Information processing device, computer readable storage medium, and computer program product

Abstract

The present invention relates to an information processing device, a computer-readable storage medium, and a computer program product. The present invention can extract object information corresponding to a keyword even if the object information is not recorded within a range determined based on the position of the keyword. When a reference character string (23) indicating a reference target of object information corresponding to the keyword is recorded within a range determined based on the position of a keyword (21) recorded in a document (20), the processor extracts the object information from a page (20b) of the reference target indicated by the reference character string (23).

Description

Technical Field

[0001] The present invention relates to an information processing device, a computer-readable storage medium and a computer program product. Background Art

[0002] There is a known technique for extracting object information from a document using a key-value extraction technique. For example, Patent Document 1 discloses a technique for determining an area containing a keyword and performing a text recognition process on the determined area to extract a value.

[0003] [Prior art literature]

[0004] [Patent Document]

[0005] Patent Document 1: Japanese Patent Application Publication No. 2018-128996 Summary of the invention

[0006] [Problems to be solved by the invention]

[0007] In the conventional key-value extraction technology, object information is extracted from the vicinity of a keyword. However, as an example, there is a case where, in a document such as a contract or a complaint, a character string indicating a reference target of the object information is recorded around the keyword instead of the object information, and the object information itself is recorded in the reference target. In this case, since the object information is not located around the keyword, the conventional key-value extraction technology cannot extract the object information.

[0008] An object of the present invention is to extract object information corresponding to a keyword even when the object information is not described within a range determined based on the position of the keyword.

[0009] [Technical means to solve the problem]

[0010] The invention of technical solution 1 is an information processing device, including a processor, which extracts the object information from the reference target represented by the character string when a character string representing a reference target of object information corresponding to the keyword is recorded within a range determined based on the position of the keyword recorded in the document.

[0011] The invention of Technical Solution 2 is an information processing device according to Technical Solution 1, wherein the keyword is a first keyword, and the processor uses pattern matching to detect a second keyword corresponding to the first keyword from the reference target, and extracts the object information from a range determined based on the position of the second keyword.

[0012] The invention of Technical Solution 3 is an information processing device according to Technical Solution 1 or 2, wherein the processor detects the title of at least one page among multiple pages contained in the document that is different from a page recording the keyword, and extracts the object information from another page among the at least one page that has the title corresponding to the string.

[0013] The invention of technical solution 4 is an information processing device according to any one of technical solutions 1 to 3, wherein the processor changes the document so that the extracted object information is displayed in association with the keyword.

[0014] The invention according to claim 5 is the information processing device according to claim 4 , wherein the processor adds the object information at a position displayed together with the keyword.

[0015] The invention according to claim 6 is the information processing device according to claim 4 , wherein the processor adds a link to the reference target to the character string.

[0016] The invention of technical solution 7 is an information processing device according to any one of technical solutions 1 to 6, wherein when the processor records multiple object information candidates corresponding to the keyword in the reference target, it extracts any one object information selected from the multiple object information candidates.

[0017] The invention of technical solution 8 is an information processing device according to technical solution 7, wherein the processor displays a screen for accepting an operation of selecting any one of the multiple object information candidates on the display unit, and extracts the object information selected by the operation.

[0018] The invention of Technical Solution 9 is an information processing device according to Technical Solution 7, wherein the processor determines the certainty of the multiple object information candidates, and selects any one object information from the multiple object information candidates based on the certainty.

[0019] The invention of claim 10 is an information processing device according to any one of claims 7 to 9, wherein the plurality of object information candidates are recorded in the plurality of reference targets represented by the character string,

[0020] The processor adds, to the character string, a link to a reference target in which the selected object information is described among the plurality of reference targets.

[0021] The invention of technical solution 11 is a computer-readable storage medium storing a program, wherein the program is used to cause a computer to execute the following steps, namely, when a character string representing a reference target of object information corresponding to the keyword is recorded within a range determined based on the position of the keyword recorded in the document, the object information is extracted from the reference target represented by the character string.

[0022] [Effects of the Invention]

[0023] According to the invention of claim 1, even when the object information is not described in the range determined based on the position of the keyword, the object information corresponding to the keyword can be extracted.

[0024] According to the invention of claim 2, the time required to extract the object information is shortened compared to the case where the object information is searched from the entire reference targets.

[0025] According to the invention of claim 3 , even when a keyword is recorded on one page and object information is recorded on another page, object information corresponding to the keyword can be extracted.

