JP5936051B2 - Document processing apparatus and program - Google Patents

Description

  The present invention relates to a document processing apparatus and a program.

  Patent Document 1 discloses that a digital image of a form read from a scanner is tilted in order to reduce the amount of work for an operator in a form dictionary registration operation and to provide a system that realizes highly accurate identification in a form identification operation. After correcting and removing the margins around the form, the image input unit stores it in the image memory, and detects the size of the form from the digital image in the image memory and also detects the character pattern in character string units. When registering the feature extraction unit that converts to feature values and stores them in the feature memory, and when registering the format of the form to be input, the feature values common to each set are extracted from the feature values of multiple sets of character strings in the feature memory. If the common feature extraction unit to be registered in the dictionary memory and the format of the input form are identified, the feature amount and dictionary of the character string of the processing form in the feature memory It discloses a structure and a collation unit that collates the feature amount of a character string for each document in the memory.

  Patent Document 2 discloses an image processing apparatus that determines the similarity between input image data and a registered image, in order to reduce the burden on the user regarding the setting operation of processing contents for input image data. Disclosure.

  A common processing rule that is commonly applied to input image data determined to be similar to the registered image identification information and the registered image is stored in the server device. An integrated processing rule obtained by integrating the individual processing rule set for each user with the processing content to be applied to the input image data determined to be similar to the registered image and the common processing rule is stored in the card. If the version of the integrated processing information matches the version of the common processing information, the processing based on the integrated processing information corresponding to the registered image determined to be similar to the input image data is performed, and does not match In this case, the processing based on the integrated processing information is not performed on the input image data.

JP 2002-288675 A JP 2009-88712 A

  An object of the present invention is to provide a document processing apparatus and program capable of performing a process of classifying a document according to a classification rule at high speed.

[Document Processing Device]
According to a first aspect of the present invention, there is provided a classification rule registration means for registering a combination of a first rule and a second rule applied to the first rule as a classification rule for classifying documents, and the classification rule registration. The document processing apparatus includes a classification rule integration unit that integrates the plurality of different classification rules registered by the unit by excluding contents overlapping between the plurality of different classification rules.

  The present invention according to claim 2 is characterized in that the classification rule registering means has a first rule defining what is used for document classification among document features, and a document feature defined by the first rule. The combination with the second rule that defines the logic to be applied is registered as a classification rule, and the classification rule integration means integrates the classification rule by eliminating duplication of the document features defined in the first rule. The document processing apparatus according to claim 1.

  The present invention according to claim 3 is characterized in that the classification rule registration means defines a rule defining a character, a photograph, a figure, a table or an encoded product included in a document as a feature of the document used for classification of the document as the first rule. The document processing apparatus according to claim 2.

  The present invention according to claim 4 is characterized in that the classification rule registering means defines a combination of two or more of characters, photos, figures, tables, and encoded products included in the document as document features used for document classification. The document processing apparatus according to claim 3, wherein the first rule is the first rule, and the second rule is a rule that defines a logic to be applied to any two or more combinations.

  The present invention according to claim 5 is the document processing according to any one of claims 1 to 4, wherein the classification rule registration means registers an individual classification rule and a common classification rule among a plurality of operators for each operator. Device.

  The present invention according to claim 6 further includes document accepting means for accepting documents, and classification means for classifying the documents accepted by the document accepting means based on the classification rules integrated by the classification rule registering means. The document processing apparatus according to claim 1.

[program]
The present invention according to claim 7 includes a step of registering a combination of a first rule and a second rule applied to the first rule as a classification rule for classifying a document, and a plurality of registered different classifications. This is a program for causing a computer to execute a step of eliminating and integrating contents overlapping between the plurality of different classification rules.

  According to the first aspect of the present invention, it is possible to provide a document processing apparatus capable of performing a process of classifying a document according to a classification rule at a high speed as compared with the case where the present configuration is not provided.

  According to the second aspect of the present invention, in addition to the effect of the first aspect of the invention, it is possible to provide a document processing apparatus that can classify based on a desired feature among document features.

  According to the present invention of claim 3, in addition to the effect of the invention of claim 2, it is possible to provide a document processing apparatus capable of classifying documents based on characters, photographs, figures, or encoded products. .

  According to the present invention according to claim 4, in addition to the effect of the present invention according to claim 3, it is possible to classify documents based on a combination of characters, photos, figures, tables or encoded products according to a desired logic. It is possible to provide a document processing apparatus that can be used.

  According to the present invention of claim 5, in addition to the effects of the present invention of claims 1 to 4, a classification based on a classification rule common to a plurality of operators, a classification based on a desired classification rule for each operator, It is possible to provide a document processing apparatus that can

  According to the sixth aspect of the present invention, in addition to the effects of the present invention according to the first to fifth aspects, a document processing apparatus capable of classifying received documents at a higher speed than the case without the present configuration. Can be provided.

  According to the seventh aspect of the present invention, it is possible to provide a program capable of performing processing for classifying documents according to the classification rules at a higher speed than in the case where the present configuration is not provided.

