015] The form acquisition unit 101 acquires form data to be subjected to character recognition. For example, the form acquisition unit 101 may acquire scanned form data from a server 20 connected via a network N, or may acquire it via a storage medium such as an optical disk or a USB (Universal Serial Bus) memory. Also, when the form determination device 10 has a scanner function, the form acquisition unit 101 may acquire form data by scanning a paper form. The form data is, for example, data of an image file having an extension such as PNG, IMG, JPG, or data of a PDF file that does not contain text information.

[0016] The network N is a communication network including a public line such as the Internet or a mobile phone communication network, a wireless LAN (Local Area Network), or a wired LAN.

[0017] The character recognition processing unit 102 performs OCR processing on the form data acquired by the form acquisition unit 101 to recognize the characters described in the form, and extracts the character information of the characters described in the form and the position information indicating the positions of the characters. Note that the OCR processing may be AI-OCR processing applying AI technology. In AI-OCR processing, by machine learning (such as deep learning) of a large number of handwritten characters and typed characters, the character recognition accuracy can be improved.

[0018] The character information analysis unit 103 extracts words from the character information extracted by the character recognition processing unit 102. For example, the character information analysis unit 103 performs morphological analysis on the character information extracted by the character recognition processing unit 102 and then performs filtering to extract words. Here, the character information analysis unit 103 mainly extracts nouns as the words to be extracted from the character information. In general, although the usage frequency is low, by registering in advance the words to be extracted that are included in the form to be recognized as the target words for extraction, the character information analysis unit 103 can improve the extraction accuracy of the words.

[0019] In addition, the character information analysis unit 103 vectorizes the extracted words and converts them into word vectors. By converting into word vectors, words with similar meanings, properties, types, etc. can be recognized as the same type of words. For example, by vectorizing the words extracted from the form, it is possible to specify whether the extracted words are words indicating company names, personal names, place names, item names, or others.

[0020] As an example, even if the words are different in the comparison between words such as Saitama and Chiba, by using word vectors, it can be recognized that both Saitama and Chiba are words indicating place names. Similarly, for company names and personal names, although the names are different in the comparison between words and they are different words, by using word vectors, even if the names are different, the company names can be recognized as company names and the personal names can be recognized as personal names.

[0021] The form identification unit 104 identifies the type of form based on the word vectors of the words extracted from the form. Here, the method for identifying the type of form will be described.

[0022] FIG. 2 is a schematic diagram showing an example of the type of form according to the present embodiment. In this figure, simple examples of three types of forms, Document A, Document B, and Document C, are shown. Since the content and location described in the form differ depending on the type, the form is divided into a predetermined number of areas (for example, six areas from Area 1 to Area 6), and the type of form can be identified based on the differences in the word vectors of the words appearing in each area.

[0023] In the illustrated example, documents A, B, and C all include company information, details / breakdowns, and remarks, but the location of this information differs depending on the document. Document A is characterized by the appearance of the company name in both areas 1 and 6, as the requesting company name is listed in area 1 and the requested company information is listed in area 6. Document B is characterized by the appearance of the company name in both areas 1 and 2, as the sending company information is listed in area 1 and the receiving company information is listed in area 2. Document C is characterized by the appearance of the company name in both area 1 or 3 and area 2 or 4, as the requesting company information is listed in areas 1 through 3 and the requested company information is listed in areas 2 through 4.

[0024] By pre-preparing document information that defines information about word vectors contained in each area of ​​the document for each type of document, the document identification unit 104 compares the word vectors and positional information (word vectors appearing in each area) of words extracted from the document with the document information and identifies it as the closest type of document defined in the document information. For example, the document identification unit 104 identifies a document in which the company name appears in areas 1 and 6 as document A.

[0025] Furthermore, in some cases, it may not be possible to determine whether a document containing a company name in both Area 1 and Area 2 is Document B or Document C. Therefore, the probability of occurrence of word vectors appearing in each area is defined in the document information, and the character information analysis unit 103 calculates the probability of occurrence of word vectors in each area of ​​the document based on the word vectors and positional information of the words extracted from the document. The document identification unit 104 may then identify the document type that is closest to the original by comparing the probabilities of occurrence of both. Also, it may not be possible to identify the document type based on the company name alone. Therefore, the document identification unit 104 may identify the document type by comparing the probability of occurrence of multiple word vectors in each area, including word vectors of other words (e.g., place names, person names, item names, etc.) that are listed in either the company information or the details / breakdown.