[0026] According to the invention of claim 4 , even when the target information is not described in the range determined based on the position of the keyword, the keyword and the target information can be associated and browsed.

[0027] According to the invention of claim 5 , even when the target information is not described in the range determined based on the position of the keyword, the keyword and the target information can be browsed together.

[0028] According to the invention of claim 6 , even when the target information is not described in the range determined based on the position of the keyword, the target information can be browsed by an operation using a link.

[0029] According to the invention of claim 7, even when a plurality of object information candidates corresponding to the keyword are described in the reference target, one piece of object information can be extracted.

[0030] According to the invention of claim 8, even when a plurality of object information candidates corresponding to the keyword are described in the reference target, one piece of object information desired by the user can be extracted.

[0031] According to the invention of claim 9, even when a plurality of object information candidates corresponding to the keyword are described in the reference target, one piece of object information can be extracted without requiring any operation by the user.

[0032] According to the invention of claim 10, even when a plurality of object information candidates corresponding to a keyword are described in a reference destination, the selected object information can be browsed by an operation using a link.

[0033] According to the invention of claim 11, even when the object information is not described in the range determined based on the position of the keyword, the object information corresponding to the keyword can be extracted. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 1 is a diagram showing an example of the configuration of the image processing device 10 .

[0035] Figure 2 This is a diagram showing an example of a document.

[0036] Figure 3 This is a diagram showing another example of a document.

[0037] Figure 4 This is a flowchart showing an example of the operation of the image processing device 10 .

[0038] Figure 5 This is a flowchart showing an example of the operation of the image processing device 10 .

[0039] Figure 6 This is a diagram showing another example of a document.

[0040] Figure 7 This is a flowchart showing an example of a process of extracting object information from a reference target.

[0041] Figure 8 This is a diagram showing an example of a selection screen.

[0042] Explanation of symbols

[0043] 10: Image processing device

[0044] 11: Processor

[0045] 12: Memory

[0046] 13: Image reading unit

[0047] 14: Storage

[0048] 15: Operation Department

[0049] 16: Display unit DETAILED DESCRIPTION

[0050] 1. Structure

[0051] Figure 1: is a diagram showing an example of the structure of an image processing device 10 (an example of an information processing device) of the present embodiment. The image processing device 10 has a scanning function and reads a document. At this time, the image processing device 10 extracts object information based on keywords recorded in the document. The object information is also called a value, which is information desired by the user. The object information may also include at least one of text, numbers, symbols and images. The image may also be a photo, a stamp or a graphic. The keyword is a string of characters that serves as a mark for the object information. The keyword may also include at least one of text, numbers and symbols. That is, in the present embodiment, the words "keyword", "string" or "text" are not only text, but may also include numbers or symbols. The object information extracted from the document is used for various processing. For example, when the object information is a deadline, the object information may also be used for deadline management. At this time, in order to prevent the user from missing the deadline, the user may also be notified when the deadline approaches.

[0052] The image processing device 10 includes a processor 11, a memory 12, an image reading unit 13, a storage unit 14, an operation unit 15, and a display unit 16. These hardware elements are connected via a bus 17. The processor 11 controls each unit of the image processing device 10 by executing a program, and performs a process of extracting object information from a document. For the processor 11, for example, a central processing unit (CPU) can also be used. The memory 12 stores the program executed by the processor 11. For example, a read-only memory (ROM) and a random access memory (RAM) can also be used for the memory 12. The image reading unit 13 reads a document and converts it into document data. For example, an image scanner can also be used for the image reading unit 13. The storage unit 14 stores the document data converted by the image reading unit 13. For example, a hard disk drive (hard disk drive) or a solid state drive (SSD) can also be used for the storage unit 14. The operation unit 15 supplies an operation signal corresponding to the user's operation to the processor 11. For example, a touch panel and buttons may be used for the operation unit 15. The display unit 16 displays a document corresponding to the document data. For example, a liquid crystal display may be used for the display unit 16.

[0053] Figure 2 This is a diagram showing an example of a document. Figure 2 The document 20 shown is a bill having a plurality of pages. The document 20 includes a first page 20a and a page 20b titled "Attachment". Here, it is assumed that the target information desired by the user is a date indicating a payment deadline.