1 is a cross-sectional view illustrating an image forming apparatus 10 that is an example of a document processing apparatus according to an embodiment of the present invention. 2 is a block diagram illustrating a hardware configuration of the image forming apparatus 10. FIG. 3 is a block diagram showing an information processing program 70 that operates in the image forming apparatus 10. FIG. 3 is a block diagram illustrating a configuration of a feature extraction unit 76. FIG. 6 is a schematic diagram illustrating an example of a classification rule generated by a classification rule generation unit 82. FIG. 7 is a table showing an example of a table stored in a rule template storage unit 84. FIG. 10 is a plan view illustrating a first example of a screen related to classification rule registration displayed on the UI device 50. 10 is a plan view showing a second example of a screen related to registration of classification rules displayed on the UI device 50. FIG. 10 is a plan view showing a third example of a screen related to registration of a classification rule displayed on the UI device 50. FIG. 10 is a plan view illustrating a fourth example of a screen related to registration of classification rules displayed on the UI device 50. FIG. FIG. 10 is a schematic diagram of classification rules registered in a classification rule storage unit 92 before integration by a classification rule integration unit 96; FIG. 10 is a schematic diagram of classification rules registered in a classification rule storage unit 92 after integration by a classification rule integration unit 96; It is an example of the flowchart which shows the operation | movement about registration of a classification rule. It is an example of the flowchart which shows the operation | movement which classify | categorizes a document and performs the process according to a classification | category. It is a table | surface which shows an example of the processing rule for every classification | category applied when processing according to a classification | category.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
FIG. 1 is a cross-sectional view showing an image forming apparatus 10 which is an example of a document processing apparatus according to an embodiment of the present invention.

  The image forming apparatus 10 includes a printing device 12 and a reading device 14. The printing device 12 includes, for example, a three-stage recording medium supply cassette 16, and each of these recording medium supply cassettes 16 is provided with a supply head 18. It has been.

  When one of the recording medium supply cassettes 16 is selected, the supply head 18 is operated to supply the selected recording medium supply cassette 16 to the image forming unit 22 through the recording medium supply path 20.

  Further, the image forming apparatus 10 may have a facsimile function (not shown).

  The image forming unit 22 is provided with yellow, magenta, cyan, and black photoconductors 24 and an intermediate transfer belt 26.

  A charging device, an exposure device, a developing device, a primary transfer device, a cleaning device, and the like (not shown) are arranged around each photoconductor 24, and a toner image formed on each photoconductor 24 is transferred to the intermediate transfer belt 26. Transcribed. When black and white is set, only black is enabled.

  The toner image on the intermediate transfer belt 26 is transferred to the recording medium sent by the secondary transfer roll 28 and fixed by the fixing device 30. The recording medium on which the toner image is fixed passes through the recording medium discharge path 32. And discharged to the discharge unit 34.

  However, when duplex printing is set, the recording medium whose surface has been fixed by the fixing device 30 is sent from the recording medium discharge path 32 to the reversing device 36 and is reversed by the reversing device 36, and the recording medium reversing path 38. Is sent back to the recording medium supply path 20 and sent to the image forming unit 22 to print the back surface.

  The reading device 14 includes an automatic document feeder 40 capable of reading a double-sided document. The document is fed to the platen 42 by the automatic document feeder 40, and the document is scanned on the platen 42 by a reading unit 44 composed of a CCD or the like. Is read. The reading device 14 may read a document placed on the platen 42 by the operator using the reading unit 44.

  A document set detector 46 for detecting whether or not a document is set on the automatic document feeder 40 is provided. The automatic document feeder 40 also serves as a platen cover, and the document can be placed on the platen 42 by opening the platen cover. The opening / closing of the platen cover can be detected by a platen cover opening / closing detector 48.

  The reading device 14 outputs the reading conditions to the classification rule generation unit 82 described later. Here, the reading condition is, for example, a setting for reading in color, a setting for reading in black and white (two gradations), a setting for reading only a character document, or a document including characters and figures. The setting in the reading process is included such as whether to set for reading. In addition, a character means a character or a character string. The reading conditions also include information related to the read document, such as information that a plurality of continuous documents have been read by the automatic document feeder 40, for example.

  The user interface device 50 (hereinafter “UI device 50”) is provided integrally with the image forming apparatus 10 or via a network. Further, the UI device 50 includes, for example, a touch panel, and is configured to accept an instruction about processing content from the operator to the image forming apparatus 10 as an input or display information.

  The image forming apparatus 10 is provided with a facsimile modem connected to a public line and a network communication device connected to a network such as a LAN. By using a network communication device provided in the image forming apparatus 10, an image read by the reading device 14 can be transmitted to a terminal connected to the network. In addition, the image forming apparatus 10 receives data from a terminal connected to the network via the network communication apparatus.

  FIG. 2 is a block diagram illustrating a hardware configuration of the image forming apparatus 10.

  As shown in FIG. 2, the image forming apparatus 10, along with the UI apparatus 50 of FIG. 1, has a CPU 52, a memory 54, an input / output interface 56, a scan interface 58, a print interface 60, a network communication interface 62, and a facsimile modem interface 64. The storage device 66 is connected by a bus.

  That is, the image forming apparatus 10 includes a component as a computer capable of information processing and communication with other apparatuses.

  The CPU 52 controls each circuit by executing an information processing program 70 described later written in the memory 54 or the storage device 66. Input received via the UI device 50 is transmitted to the CPU 52, and display information from the CPU 52 is transmitted to the UI device 50.

  The CPU 52 may execute the information processing program 70 stored in a storage medium such as a CD ROM, or may execute the information processing program 70 provided via the network communication interface 62.

  A document set detection signal from the document set detector 46 in FIG. 1 and a platen cover open / close signal from the platen cover open / close detector 48 in FIG. 1 are input to the input / output interface 56.

  The scan interface 58 is connected to the reading device 14 of FIG. 1, and the print interface 60 is connected to the printing device 12 of FIG.