[0026] In the example shown in Figure 2, the form is divided into a total of six areas: 2 (horizontal) x 3 (vertical). However, this is not the only way to divide the form; the number of divisions in each direction, the shape of the areas, etc., can be determined arbitrarily.

[0027] Figure 3 shows an example of document information defined in this embodiment. As shown in the figure, the document information defines the occurrence probability of word vectors for each area of ​​the document, for each type of document. The occurrence probability indicates the proportion of word vectors in a particular area. The specific method for calculating the occurrence probability is to use the total number of recognized words in a particular area (for example, in each of areas 1 to 6 shown in Figure 2) as the denominator and calculate the proportion by using the number of words aggregated for each word vector (company name, person's name, etc.) as the numerator. For example, for each type of document, the occurrence probability is defined by examining the word vectors appearing in each area of ​​multiple document samples (for example, 100 documents, etc.) and calculating it. This document information is stored in the document information storage unit 110. In other words, the document information is a list of the occurrence probabilities of word vectors for each area of ​​the document, for each type of document.

[0028] In this way, the document identification unit 104 identifies the type of document to be recognized based on document information, which defines information about word vectors contained in each area of ​​the document for each type of document, and word vectors and position information extracted from the document to be recognized. For example, the document information defines the probability of word vectors appearing in each region of the document for each type of document. The character information analysis unit 103 calculates the probability of word vectors appearing in each region of the document based on the word vectors and positional information extracted from the document to be recognized. The document identification unit 104 then identifies the type of document to be recognized based on the calculated probability of word vectors appearing in each region and the document information. The document identification unit 104 outputs information indicating the type of document that has been identified. The output format may be text data or display data.

[0029] The update unit 105 updates the document information based on the type of document to be recognized, as identified by the document identification unit 104, and the word vectors and position information extracted from the document to be recognized. For example, the update unit 105 updates the document information based on the type of document to be recognized, as identified by the document identification unit 104, and the occurrence probability of word vectors for each region, calculated by the character information analysis unit 103.

[0030] Specifically, the user confirms the type of document targeted for character recognition, as identified by the document identification unit 104. If the identification is correct, the update unit 105 adds the probability values ​​of word vector occurrences for each region of that document to the document information as one of the identified document samples. If, as a result of the user's confirmation, the type of document targeted for character recognition identified by the document identification unit 104 is incorrect, the update unit 105 adds the probability values ​​of word vector occurrences for each region of that document to the document information as one of the correct document samples. In this way, by feeding back the results of document type identification to the document information, the accuracy of document type identification can be improved.

[0031] Figure 4 is a block diagram showing an example of the hardware configuration of the document determination device according to this embodiment. The document determination device 10 includes a CPU (Central Processing Unit) 11, RAM (Random Access Memory) 12, ROM (Read Only Memory) 13, storage device 14, communication unit 15, input unit 16, and output unit 17 as its hardware configuration.

[0032] The CPU 11 is a processor that performs various processes by executing programs stored in the ROM 13 or storage device 14. For example, the CPU 11 performs the processes of the form acquisition unit 101, character recognition processing unit 102, character information analysis unit 103, form identification unit 104, and update unit 105 shown in Figure 1.

[0033] RAM12 is used as a reading area for programs executed by the CPU11, or as a work area for writing data used for processing by said programs.

[0034] ROM13 consists of electrically rewritable non-volatile memory, such as EEPROM (Electrically Erasable Programmable Read Only Memory) or flash ROM. For example, ROM13 stores at least a portion of the system program and programs that execute various processes.

[0035] The storage device 14 is comprised of an HDD (Hard Disk Drive), an SSD (Solid State Drive), and the like. For example, the storage device 14 may store at least a portion of a system program, a program that executes various processes, and so on. The storage device 14 also includes a report information storage unit 110 as shown in Figure 1, which stores, for example, the report information shown in Figure 3.