[0054] On page 20a, a keyword 21 (an example of a first keyword) "payment deadline" is recorded as a mark of the object information. Within a range (hereinafter referred to as "surrounding range") 22 based on the position of the keyword 21, a character string (hereinafter referred to as "reference character string") 23 "reference attached page" indicating a reference target for the payment deadline is recorded. This means that the object information corresponding to keyword 21 is recorded on page 20b with the title "Attached Page". In addition, Figure 2 The reference character string 23 shown is text, but the reference character string 23 is not limited to text, and may include at least one of text, numbers, and symbols. Here, for the keyword 21, the direction and distance from the position of the keyword 21 where the object information is generally considered to exist are predefined. The surrounding range 22 is set to a size range of less than a specified distance extending from the position of the keyword 21 in a specified direction. For example, the surrounding range 22 may be as follows: Figure 2 As shown in FIG. 1 , the range extends in the front, back, left, and right directions from the position of the keyword 21. In addition, the above directions and distances may be different for each keyword.

[0055] At the top of the page 20b, a title 24 of "Attachment" is recorded in a font size larger than the base font size. In addition, on the page 20b, local keywords 25-1 and 25-2 (an example of a second keyword) of "Payment" corresponding to the keyword 21 of "Payment Deadline" are recorded. In the surrounding area 26 of the local keyword 25-1, the date 27 of "October 31, 2018" is recorded.

[0056] Figure 3 This is a diagram showing another example of a document. Figure 3 The document 30 shown is a bill having multiple pages. The document 30 includes the first page 30a, a page 30b with a title "Attachment 1", and a page 30c with a title "Attachment 2". Figure 2 In the example shown, it is assumed that the target information desired by the user is a date indicating a payment deadline.

[0057] On page 30a, with Figure 2 Similarly, the page 20a shown in FIG. 1 contains a keyword 31 (an example of a first keyword) called "payment deadline". Figure 2 Similarly, in the illustrated page 20a, a reference character string 33 of "see attached page" is described in the surrounding area 32 of the keyword 31. This means that the object information corresponding to the keyword 31 is described in the page 30b or the page 30c having a title including "attached page".

[0058] At the top of the page 30b, a title 34b "Attachment 1" is recorded in a font size larger than the base font size. In addition, on the page 30b, local keywords 35b-1 and 35b-2 (an example of a second keyword) "Payment" corresponding to the keyword 31 "Payment Deadline" are recorded. In the surrounding area 36b of the local keyword 35b-1, the date 37b "October 31, 2018" is recorded.

[0059] At the top of the page 30c, a title 34c "Attachment 2" is recorded in a font size larger than the base font size. In addition, on the page 30c, local keywords 35c-1 and 35c-2 "payment" corresponding to the keyword 31 "payment deadline" are recorded. In the surrounding area 36c of the local keyword 35c-1, a date 37c "November 1, 2018" is recorded.

[0060] 2. Action

[0061] Figure 4 and Figure 5 1 is a flowchart showing an example of the operation of the image processing device 10. In the following description, the processor 11 is described as the main body of the processing, but the meaning is that the processor 11 performs calculations or controls the operation of other hardware elements through the cooperation of the program stored in the memory 12 and the processor 11 executing the program, thereby performing processing.

[0062] Before performing the above-mentioned operation, at least one keyword is set by the user. Figure 4 and Figure 5 In the example shown, the first keyword is first retrieved in order from the first page of the document. When the keyword is detected, the object information corresponding to the keyword is extracted. When the object information corresponding to the first keyword is extracted, the same process is performed for the next keyword. In this way, the process is repeated until the object information is extracted for all keywords.

[0063] Figure 4 The process shown is started, for example, in response to an operation in which a user sets a document in the image processing apparatus 10 and instructs the start of reading using the operation unit 15. In step S11, in response to the operation, the processor 11 causes the image reading unit 13 to read the document.

[0064] In step S12, the characters recorded in the document read in step S11 are recognized. For the character recognition, optical character recognition (OCR) is used, for example.

[0065] In step S13, the processor 11 searches for the object keyword in the object page of the document. In the initial state, the object page is the first page, and the object keyword is the first keyword. Here, it is assumed that the first keyword is "payment deadline". At this time, the keyword "payment deadline" is retrieved in the first page.

[0066] In step S14, it is determined whether the target keyword is detected in the search result of step S13. If the target keyword is not detected (the determination of step S14 is NO), the process proceeds to step S15, where the processor 11 determines whether the target page is the last page. If the target page is the last page (the determination of step S15 is YES), the process proceeds to step S16. Figure 5 On the other hand, if the target page is not the last page (the determination of step S15 is no), the process proceeds to step S16, and the processor 11 sets the next page as the target page, and then returns to the step S13. Thus, the process of step S13 to step S14 is repeated for the next page. On the other hand, if the target keyword is detected (the determination of step S14 is yes), the process proceeds to step S17. Moreover, the target keyword detected in step S14 is stored in the memory 12.