  The network communication interface 62 is connected to a network communication device provided in the image forming apparatus 10, and the modem interface 64 is connected to a facsimile modem provided in the image forming apparatus 10.

  Further, a storage device 66 is connected to the bus, and for example, an image transmitted from a terminal connected to the network can be stored in the storage device 66 provided in the image forming apparatus 10.

  FIG. 3 is a block diagram showing an information processing program 70 that operates in the image forming apparatus 10. As shown in FIG. 3, the information processing program 70 includes a document reception unit 72, an object separation unit 74, a feature extraction unit 76, a document type identification unit 78, a document type storage unit 80, a classification rule generation unit 82, and a rule template storage unit. 84, a character information reception unit 86, a classification rule presentation unit 88, a classification rule registration unit 90, a classification rule storage unit 92, an editing reception unit 94, a classification rule integration unit 96, and a classification unit 98.

  The document receiving unit 72 receives a document input to the image forming apparatus 10. In the present embodiment, the document receiving unit 72 receives a document read by the reading device 14 or a document sent from a terminal connected to the network. Note that a document refers to data in which information is described, and includes not only data including characters but also drawings or photographs that do not include characters. In the description of the present embodiment, “character” means a character or a character string. The document reception unit 72 outputs the received document to the object separation unit 74.

  The object separation unit 74 separates the document for each object in the document and outputs the document to the feature extraction unit 76. Here, an object refers to an object that constitutes the content of a document, and is broadly divided into a character and an image. The image includes a table, a photograph, an encoded object, a figure, and the like. Note that the encoded product refers to an image of encoded information, and corresponds to, for example, a barcode or a QR code (registered trademark). The feature extraction unit 76 of this embodiment separates a document into a character region and an image region (specifically, a table region, a photographic region, and an encoded product region), and the data of each separated region is sent to the feature extraction unit 76. Output.

  The feature extraction unit 76 extracts features included in the document. The feature extraction unit 76 of the present embodiment extracts features for each area of data in the document separated by the object separation unit 74. The feature extraction unit 76 outputs the extracted features to the document type identification unit 78 and the classification rule generation unit 82 when generating the document classification rules, and extracts the features when classifying the documents. Is output to the classification unit 98. The detailed configuration of the feature extraction unit 76 will be described later.

  The document type identification unit 78 identifies which of the predetermined document types the document corresponds to based on the features extracted from the document. In the present embodiment, the document type identification unit 78 compares the features for each document type stored in the document type storage unit 80 with the features extracted by the feature extraction unit 76, and identifies the document type. The document type identification unit 78 notifies the classification rule generation unit 82 of the identified document type.

  The document type storage unit 80 stores a document type and a feature common to the document type in association with each other for each predetermined document type. The document type storage unit 80 in the present embodiment stores the document type and features common to the document type as a table.

  The document type storage unit 80 includes, for example, “application” document, “drawing” document, “photo” document, “barcode-added document” document, “description” document, “approval document” document, “ For each predetermined document type such as an “educational document” document, the characteristics of the document type are stored in association with each other.

  The features stored in the document type storage unit 80 are unique to each document type among the features extracted by the feature extraction unit 76. For example, for the document type “application”, the subject (title) in the document It memorize | stores as a characteristic that there exists a character of "application" in a position. Further, for example, the document type “photo” is stored as a feature that there is a photo in the document, and the document type “document with barcode” is stored as a feature that there is a barcode in the document.

  The document type storage unit 80 may store a combination of a plurality of features instead of one feature for each document type. For example, the document type “application form” may be stored as a feature that there is a character “application form” at the title position in the document and a table image is included in the document.

  The classification rule generation unit 82 generates a combination of a first rule and a second rule applied to the first rule as a classification rule for classifying documents. The classification rule generation unit 82 according to the present embodiment uses, as a first rule, a rule that defines a feature used for classification among features extracted from a document, and applies the rule to the document feature defined by the first rule. A rule that defines logic is set as a second rule, and a classification rule including a combination of the first rule and the second rule is generated.

  Specifically, the classification rule generation unit 82 extracts a template and a feature corresponding to the document type specified by the document type specifying unit 78 among the rule templates for each document type stored in the rule template storage unit 84. A classification rule is generated using the features extracted by the unit 76.

  Further, the classification rule generation unit 82 may generate a classification rule based on the character information received by the character information reception unit 86.

  The character information receiving unit 86 receives characters input by the operator via the UI device 50 and outputs them to the classification rule generating unit 82. For example, when the operator wants to include a specific character in the classification rule or to include a character related to the specific character in the classification rule, the operator inputs the specific character.

  The classification rule generation unit 82 generates a classification rule using characters selected based on the characters received by the character information reception unit 86 among the characters extracted by the feature extraction unit 76. For example, the classification rule generation unit 82 generates a classification rule using a character received by the character information reception unit 86 or a character related to the character received by the character information reception unit 86.

  Further, the classification rule generation unit 82 may generate a classification rule based on a reading condition from the reading device 14.

  The classification rule generation unit 82 outputs the generated classification rule to the classification rule presentation unit 88 and the classification rule registration unit 90. Details of the classification rule generated by the classification rule generation unit 82 will be described later.

  The classification rule generation unit 82 may generate a classification rule for classifying the document based on the features extracted from the document, and the feature extraction unit 76 does not specify the document type by the document type specification unit 78. Classification rules may be generated based on the extracted features.

  The rule template storage unit 84 stores a document type and a rule template used for classification of the document type in association with each other for each predetermined document type. The rule template storage unit 84 in the present embodiment stores document types and rule templates corresponding to the document types as a table.