[0036] The communication unit 15 connects to a network via a wireless LAN (Local Area Network) or wired LAN to perform data communication with external devices (such as a server 20). The communication unit 15 may also perform data communication with peripheral devices by being equipped with short-range wireless communication such as Bluetooth® or an interface such as USB (Universal Serial Bus).

[0037] The input unit 16 includes, for example, an input device such as a keyboard, touchpad, touch panel, or microphone. Furthermore, if the document determination device 10 has a scanner function, the input unit 16 may be configured as a scanner. The output unit 17 includes a display device such as a liquid crystal display or an organic EL display, and an output device such as a speaker.

[0038] As mentioned above, the document determination device 10 may consist of a single computer device, or it may be a configuration in which multiple computer devices work together. Therefore, it may have one or more hardware configurations as shown in Figure 4. For example, in a configuration where the document determination device 10 is connected in a way that allows communication between a computer device including a document acquisition unit 101 and a character recognition processing unit 102, and a computer device including a character information analysis unit 103, a document identification unit 104, and an update unit 105, each computer device may have the hardware configuration shown in Figure 4.

[0039] Next, we will explain the operation of the document determination process performed by the document determination device 10 described above. Figure 5 is a flowchart showing an example of the document determination process according to this embodiment.

[0040] First, the document acquisition unit 101 determines whether or not it has acquired document data to be used for character recognition (step S101). If the document acquisition unit 101 determines that it has not acquired document data (step S101: NO), it repeats the process in step S101 and waits for input of document data. On the other hand, if the document acquisition unit 101 determines that it has acquired document data (step S101: YES), it saves the acquired document data to the storage device 14 and proceeds to the process in step S103.

[0041] In step S103, the character recognition processing unit 102 reads the document data acquired by the document acquisition unit 101 from the storage device 14, performs OCR processing on the document data to recognize the characters written on the document, and extracts the character information of the characters written on the document and position information indicating the position of those characters.

[0042] Next, the character information analysis unit 103 extracts words from the character information extracted by the character recognition processing unit 102 (step S105). Subsequently, the character information analysis unit 103 vectorizes the extracted words and converts them into word vectors (step S107). Then, the character information analysis unit 103 calculates the probability of occurrence of word vectors for each area of ​​the form based on the word vectors and positional information of the words extracted from the form (step S109).

[0043] Next, the document identification unit 104 compares the occurrence probability of word vectors for each region calculated in step S109 with document information defined for each type of document (step S111). Then, based on the comparison result, the document identification unit 104 identifies the type of document to be subject to character recognition and outputs information indicating the identified type of document (step S113). The output format may be text data or display data.

[0044] Through the above processing, the document determination device 10 can identify the type of document. Subsequent processing is feedback processing that can be performed to further improve the accuracy of document type identification. This feedback processing may be performed each time a document type is identified, or it may be performed all at once later. Furthermore, the feedback processing may be performed on all identified documents, or on some of them.

[0045] The user confirms the information indicating the type of document identified by the document identification unit 104, and inputs the result of confirming whether the identified type is correct using the input unit 16 (e.g., keyboard, mouse, etc.) (step S115).

[0046] The update unit 105 updates the occurrence probabilities of the report information based on the user's confirmation results (step S117). For example, if the type of report is correctly identified based on the user's confirmation results, the update unit 105 adds the occurrence probability values ​​of the word vectors for each region of that report to the report information as one of the identified report samples. On the other hand, if the type of report is incorrectly identified based on the user's confirmation results, the update unit 105 adds the occurrence probability values ​​of the word vectors for each region of that report to the report information as one of the correct report samples.

[0047] As described above, the document determination system SYS according to this embodiment recognizes the characters written in the document to be recognized, extracts character information and position information of the characters written in the document, further extracts words from the extracted character information, and converts the extracted words into vectorized word vectors. The document determination device 10 then identifies the type of document to be recognized based on document information, which defines information about word vectors contained in each area of ​​the document for each type of document, and the word vectors and position information of the words extracted from the character information.

[0048] As a result, the document determination system SYS can appropriately identify the type of document by recognizing the characters written in the document and converting them into word vectors, since the content written in each area differs depending on the type of document. The document determination device 10, rather than judging the characters written in the document as words, judges them as word vectors, so even if the words are different when compared to each other, it can recognize company names as company names and personal names as personal names, and can appropriately identify the type of document.