[0067] In step S17, the processor 11 searches for a reference string in the surrounding range of the object keyword detected in step S14. Here, it is assumed that the reference string is "refer to attached sheet". In step S18, the processor 11 determines whether the reference string is detected in the search result of step S17. If the reference string is not detected (the determination of step S18 is no), the process proceeds to step S19, and the processor 11 extracts object information from the surrounding range of the object keyword. The object information extracted in step S19 is stored in the memory 12.

[0068] Figure 6 This is a diagram showing another example of a document. Figure 6 The document 40 shown is a bill with multiple pages. The first page 40a is included in the document 40. In page 40a, the keyword 41 "payment deadline" is recorded, so the keyword 41 is detected in the step S14. Moreover, in the surrounding range 42 of the keyword 41, although the date 43 "October 31, 2018" is recorded, the reference string "refer to the attached page" is not recorded, so the reference string cannot be detected in the step S18. Here, for the keyword "payment deadline", the format of the corresponding object information is pre-defined as a date format. At this time, in the step S19, the date 43 "October 31, 2018" in the date format is extracted from the surrounding range 42 of the keyword 41 recorded on the page 40a as the object information, and stored in the memory 12.

[0069] On the other hand, in step S18, if a reference character string is detected (step S18 is determined to be yes), the process proceeds to step S20, and the processor 11 performs the following processing, namely, extracting object information corresponding to the object keyword from the reference target represented by the reference character string detected in step S18. Figure 2 In the example shown, since the keyword 21 "payment period" is recorded in the first page 20a, the keyword 21 is detected in step S14. Since the reference character string 23 "see attached sheet" is recorded in the surrounding range 22 of the keyword 21, the reference character string 23 is detected in step S18. At this time, the following processing is performed in step S20, that is, the object information corresponding to the object keyword is extracted from the reference target represented by the reference character string 23.

[0070] Figure 7 1 is a flowchart showing an example of the process of extracting object information from the reference target in step S20. In step S31, the processor 11 detects titles from each page of the document. In addition, the title detection may not necessarily be performed on all pages, for example, it may be performed on pages after the target page, but not on the target page. The title detection is performed based on, for example, the position where the title is thought to be recorded in the page and the font size that is thought to be used for the title. Figure 2 In the document 20 shown, a title is recorded at the top of each page in a font size larger than the base font size. At this time, the character string recorded at the top of each page in a font size larger than the base font size is detected as a title. For page 20b, the character string "Appendix" is detected as the title 24. The same method is used to detect the title for other pages included in the document 20. In addition, if the title cannot be detected, the character string recorded in the first line of the page can also be used as the title. Moreover, if the title has been detected in step S12, step S31 can also be skipped.

[0071] In step S32, the processor 11 determines the reference target based on the title detected in step S31. For example, a page having a title corresponding to the reference string is determined as the reference target. The title corresponding to the reference string may also include a title containing at least a portion of the reference string or a title similar to the reference string. Figure 2 In the example shown, the title 24 of the page 20b includes a character string "Attached Page" which is a part of the reference character string 23. In this case, the page 20b having the title 24 is determined as the reference target.

[0072] In step S33, the processor 11 uses pattern matching to search for the object keyword in the reference target determined in step S32. For the pattern matching, for example, a local keyword represented by a regular expression of the object keyword is used. This is because in the reference target, the keyword is sometimes not directly recorded in this expression, but recorded in other expressions. Figure 2 In the example shown, partial keywords such as "payment" and "date" are used, which are represented by the regular expression of the keyword 21 of "payment deadline". At this time, the page 20b serving as the reference target is moved to search for the partial keywords.

[0073] In step S34, the processor 11 determines whether the pattern of the target keyword is detected. If the pattern of the target keyword is not detected (the determination of step S34 is No), the process proceeds to step S35, and the processor 11 determines whether the reference target is the last page. If the reference target is the last page (the determination of step S35 is Yes), the process proceeds to step S36. Figure 5 On the other hand, if the reference target is not the last page (the determination of step S35 is No), the process proceeds to step S36, and the processor 11 sets the next page as the reference target, and then returns to step S33. Thus, the process of steps S33 to S34 is repeated for the next page. On the other hand, if the pattern of the target keyword is detected in step S34 (the determination of step S34 is Yes), the process proceeds to step S37. Figure 2 In the example shown, the local keywords 25-1 and 25-2 of "payment" are described in the page 20b, and therefore these local keywords 25-1 and 25-2 are detected. Then, the process proceeds to step S37.