  The classification rule presentation unit 88 presents the classification rule generated by the classification rule generation unit 82 to the operator. The classification rule presentation unit 88 of the present embodiment displays the classification rule generated by the classification rule generation unit 82 on the UI device 50 and presents it to the operator.

  The classification rule registration unit 90 registers in the classification rule storage unit 92 a classification rule that is a combination of the first rule and the second rule applied to the first rule. In this embodiment, the classification rule registration unit 90 registers the classification rule generated by the classification rule generation unit 82 or the classification rule edited by the editing reception unit 94 in the classification rule storage unit 92.

  The classification rule storage unit 92 is a database that stores classification rules used for document classification.

  The edit accepting unit 94 receives an editing operation on the classification rule generated by the classification rule generating unit 82 or the classification rule stored in the classification rule storage unit 92, and edits corresponding to the editing operation are performed on the classification rule and edited. The obtained classification rule is output to the classification rule registration unit 90. The edit receiving unit 94 of the present embodiment receives an editing operation for the classification rule via the UI device 50, and performs editing corresponding to the received editing operation on the classification rule.

  The classification rule integration unit 96 integrates a plurality of different classification rules registered in the classification rule storage unit 92 by the classification rule registration unit 90 by eliminating content overlapping between the plurality of different classification rules. The classification rule integration unit 96 according to the present embodiment reads out the classification rules stored in the classification rule storage unit 92, integrates the read out classification rules by excluding duplicate contents regarding the first rule, and integrates them. The classified classification rule is registered in the classification rule storage unit 92. Details of the classification rule integration unit 96 will be described later.

  The classification unit 98 applies the classification rule integrated by the classification rule integration unit 96 stored in the classification rule storage unit 92 to the features extracted by the feature extraction unit 76, and classifies the document.

Next, a detailed configuration of the feature extraction unit 76 will be described.
FIG. 4 is a block diagram illustrating a configuration of the feature extraction unit 76. As shown in FIG. 4, the feature extraction unit 76 includes a first feature extraction unit 100 and a second feature extraction unit 102.

  The first feature extraction unit 100 extracts features about character regions in the document. The first feature extraction unit 100 of this embodiment extracts characters from the data of the character area separated by the object separation unit 74. The first feature extraction unit 100 includes a character recognition unit 104 and a character feature extraction unit 106.

  The character recognition unit 104 performs character recognition on the character region in the document, recognizes the character included in the character region, and outputs it to the character feature extraction unit 106.

  The character feature extraction unit 106 performs, for example, morphological analysis on the characters recognized by the character recognition unit 104, and extracts characters such as nouns and positions of the characters in the document as features. In addition, when there are a plurality of extracted characters, the character feature extraction unit 106 may set a priority order corresponding to the appearance frequency in the document for the extracted characters. In the following description, the characters extracted by the character feature extraction unit 106 may be referred to as keywords.

  The second feature extraction unit 102 extracts features about an image region in the document as information about the image. The second feature extraction unit 102 of the present embodiment extracts features from the data of the table area, the photograph area, and the encoded product area separated by the object separation unit 74. The second feature extraction unit 102 includes a table feature extraction unit 108, a photographic feature extraction unit 110, and an encoded feature extraction unit 112.

  A table feature extraction unit 108 extracts features of a table image in a document. The table feature extraction unit 108 of the present embodiment extracts, as features, the size of the table image in the document, the position of the table image in the document, the total number of table images in the document, and the like.

  The photograph feature extraction unit 110 extracts the feature of the photograph image in the document. The photo feature extraction unit 110 of the present embodiment extracts features such as the size of a photo image in the document, the position of the photo image in the document, the total number of photo images in the document, and the like.

  The encoded product feature extraction unit 112 extracts features of the encoded product image in the document. The encoded product feature extraction unit 112 of the present embodiment extracts the size of the encoded product image in the document, the position of the encoded product image in the document, the total number of encoded images in the document, and the like as features. The encoded product feature extraction unit 112 may decode the encoded product image and extract the encoded information as a feature.

  Note that the second feature extraction unit 102 may extract information of a figure included in the document as the feature of the image.

  Next, the classification rule generated by the classification rule generation unit 82 in this embodiment will be described. The classification rule generation unit 82 defines, as a first rule, a rule that defines a feature used for classification among features extracted from a document, and defines logic to be applied to the document feature defined by the first rule. The rule is set as the second rule, and a classification rule including a combination of the first rule and the second rule is generated.

  The first rule in the present embodiment is a rule that defines which of the features extracted by the feature extraction unit 76 is to be used for classification determination.

  Here, the feature defined in the first rule may be one or plural. In addition, the first rule may be defined by combining the feature extracted by the first feature extraction unit 100 and the feature extracted by the second feature extraction unit 102. Either the feature extracted by the unit 100 or the feature extracted by the second feature extraction unit 102 may be defined.

  In addition, the first rule may be defined for a plurality of features extracted by the table feature extraction unit 108, the photograph feature extraction unit 110, and the encoded feature extraction unit 112 in the second feature extraction unit 102.

  The second rule in the present embodiment is a logic applied to the document features defined in the first rule. The logic is negation (NOT), logical sum (OR), logical product (AND), negative logical sum ( NOR) and logical operations such as NAND (NAND) are defined.

  FIG. 5 is a schematic diagram illustrating an example of a classification rule generated by the classification rule generation unit 82. As shown in FIG. 5, each classification rule includes a first rule that defines a feature used for classification determination, and a logical operation that is performed on the feature defined in the first rule. It consists of two rules.