[0049] For example, the document information defines the probability of occurrence of the word vector for each region of the document, for each type of document. The document determination device 10 then calculates the probability of occurrence of the word vector for each region of the document based on the word vector and position information of the words extracted from the character information, and identifies the type of document to be subject to character recognition based on the calculated probability of occurrence of the word vector for each region and the document information.

[0050] As a result, the document determination system SYS can appropriately identify the type of document by recognizing the characters appearing in each area of ​​the document and converting them into word vectors, since the probability of word vectors appearing in each area differs depending on the type of document.

[0051] Furthermore, the document determination device 10 updates the document information based on the type of document identified as the target of character recognition, and the word vectors and positional information of the words extracted from the character information.

[0052] This allows the SYS document type determination system to improve its accuracy in identifying document types by feeding back the results of the document type identification process.

[0053] For example, the document recognition system SYS updates the document information based on the type of document targeted for character recognition and the calculated probability of word vector occurrence for each region.

[0054] This allows the SYS document type determination system to improve its accuracy in identifying document types by feeding back the results of the document type identification process.

[0055] Furthermore, the document determination device 10 according to this embodiment recognizes the characters written in the document to be recognized, extracts character information and position information of the characters written in the document, further extracts words from the extracted character information, and converts the extracted words into vectorized word vectors.Then, the document determination device 10 identifies the type of document to be recognized based on document information, which defines information about word vectors contained in each area of ​​the document for each type of document, and the word vectors and position information of the words extracted from the character information.

[0056] As a result, the document determination device 10 can appropriately identify the type of document by recognizing the characters written on the document and converting them into word vectors, since the content written in each area differs depending on the type of document. Because the document determination device 10 judges the characters written on the document using word vectors rather than words, it can recognize company names as company names and personal names as personal names even if the words are different when compared to each other, and can appropriately identify the type of document.

[0057] Furthermore, the document determination method in the document determination system SYS and document determination device 10 according to this embodiment includes the steps of: a character recognition processing unit 102 recognizing characters written in the document to be determined and extracting character information and position information of the characters written in the document; a character information analysis unit 103 extracting words from the character information extracted by the character recognition processing unit 102 and converting the extracted words into vectorized word vectors; and a document identification unit 104 identifying the type of document to be determined based on document information in which information regarding word vectors contained in each area of ​​the document is defined for each type of document, and the word vectors and position information of the words extracted from the character information by the character information analysis unit 103.

[0058] As a result, the document determination system SYS and the document determination device 10 can appropriately identify the type of document by recognizing the characters written in the document and converting them into word vectors, since the content written in each area differs depending on the type of document. Because the document determination device 10 determines the characters written in the document using word vectors rather than words, it can recognize company names as company names and personal names as personal names even if the words are different when compared to each other, and can appropriately identify the type of document.

[0059] Although each embodiment of this invention has been described in detail above with reference to the drawings, the specific configuration is not limited to the embodiments described above, and includes designs and the like that do not depart from the gist of this invention. For example, the configurations described in each of the embodiments described above can be combined in any way.

[0060] The document determination device 10 described above has an internal computer system. Furthermore, programs for realizing the functions of each configuration of the document determination device 10 may be recorded on a computer-readable recording medium, and the programs recorded on this recording medium may be loaded into the computer system and executed to perform the processing in each configuration of the document determination device 10. Here, "loading the programs recorded on the recording medium into the computer system and executing them" includes installing the programs into the computer system. Here, "computer system" includes hardware such as the OS and peripheral devices. Also, "computer system" may include multiple computer devices connected via a network including communication lines such as the Internet, WAN, LAN, and dedicated lines. Furthermore, "computer-readable recording medium" refers to portable media such as flexible disks, magneto-optical disks, ROMs, CD-ROMs, and storage devices such as hard disks built into the computer system. Thus, the recording medium storing the program may be a non-transient recording medium such as a CD-ROM.