[0074] In step S37, the processor 11 extracts object information from the surrounding area of ​​the local keyword detected in step S34. The object information extracted in step S37 is stored in the memory 12. Here, as described above, for the keyword "payment deadline", the format of the corresponding object information is predefined as a date format. At this time, Figure 2 In the example shown, a date 27 in a date format of “October 31, 2018” is extracted from a surrounding range 26 of a local keyword 25 - 1 recorded on the page 20 b as object information and stored in the memory 12 .

[0075] When the process of extracting object information from the reference target is completed, proceed to Figure 5In step S21, the processor 11 determines whether the object information corresponding to all the keywords has been extracted. If the object information corresponding to all the keywords has not been extracted (the determination of step S21 is no), the process proceeds to step S22, and the processor 11 sets the next keyword as the object keyword and returns to step S23. Figure 4 Step S13 is shown. Thus, the processing from step S13 to step S21 is repeated for the next keyword. On the other hand, if the object information corresponding to all the keywords has been extracted (the determination of step S21 is yes), the process proceeds to step S23.

[0076] In step S23, the processor 11 determines whether the number of object information corresponding to all keywords is less than 1. For example, if object information is extracted one by one for all keywords (the determination of step S23 is yes), the process proceeds to step S26. On the other hand, if a plurality of object information candidates are extracted for at least one keyword (the determination of step S23 is no), the process proceeds to step S24.

[0077] Figure 3 In the example shown, Figure 2 In the example shown, the keyword 31 "payment deadline" recorded on page 30a and the reference character string 33 "see attached page" recorded in the surrounding range 32 of the keyword 31 are detected. However, the title 34b of page 30b and the title 34c of page 30c both contain the character string "attached page" which is a part of the reference character string 33, so both page 30b and page 30c are determined as reference targets. In addition, for page 30b, the local keywords 35b-1 and 35b-2 of "payment" recorded on page 30b are detected, and further, the date 37b of "October 31, 2018" is extracted from the surrounding range 36b of the local keyword 35b-1. For page 30c, the local keywords 35c-1 and 35c-2 of "payment" recorded on page 30c are detected, and further, the date 37c of "November 1, 2018" is extracted from the surrounding range 36c of the local keyword 35c-1. As described above, when a plurality of object information candidates are extracted for a single keyword, the process proceeds to step S24.

[0078] In step S24 , the processor 11 causes the display unit 16 to display a selection screen that accepts an operation to select any one of the plurality of object information candidates. Figure 8 FIG. 1 is a diagram showing an example of a selection screen. Figure 8 The selection screen 50 shown includes Figure 3The two object information candidates extracted by the keyword "payment deadline" 31 are date 37b "October 31, 2018" and date 37c "November 1, 2018", selection buttons 51-1 and 51-2 used to select these object information candidates, and an "OK" button 52 used to confirm the selection. For example, when the date 37b "October 31, 2018" is selected, the user selects the date 37b using the selection button 51-1 and then presses the "OK" button 52. In addition, when a plurality of object information candidates are extracted for a plurality of keywords, the selection screen may be configured to accept an operation of selecting any one of the plurality of object information candidates for the plurality of keywords.

[0079] In step S25, the processor 11 selects any one of the plurality of object information candidates according to the user's operation. Figure 8 As shown, when the operation of selecting the date 37b of "October 31, 2018" is performed, the date 37b is selected as the object information. At this time, only the selected date 37b among the multiple object information candidates is used as the object information, and the date 37c of "November 1, 2018" may not be used as the object information.

[0080] In step S26 , the processor 11 changes the document so that the object information extracted in step S19 or step S37 is displayed in association with the corresponding keyword, and stores it in the storage unit 14 . Figure 2In the example shown, a link to page 20b is attached to the reference character string 23 "see attached page" recorded on page 20a. For the link, for example, a hyperlink may be used. Furthermore, the display format of the document is changed so that the date 27 "October 31, 2018" recorded on page 20b is highlighted. For the highlighted display, for example, a highlight display may be used. When the user wants to view the document 20, the user displays the document 20 on the display unit 16. At this time, when the user clicks the link attached to the reference character string 23, page 20b is displayed in accordance with the operation, so that the object information, that is, the date 27 "October 31, 2018", is easily viewed. Furthermore, document data representing the document is created and stored in the storage unit 14. At this time, attribute information including a set of the keyword 21 "payment deadline" stored in the memory 12 and the date 27 "October 31, 2018", that is, the object information, is attached to the document data. For example, when the file format of the document data is Portable Document Format (PDF), a set of an object keyword and object information may also be included in the PDF attributes.