  In FIG. 5, the @ mark in the first rule indicates a characteristic that the character on the left side of the @ mark exists at the position of the document designated on the right side of the @ mark. Here, the position of the document includes the title position, the body description position, the lower part of the document, and the like, and may be an area designated by the operator. In the second rule, the x mark indicates a logical product (AND), the + mark indicates a logical sum (OR), and the-mark indicates a negative (NOT).

  According to such a classification rule, documents are classified as follows. If the result of the logical operation of the second classification rule is true based on the authenticity of the feature defined in the first classification rule, the document is classified into the classification defined by the classification rule.

  In the classification rule for classification 1 shown in FIG. 5 (a), the feature A that the text “instructions” exists at the title position of the document, and the text “abc” exists at the text position of the document. A feature B is defined in the first rule. Here, since the logical operation shown in the second rule is A × B, there is a character “description” at the title position of the document, and a character “abc” in the text position of the document. True for a document that satisfies the condition, and the document is classified into category 1.

  In the classification rule for classification 2 shown in FIG. 5B, the feature A and the feature C that a photographic image exists in the document are defined in the first rule. Here, since the logical operation shown in the second rule is A × C, for a document that satisfies the condition that the text “explanation” exists at the title position of the document and that a photographic image exists in the document. True, and the document is classified into Category 2.

  In the classification rule for classification 3 shown in FIG. 5C, the feature A and the feature B are defined in the first rule. Here, since the logical operation shown in the second rule is A + B, a document in which the characters “instructions” exist at the title position of the document, or a document in which the characters “abc” exists in the text position of the document. On the other hand, the result is true, and the document is classified into classification 3.

  In the classification rule for classification 4 shown in FIG. 5D, the feature A and the feature B are defined in the first rule. Here, since the logical operation shown in the second rule is A × (−B), there is a character “description” at the title position of the document, and “abc” is at the body position of the document. The result is true for a document that satisfies the condition that no character exists, and the document is classified into category 4.

  In the example shown in FIG. 5, regarding the character of the presence of the character, the position in the document is specified by using the @ mark, but the character (for example, “Description” is not specified without specifying the position). The presence or absence of a note ") may be a condition.

  In the above example, the feature C only requires whether or not a photographic image exists in the document. For example, whether or not a photographic image having N × N pixels or more exists in the document, May be present in the designated area in the document, or whether a photographic image of N × N pixels or more is present in the designated area in the document.

  In this way, the classification rule generation unit 82 generates a classification rule that is freely combined with respect to the features extracted by the feature extraction unit 76, and performs classification according to the operator's request.

  Next, generation of classification rules in the present embodiment will be described in detail. As described above, the classification rule generation unit 82 according to the present embodiment generates a classification rule using the table that associates the document type stored in the rule template storage unit 84 with the rule template corresponding to the document type.

  FIG. 6 is a table showing an example of a table stored in the rule template storage unit 84. The table shown in FIG. 6 stores a rule template generated by the classification rule generating unit 82 for each document type specified by the document type specifying unit 78.

  Note that the template of the classification rule shown as an example in FIG. 6 is a template related to the feature by the first feature extraction unit 100 (the classification rule of the row in which the column of the object is “character” in FIG. 6). And a template related to the feature by the second feature extraction unit 102 (the classification rule of the row in which the object column is “image” in FIG. 6).

  In FIG. 6, for example, when the document type specified by the document type specifying unit 78 is an application form, the classification rule generation unit 82 generates a template ““ application form ”@title for the character object in generating the classification rule. × Use the keyword “text” and the template “table image” for the image object.

  Here, "" Application Form "@Title" means that, as mentioned above, there is a character "Application Form" at the title position in the document, and "Keyword @ Body" This means that there is a condition that there is a keyword at the text position in the document. In addition, “x” connecting both conditions means that logical product is used for logical operation in the second rule.

  Further, the “keyword” in the template means that a classification rule is generated by applying any of the characters extracted by the first feature extraction unit 100. For example, the classification rule generation unit 82 generates a classification rule by applying the character with the highest priority among the characters extracted by the first feature extraction unit 100. The priority order is determined by, for example, the appearance frequency in the document, the relevance with the character received by the character information receiving unit 86 described above, and the like.

  The classification rule generation unit 82 according to the present embodiment generates a classification rule by combining the condition for the character object shown in the template table and the condition for the image object by a logical product. For example, other logical operations such as logical sum may be combined.

  In the above example, for example, if the keyword K having the highest priority among the keywords extracted by the first feature extraction unit 100 is the keyword K for the document P that is the basis for generating the classification rule, the classification rule generation is performed. The classification rule generated by the unit 82 has the following configuration. That is, the classification rule generation unit 82 sets, as the first rule, the feature D ““ application form ”@title”, the feature E ““ keyword K ”@text”, and the feature F “table image, size and position of the document P”. Are defined, and a classification rule in which logic “D × E × F” is defined as the second rule is generated and output to the classification rule presentation unit 88. Note that the same degree means that the difference between the two is within a predetermined threshold.

  In the feature F, the size and position of the front image are required. However, the presence or absence of the image may be required, the size of the image may be required, or the position of the image is required. It is good. Further, it may be stored as a template whether the presence / absence, size, or position of an image is a requirement.

  In addition to the features extracted by the feature extraction unit 76, the classification rule generation unit 82 may put a reading condition from the reading device 14 in the classification rule. For example, the first rule defines the feature D, the feature E, the feature F, and the feature G “a document obtained by reading n or more originals”, and the logic “D × E ×” as the second rule. A classification rule in which “F × G” is defined may be generated.