[0061] Furthermore, the recording medium also includes internal or external recording media accessible from the distribution server for distributing the program. The program may be divided into multiple parts, downloaded at different times, and then combined in each configuration of the report determination device 10. The distribution servers distributing each of the divided programs may also be different. Moreover, "computer-readable recording media" includes volatile memory (RAM) within computer systems that act as servers or clients when a program is transmitted over a network, which retains the program for a certain period of time. The program itself may also be intended to implement some of the functions described above. Furthermore, the program may be a so-called differential file (differential program) that can implement the functions described above in combination with a program already recorded in the computer system.

[0062] Furthermore, some or all of the functions of the document determination device 10 in the above-described embodiment may be implemented as an integrated circuit such as an LSI (Large Scale Integration). Each function may be individually processorized, or some or all of them may be integrated into a single processor. In addition, the method of implementing the integrated circuit is not limited to LSIs; it may also be implemented using dedicated circuits or general-purpose processors. Furthermore, if an integrated circuit technology that can replace LSIs emerges due to advances in semiconductor technology, an integrated circuit using that technology may be used. [Explanation of symbols]

[0063] SYS Document Judgment System, 10 Document Judgment Device, 11 CPU, 12 RAM, 13 ROM, 14 Storage Device, 15 Communication Unit, 16 Input Unit, 17 Output Unit, 20 Server, 101 Document Acquisition Unit, 102 Character Recognition Processing Unit, 103 Character Information Analysis Unit, 104 Document Identification Unit, 105 Update Unit, 110 Document Information Storage Unit

Claims

1. A character recognition processing unit recognizes characters written on a document to be recognized and extracts character information and positional information of the characters written on the document. A character information analysis unit extracts words from the character information extracted by the character recognition processing unit and converts the extracted words into vectorized word vectors. A document identification unit identifies the type of document to be subject to character recognition based on document information, which defines information about word vectors contained in each area of ​​the document for each type of document, and the word vectors and position information of words extracted from the character information by the character information analysis unit. Equipped with, The aforementioned report information includes a definition for each type of report, specifying the probability of occurrence of the word vector for each area of ​​the report. The character information analysis unit calculates the probability of occurrence of the word vector for each area of ​​the form based on the word vector and position information of the word extracted from the character information. The document identification unit identifies the type of document to be subject to character recognition based on the occurrence probability of the word vector for each region calculated by the character information analysis unit and the document information. Document review system.

2. An update unit updates the document information based on the type of document to be recognized, identified by the document identification unit, and the word vectors and position information of the words extracted from the character information by the character information analysis unit. The document determination system according to claim 1, further comprising:

3. An update unit updates the document information based on the type of document targeted for character recognition identified by the document identification unit and the probability of occurrence of the word vector for each region calculated by the character information analysis unit. The document determination system according to claim 1, further comprising:

4. A character recognition processing unit recognizes characters written on a document to be recognized and extracts character information and positional information of the characters written on the document. A character information analysis unit extracts words from the character information extracted by the character recognition processing unit and converts the extracted words into vectorized word vectors. A document identification unit identifies the type of document to be subject to character recognition based on document information, which defines information about word vectors contained in each area of ​​the document for each type of document, and the word vectors and position information of words extracted from the character information by the character information analysis unit. Equipped with, The aforementioned report information includes a definition for each type of report, specifying the probability of occurrence of the word vector for each area of ​​the report. The character information analysis unit calculates the probability of occurrence of the word vector for each area of ​​the form based on the word vector and position information of the word extracted from the character information. The document identification unit identifies the type of document to be subject to character recognition based on the occurrence probability of the word vector for each region calculated by the character information analysis unit and the document information. Document judgment device.

5. The character recognition processing unit performs the steps of recognizing the characters written in the document to be recognized and extracting the character information and position information of the characters written in the document, The character information analysis unit extracts words from the character information extracted by the character recognition processing unit, and converts the extracted words into vectorized word vectors. The character information analysis unit calculates the probability of occurrence of the word vector for each area of ​​the form based on the word vector and position information of the word extracted from the character information. The document identification unit identifies the type of document to be subject to character recognition based on document information in which the occurrence probability of word vectors contained in each area of ​​the document is defined for each type of document, and the occurrence probability of the word vectors for each area calculated by the character information analysis unit. A method for determining the type of document, including the type of document.

Images