[0081] Furthermore, if the target information is selected in step S25 , the document may be changed in step S26 so that the target information is displayed in association with the corresponding keyword and stored in the storage unit 14 . Figure 3 In the example shown, for the reference character string 33 "See Appendix" recorded on page 30a, only a link to page 30b recording the date 37b selected as the object information may be attached. That is, a link to page 30c recording the date 37c not selected as the object information may not be attached. Moreover, in the attribute information of the document data, only the group of the keyword 31 "Payment Deadline" and the date 37b "October 31, 2018" selected as the object information may be included. That is, the group of the keyword 31 "Payment Deadline" and the date 37c "November 1, 2018" not selected as the object information may not be included in the attribute information. After step S26, the processing ends.

[0082] According to the embodiment, when a reference string is recorded in the surrounding area of ​​the keyword, object information corresponding to the keyword is extracted from the reference target represented by the reference string, so that even if the object information is not recorded in the surrounding area of ​​the keyword, the object information corresponding to the keyword can be extracted. Figure 2 and Figure 3As shown, even if the keyword is recorded in one page and the object information is recorded in another page, the object information corresponding to the keyword can be extracted. Moreover, since the object information is extracted from the surrounding range of the local keyword recorded in the reference target, the time consumed for extracting the object information is shortened compared with the case where the object information is searched from the entire reference target. Furthermore, since a link to the reference target where the object information is recorded is attached to the reference string, the user can browse the object information in association with the keyword by performing an operation using the link. Furthermore, even if a plurality of object information candidates corresponding to the keyword are recorded in the reference target, the user can extract one object information desired by the user by performing an operation of selecting the object information from the plurality of object information candidates. In this case, since a link to the reference target where the object information selected from the plurality of object information candidates is recorded is attached to the reference string, the user can browse the selected object information in association with the keyword by performing an operation using the link.

[0083] 3. Modifications

[0084] The above embodiment is an example of the present invention. The present invention is not limited to the above embodiment. Moreover, the above embodiment can also be implemented by modifying as in the following example. In this case, two or more of the following modification examples can also be combined and used.

[0085] (1) In the above embodiment, the method of changing a document to display object information in association with a keyword is not limited to the method of adding a link to a reference target. For example, the processor 11 may also add object information to a position displayed together with the keyword. The position displayed together with the keyword is, for example, a position displayed on the same screen as the keyword. The position may also include, for example, a position adjacent to the keyword or the reference string. Figure 2 In the example shown, the date 27 of "October 31, 2018" which is the target information may be recorded in the blank space below the reference character string 23 "See attached page" recorded on the page 20a. Figure 2 In the page 20a shown, the reference character string 23 "see attached page" is replaced with the date 27 "October 31, 2018" which is the object information. According to the modification, even when the object information is not recorded in the surrounding area of ​​the keyword, the keyword and the object information can be browsed together.

[0086] (2) In the above embodiment, Figure 3As shown, in the case where a plurality of object information candidates corresponding to a keyword are recorded in the reference target, any one of the plurality of object information candidates may be selected without user operation. At this time, the processor 11 determines the confidence of the plurality of object information candidates. The confidence is a degree indicating the accuracy. The higher the confidence, the higher the possibility that the object information candidate is the object information corresponding to the keyword. The confidence may also be determined based on, for example, the location where the object information candidate appears, other character strings located around the object information candidate, the date represented by the object information candidate, the font size of the object information candidate, or at least one combination thereof. Furthermore, the processor 11 selects any one of the object information from the plurality of object information candidates based on the confidence. For example, the object information with the highest confidence among the plurality of object information candidates may be selected.

[0087] When the certainty is determined based on the appearance position of the object information candidate, the earlier the appearance order of the object information candidate is, the higher the certainty of the object information candidate may be. In this case, the first object information that appears among the plurality of object information candidates is selected.

[0088] In the case where the certainty is determined based on other character strings located around the object information candidate, when a character string that tends not to be recorded together with the object information corresponding to the keyword is included within a specified distance from the object information candidate, the certainty of the object information candidate may also become low. Conversely, when a character string that tends to be recorded together with the object information corresponding to the keyword is included within a specified distance from the object information candidate, the certainty of the object information candidate may also become high. This is because, if a character string that tends not to be recorded together with the object information corresponding to the keyword is included around the object information candidate, there is a high possibility that the object information candidate is not the object information, and conversely, if a character string that tends to be recorded together with the object information corresponding to the keyword is included around the object information candidate, there is a high possibility that the object information candidate is the object information.