  Further, according to the reading condition, it may be determined whether or not the feature of the image object shown in the template table is defined in the first rule. For example, if the reading condition is set to read in color, the operator may consider the image included in the document as important, so the classification rule generation unit 82 at least features the image object. May be defined as a first rule to generate a classification rule.

Next, presentation of classification rules by the classification rule presentation unit 88 will be described.
FIG. 7 is a plan view illustrating a first example of a screen related to registration of classification rules displayed on the UI device 50. Note that the screens shown in FIG. 7 and FIGS. 8, 9, and 10 to be described later are displayed by the classification rule presentation unit 88, for example.

  In the screen shown in FIG. 7, the display screen can be switched by the display switching tab 120 and the display switching tab group 122. In the display switching tab 120, a screen for registering a new classification rule is displayed. In any of the tabs 122, a screen for editing a classification rule already stored in the classification rule storage unit 92 is displayed. In the example illustrated in FIG. 7, a state in which display is performed using the display switching tab 120 is illustrated.

  In the screen shown in FIG. 7, a document type name setting unit 124 for setting a document type name, a document classification rule setting unit 126 for setting a classification rule for classifying a document, and a file name to be given to the classified document are set. And a classification rule type setting unit 129 for setting a classification rule (common classification rule) for a plurality of operators such as a group or a classification rule (individual classification rule) for individual operators.

  Here, the classification rule presentation unit 88 presents the classification rule generated by the classification rule generation unit 82 to the document classification rule setting unit 126. Similar to the generation of the classification rule, the classification rule generation unit 82 also generates a document type name or file name based on the document type specified by the document type specification unit 78, for example. The document type name or file name generated by the classification rule generation unit 82 may be presented to the document type name setting unit 124 or the file name rule setting unit 128 to be set.

  In the example of the screen shown in FIG. 7, the following classification rule is presented as the classification rule generated by the classification rule generation unit 82. That is, in the example of the screen shown in FIG. 7, a document having the text “instruction” in the title portion of the document and the text “abc” in the body portion of the document is set as the document type “instruction”. A classification rule to classify is presented.

  Further, the screen shown in FIG. 7 includes a keyboard input button 130 for editing the settings presented in the document type name setting unit 124, the document classification rule setting unit 126, or the file name rule setting unit 128 by keyboard input. , An erasing button 132 for erasing, and an automatic button 134 for causing the classification rule presenting unit 88 to present it.

  Also, a registration button 136 for registering the settings presented in the document type name setting unit 124, the document classification rule setting unit 126, and the file name rule setting unit 128 in the classification rule storage unit 92, and a delete button for deleting the registered settings. 138.

  Here, when a classification rule is registered as a common classification rule by the classification rule type setting unit 129, it is registered as a classification rule that is commonly applied among a plurality of operators such as a group, and any one of the plurality of operators is registered. This classification rule is applied when a person classifies a document. On the other hand, when a classification rule is registered as an individual classification rule by the classification rule type setting unit 129, it is registered as a classification rule to be applied to each operator, and this classification rule is applied when the operator performs document classification. It does not apply to document classification by other operators.

  When the registration button 136 is operated by the operator, the classification rule registration unit 90 converts the classification rule composed of the first rule and the second rule to the classification rule storage unit 92 as a common classification rule or an individual classification rule. sign up.

  Further, in the screen shown in FIG. 7, in order to edit the classification rule generated by the classification rule generation unit 82, a keyword changing unit 140 for changing a keyword used for the classification rule, and an area for changing a region in the document used for the classification rule. A change unit 142, an image change unit 144 that changes an image used for a classification rule, a logic change unit 146 that changes a logic used for a classification rule, and a document display unit 148 that displays a document that is a basis for generating a classification rule.

  In the keyword changing unit 140, the keywords extracted by the first feature extracting unit 100 are listed, and the operator selects a keyword from among the listed keywords and sets a classification rule to use the keyword. Can be edited.

  The area changing unit 142 lists positions in the document, and the operator can edit the classification rule to use the position by selecting one of the listed positions.

  The image change unit 144 lists the types of images extracted by the second feature extraction unit 102, and the operator selects one of the listed image types to select the type of the image. The classification rules can be edited to use

  The logic changing unit 146 lists logics that can be used as classification rules as the second rule, and the operator can edit the classification rules using the listed logics.

  FIG. 8 is a plan view showing a second example of a screen related to registration of classification rules displayed on the UI device 50.

  In the screen shown in FIG. 8, the feature extracted by the first feature extraction unit 100 and the second feature extraction unit 102 (photo feature extraction unit 110) are extracted as the classification rule generated by the classification rule generation unit 82. A classification rule that combines these features is presented. Specifically, the screen shown in FIG. 8 has characters of “instructions” in the title portion of the document, and is at the same position as the photo 150 in the document displayed on the document display unit 148. A classification rule for classifying a document having a photograph of a size into a document type “instruction” is presented. Note that the same degree means that the difference between the two is within a predetermined threshold.

  Further, the screen shown in FIG. 8 displays a screen condition changing unit 152 that is a pop-up display for changing the condition for the image. The operator can select a desired condition among the conditions displayed on the screen condition changing unit 152. By selecting this condition, the condition for the image can be changed.

  FIG. 9 is a plan view illustrating a third example of a screen related to registration of classification rules displayed on the UI device 50.

  The screen shown in FIG. 9 includes the features extracted by the first feature extraction unit 100 and the second feature extraction unit 102 (encoded product feature extraction unit 112) as the classification rules generated by the classification rule generation unit 82. A classification rule that combines the extracted features is presented. Specifically, the screen shown in FIG. 9 has characters of “instructions” in the title portion of the document, and has the same size as the barcode 154 in the document displayed on the document display unit 148. A classification rule for classifying a document having a barcode into a document type “instruction” is presented. Note that the same degree means that the difference between the two is within a predetermined threshold.

  FIG. 10 is a plan view illustrating a fourth example of a screen related to registration of classification rules displayed on the UI device 50.

  In the screen shown in FIG. 10, a classification rule including only the features extracted by the second feature extraction unit 102 (photo feature extraction unit 110) is presented as the classification rule generated by the classification rule generation unit 82. . Specifically, on the screen shown in FIG. 10, a document having a photo of the same size at the same position as the photo 156 in the document displayed on the document display unit 148 is set as the document type “instruction”. A classification rule to classify is presented. Note that the same degree means that the difference between the two is within a predetermined threshold.

Next, details of the classification rule integration unit 96 will be described.
FIG. 11 is a schematic diagram of classification rules registered in the classification rule storage unit 92 before the integration by the classification rule integration unit 96. FIG. 12 is a schematic diagram of classification rules registered in the classification rule storage unit 92 and integrated by the classification rule integration unit 96.

  As shown in FIG. 11, in the classification rule before integration, a classification rule corresponding to the classification exists for each classification. When the documents are classified based on the classification rule before integration, first, it is determined whether or not the classification rule 1 is satisfied, next, whether or not the classification rule 2 is satisfied, and then, Whether or not the classification rule 3 is satisfied is determined, and so on.

  The case where document classification is performed according to the classification rule before integration will be described in detail according to the example shown in FIG. 11. First, in order to determine whether or not the classification rule 1 is met, the first of the classification rule 1 is used. It is determined whether or not the feature A and the feature B defined in the rule exist in the document, and based on the result, the logical operation (A × B) defined in the second rule is performed, and the classification rule 1 (Whether the result of the logical operation is true) is determined.

  Next, similarly, in order to determine whether or not the classification rule 2 is met, it is determined whether or not the feature A and the feature C defined in the first rule of the classification rule 2 exist in the document. Based on the result, the logical operation (A × C) defined in the second rule is performed to determine whether the classification rule 1 is met (whether the result of the logical operation is true).

  The same determination is made for the subsequent classification rules. Further, it is necessary to determine not only the classification rule belonging to the common classification rule but also the classification rule belonging to the individual classification rule.

  However, in the classification rule 1 and the classification rule 2 described above, the feature A is commonly defined in the first rule, and the presence or absence of the feature A is repeated in the determination of the classification rule 1 and the determination of the classification rule 2. There is no need to do it.

  The classification rule integration unit 96 integrates the classification rules so that the features defined in other classification rules among the features of the document defined in each classification rule do not overlap.

  As shown in FIG. 12, the classification rule integration unit 96 integrates the first rule of each classification rule so as not to overlap. Specifically, operand rules are generated and integrated based on the first rule. In the operand rule, the features defined in the first rule of any of the classification rules before integration are defined one by one.

  As shown in FIG. 12, the classification rule integration unit 96 integrates the second rule of each classification rule in association with the classification. Specifically, a logical rule is generated and integrated by associating the second rule in each classification rule with the classification based on the second rule. The logical rule includes the second rule (“A × B” and “A × C” illustrated in FIG. 12) used for classification into each document type, and the document type (exemplified in FIG. 12). They are defined in association with classification 1 "," class 2 ").

  The operand rule may be generated for each of the common classification rule and the individual classification rule, or one operand rule may be generated for the common classification rule and the individual classification rule.

  According to the classification rules after the integration, the document classification process according to the classification rules is classified based on the classification rules before the integration without redundant determination of the presence or absence of the same feature in the document. It is faster than the case.

Next, an operation flow of registration of the classification rule described above will be described using a flowchart.
FIG. 13 is an example of a flowchart showing an operation for registering a classification rule.

  In step 100, the document reception unit 72 receives a document that is a basis for generating a classification rule, and outputs the received document to the object separation unit 74.

  In step 102, the object separation unit 74 separates the object in the document and outputs it to the feature extraction unit 76.

  In step 104, the feature extraction unit 76 extracts features included in the document, and outputs the extracted features to the document type identification unit 78 and the classification rule generation unit 82.

  In step 106, the document type specifying unit 78 specifies which of the predetermined document types the document received by the document receiving unit 72 in step 100 corresponds to based on the extracted features.

  In step 108, the classification rule generation unit 82 selects a template corresponding to the document type identified by the document type identification unit 78 from the rule templates for each document type stored in the rule template storage unit 84 and a feature extraction unit. A classification rule is generated using the features extracted by 76.

  In step 110, the classification rule presentation unit 88 presents the classification rule generated by the classification rule generation unit 82 to the UI device 50.

  In step 112, it is determined whether or not an editing operation has been performed by the operator for the classification rule presented in step 110. If an editing operation has been performed, the process proceeds to step 114, and if no editing operation has been performed. Control goes to step 116.

  In step 114, the edit receiving unit 94 receives an editing operation by the operator and performs editing corresponding to the editing operation on the classification rule.

  In step 116, the classification rule registration unit 90 registers the classification rule generated by the classification rule generation unit 82 or the classification rule edited by the editing reception unit 94 in the classification rule storage unit 92.