[0089] When the degree of certainty is determined based on the date indicated by the object information candidate, the earlier the date indicated by the object information candidate, the higher the degree of certainty of the object information candidate. In this case, the object information indicating the earliest date is selected from the plurality of object information candidates. This is because, for example, when the object information is used for deadline management, by using the object information indicating the earliest date among the plurality of object information, even if the selected object information is incorrect, the user can be notified before the deadline is missed.

[0090] When the degree of certainty is determined based on the font size of the object information candidate, the smaller the font size of the object information candidate, the lower the degree of certainty of the object information candidate. This is because if the font size of the object information candidate is small, there is a high possibility that the object information candidate is, for example, part of a note or annotation rather than the object information.

[0091] According to the modification, since any one of the plurality of object information candidates is selected based on the degree of certainty, even when a plurality of object information candidates corresponding to the keyword are described in the reference target, a single accurate object information can be extracted without user operation.

[0092] (3) In the above embodiment, Figure 3 As shown, when multiple object information is extracted from multiple reference targets for a single keyword, links to the reference targets may be added to the corresponding reference characters. For example, characters "1" and "2" may be added after the reference character string 33 "Refer to attached page" recorded on page 30a, and a link to page 30b may be added to character "1", and a link to page 30c may be added to character "2". In addition, the display format may be changed so that both the date 37b "October 31, 2018" recorded on page 30b and the date 37c "November 1, 2018" recorded on page 30c are highlighted. Furthermore, attribute information including the following two may be added to the document data representing the document 30, namely, a group of the keyword 31 "Payment deadline" and the date 37b "October 31, 2018", and a group of the keyword 31 "Payment deadline" and the date 37c "November 1, 2018". According to the modification, when the user displays the document 30 on the display unit 16 and clicks the link attached to the characters "1" and "2" after the reference character string 33, the pages 30b and 30c are displayed in accordance with the operation, so that the date 37b and the date 37c, which are the object information candidates, can be easily viewed. As described above, after viewing the date 37b and the date 37c, the user can also select the desired object information from these object information candidates.

[0093] (4) In the above-described embodiment, keywords and object information may not necessarily be recorded in different pages. For example, object information may be recorded in a location away from keywords on a single page. In documents such as contracts, there is the following situation: on a single page, a character string "Refer to Article 5 of this contract" is recorded in the surrounding area of ​​the keyword, and object information corresponding to the keyword is recorded in Article 5 which is located outside the surrounding area of ​​the keyword. In this case, the character string "Refer to Article 5 of this contract" becomes a reference character string, and object information is extracted from the reference target "Article 5" represented by the reference character string. As described above, keywords and object information may also be recorded in different lines, paragraphs or chapters on a single page.

[0094] (5) In the above embodiment, Figure 4 When detecting the target keyword in step S13 or searching for the reference string in step S17, pattern matching may be used. For example, when searching for the reference string in the vicinity of the keyword using pattern matching, the string recorded in the vicinity of the keyword may be detected as the reference string regardless of whether it is "refer to the attached sheet" or "specified in the attached sheet".

[0095] (6) In the above-described embodiment, the reference target may not necessarily include a local keyword. In this case, object information may be retrieved from the entire reference target.

[0096] (7) In the above embodiment, the object information is not limited to the date indicating the payment deadline. The object information may be any information as long as it is recorded in the document. For example, when the document is a bill, the object information may also be the date of the bill, the name of the request target or requesting party, or the requested amount. When the document is a contract, the object information may also be the date of the contract, the name of the signatory, the date indicating the release period or authorization period, the payment conditions or the content of the disputed points. In this case, a character string that marks these object information is used as a keyword. Moreover, the use of the object information is not limited to deadline management. For example, when the object information is the name of the request target, the document data may be saved in a folder having a folder name containing the name. Furthermore, the following process may be performed, namely, a file containing a group of keywords and object information in a prescribed format is created and output to other systems.

[0097] (8) In the above-mentioned embodiments, the so-called processor refers to a processor in a broad sense, including a general-purpose processor (such as a central processing unit (CPU)) or a special-purpose processor (such as a graphics processing unit (GPU), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), a programmable logic device, etc.)