  In step 118, the classification rule integration unit 96 integrates the classification rules stored in the classification rule storage unit 92.

  Through the above flow, an integrated classification rule used for document classification is created.

Next, the flow of operations for classifying documents based on the integrated classification rules and performing processing according to the classification will be described.
FIG. 14 is an example of a flowchart illustrating an operation of classifying documents and performing processing according to the classification.

  In step 200, the document reception unit 72 receives a document to be classified and outputs the received document to the object separation unit 74.

  In step 202, the object separation unit 74 separates the object in the document and outputs it to the feature extraction unit 76.

  In step 204, the feature extraction unit 76 extracts features included in the document and outputs the extracted features to the classification unit 98.

  In step 206, the classification unit 98 applies the classification rule integrated by the classification rule integration unit 96 to the features extracted by the feature extraction unit 76 to classify the document.

  Specifically, the classification unit 98 performs the following processing for the common classification rule and the individual classification rule. The classification unit 98 determines whether or not the document to be classified has each feature defined in the operand rule of the integrated classification rule. A logical rule is applied based on the determination result for each feature (true if the feature is present, false if the feature is not present). The classifying unit 98 classifies the document into types corresponding to the logical operation whose operation result is true among the logical operations defined in the logic rule.

  In step 208, processing corresponding to the classification made in step 206 is performed. In step 208, processing is performed in accordance with, for example, a processing rule determined for each classification.

  FIG. 15 is a table showing an example of processing rules for each classification applied when processing according to the classification. FIG. 15A shows the rules for the process of assigning the file name corresponding to the classification to the document data, and FIG. 15B shows the rules for the process of outputting the document data to the place corresponding to the classification. ing.

  In the processing rule shown in FIG. 15A, a rule for assigning a file name is defined for each classification. Note that the file name assignment rule shown in FIG. 15A corresponds to the rule set in the file name rule setting unit 128 described above.

  In the processing rule shown in FIG. 15B, a transmission destination folder is defined for each classification. Note that the transmission destination folder shown in FIG. 15B is set by the operator for each classification, for example.

  Further, as processing according to the classification, printing by the printing apparatus 12 may be performed. For example, processing for printing out may be performed with different settings in printing according to the classification.

  According to the above flow, document classification and processing corresponding to the classification are performed.

  In the above description, classification rule integration (step 118) is performed after classification rule registration (step 116). However, classification rule integration may be performed before document classification processing. For example, classification rules may be integrated between step 204 and step 206. The integration of the classification rules may be performed each time a predetermined number of classification rules are registered in the classification rule registration unit 90.

  In addition, the classification rule integration unit 96 may integrate the classification rules before or after the integration similarly generated by another document processing apparatus and the classification rules before or after the integration of the own apparatus.

  In the above description, the classification rule integration unit 96 integrates the classification rules stored in the classification rule storage unit 92 and stores the integrated classification rules in the classification rule storage unit 92. However, the integrated classification rules are classified. You may memorize | store in another database different from the rule storage part 92. FIG.

  Further, the classification rule before integration or the classification rule after integration may be stored in an external database connected to the document processing apparatus via a network. In this case, the document processing apparatus has means for reading the classification rules from an external database during classification rule integration or document classification processing.

DESCRIPTION OF SYMBOLS 10 Image forming apparatus 14 Reading apparatus 50 UI apparatus 72 Document reception part 74 Object separation part 76 Feature extraction part 78 Document specification part 80 Document classification storage part 82 Classification rule generation part 84 Rule template storage part 86 Character information reception part 88 Classification rule presentation Unit 90 classification rule registration unit 92 classification rule storage unit 94 edit reception unit 96 classification rule integration unit 98 classification unit 100 first feature extraction unit 102 second feature extraction unit 104 character recognition unit 106 character feature extraction unit 108 table feature extraction Unit 110 photographic feature extraction unit 112 encoded product feature extraction unit

Claims (7)

Classification rule registration means for registering a combination of a first rule and a second rule applied to the first rule as a classification rule for classifying documents;
A document processing apparatus comprising: a classification rule integration unit that integrates a plurality of different classification rules registered by the classification rule registration unit based on content overlapping between the plurality of different classification rules. The classification rule registering means defines a first rule that defines a document feature used for document classification, and a second logic that defines logic applied to the document feature defined by the first rule. Register a combination with the rule of as a classification rule,
The document processing apparatus according to claim 1, wherein the classification rule integration unit integrates the classification rules by eliminating duplication of document features defined in the first rule. The document processing apparatus according to claim 2, wherein the classification rule registration unit uses, as a first rule, a rule that defines a character, photograph, figure, table, or encoded product included in a document as a feature of the document used for document classification. . The classification rule registration means is a rule that defines a combination of two or more of characters, photos, figures, tables or encoded products included in a document as a document characteristic used for document classification, as a first rule. The document processing apparatus according to claim 3, wherein a rule that defines a logic to be applied to any two or more combinations is set as a second rule. The document processing apparatus according to claim 1, wherein the classification rule registration unit registers an individual classification rule and a common classification rule among a plurality of operators for each operator. A document receiving means for receiving a document;
The document processing apparatus according to claim 1, further comprising: a classifying unit that classifies the document received by the document receiving unit based on the classification rule registered by the classification rule registering unit. Registering a combination of a first rule and a second rule applied to the first rule as a classification rule for classifying documents;
A program for causing a computer to execute a step of integrating a plurality of registered different classification rules on the basis of contents overlapping between the plurality of different classification rules.

Images