[0098] Furthermore, the actions of the processor in the above-described embodiments may not be performed by only one processor, but may be performed collaboratively by multiple processors located at physically separated locations. Furthermore, the order of the actions of the processor is not limited to the order described in the above-described embodiments, and may also be appropriately changed. For example, the process of changing the document and storing it in step S26 may also be performed following the process of extracting the object information in step S19 or step S37. Furthermore, the process of selecting the object information in steps S23 to S25 may also be performed following the process of extracting the object information in step S19 or step S37.

[0099] (9) In the above-described embodiment, the structure of the image processing device 10 is not limited to Figure 1 The example shown. The image processing device 10 may also include other hardware elements, and some of the hardware elements of the image processing device 10 may also be provided in other devices. For example, at least one of the image reading unit 13, the storage unit 14, the operation unit 15 and the display unit 16 may also be provided in other devices. Moreover, the image processing device 10 may also have a functional structure that realizes the function of executing the processing of each step performed in the image processing device 10. For example, the image processing device 10 may also include: a text recognition unit for recognizing text, a detection unit for detecting keywords, a detection unit for detecting reference character strings, an extraction unit for extracting object information, a display control unit for displaying a selection screen, a selection unit for selecting any one object information from a plurality of object information candidates, a change unit for changing a document, an attachment unit for attaching an additional link, and an attachment unit for attaching attribute information. Furthermore, the present invention may also be provided as a method including the steps of processing performed in the image processing device 10.

[0100] (10) The present invention may also be provided as a program executed in the image processing device 10. The program may be downloaded via a communication line such as the Internet, or may be provided in a state recorded in a computer-readable recording medium such as a magnetic recording medium (a magnetic tape, a magnetic disk, etc.), an optical recording medium (an optical disk, etc.), an optical magnetic recording medium, or a semiconductor memory.

Claims

1. An information processing device, characterized in that: Including processor, The processor performs the following processing: Search for keywords in the object page of the document; When a reference character string indicating a reference target page of object information corresponding to the keyword is recorded within a range determined based on a position of the detected keyword recorded in the document, detecting a title from each page of the document other than the target page; determining a page containing at least a portion of the reference string or a page having a title similar to the reference string as the reference target page; The keyword is searched for in the determined reference target page using pattern matching.

2. The information processing device according to claim 1, wherein The processor detects a local keyword represented by a regular expression of the keyword from the reference target page using pattern matching, The object information is extracted from a range determined based on the position of the local keyword.

3. The information processing device according to claim 1 or 2, wherein The processor detects a title of at least one page among a plurality of pages included in the document that is different from a page on which the keyword is recorded. The object information is extracted from another page of the at least one page having the title corresponding to the character string.

4. The information processing device according to claim 1 or 2, wherein The processor changes the document so that the extracted object information is displayed in association with the keyword.

5. The information processing device according to claim 4, wherein The processor adds the object information at a position displayed together with the keyword.

6. The information processing device according to claim 4, wherein The processor adds a link to the reference destination page to the character string.

7. The information processing device according to claim 1 or 2, wherein When a plurality of object information candidates corresponding to the keyword are described in the reference target page, the processor extracts any one of the object information selected from the plurality of object information candidates.

8. The information processing device according to claim 7, wherein the processor causes the display unit to display a screen for accepting an operation of selecting any one of the plurality of object information candidates, The object information selected by the operation is extracted.

9. The information processing device according to claim 7, wherein The processor determines the certainty of the plurality of object information candidates, Based on the certainty, any one object information is selected from the plurality of object information candidates.

10. The information processing device according to claim 8 or 9, wherein The plurality of object information candidates are recorded in a plurality of reference target pages indicated by the character string, The processor adds, to the character string, a link to a reference target page, among the plurality of reference target pages, on which the selected object information is described.

11. A computer-readable storage medium storing a program for causing a computer to execute the following steps, namely, Search for keywords in the object page of the document; When a reference character string indicating a reference target page of object information corresponding to the keyword is recorded within a range determined based on a position of the detected keyword recorded in the document, detecting a title from each page of the document other than the target page; determining a page containing at least a portion of the reference string or a page having a title similar to the reference string as the reference target page; The keyword is searched for in the determined reference target page using pattern matching.

12. A computer program product comprising a program for causing a computer to execute the following steps: Search for keywords in the object page of the document; When a reference character string indicating a reference target page of object information corresponding to the keyword is recorded within a range determined based on a position of the detected keyword recorded in the document, detecting a title from each page of the document other than the target page; determining a page containing at least a portion of the reference string or a page having a title similar to the reference string as the reference target page; The keyword is searched for in the determined reference target page using pattern matching.

Images