JP5229102B2 - Form search device, form search program, and form search method - Google Patents

Description

  The present invention relates to a form search device, a form search program, and a form search method.

  2. Description of the Related Art Conventionally, an electronic form management system that manages a management table for recording product orders and various application forms as electronic forms has been widely used.

  For example, there are the following three types of recording methods for electronic form data in this electronic form management system. In the electronic document management system of the first recording method, image data obtained by capturing a paper form by a scanner is recorded. In an electronic form management system of the second recording method, form data is recorded using a database showing the relationship between each item (heading) on the form and its contents. In an electronic form management system of the third recording method, drawing data including a character string and a drawing position of data constituting the form is recorded.

  Here, when searching for a specific electronic form from the electronic forms recorded in the electronic form management system, the search target is often limited to a specific item.

  In this case, since the electronic form management system of the first recording method records the electronic form as image data, the character code information cannot be recognized, and the electronic form is searched by limiting the search target to a specific item. Can not do it.

  In the second recording type electronic form management system, a database in which the relationship between each item constituting the electronic form and its contents is structured is used. You can easily search for forms. Even when the item is not specified, the content of the item is compared with all data in the database, and the item corresponding to the matching data is searched. As a result, the electronic form management system can search for forms using the searched items and their contents as search conditions.

  Furthermore, in the electronic document management system of the third recording method, the character code information of each character string on the form can be recognized, but since it is unknown whether the character string is an item or not, it is specified as a search condition. The relationship between the item and its contents cannot be recognized. For this reason, information indicating the relationship between items and their contents (hereinafter referred to as “item definition information”) is manually defined for each form in advance.

  Also, in such an electronic form management system, the character string of the data constituting the form is collated with the item definition information in which the item name is defined, and the character string that has been successfully collated is used as the item name character string. A character string is used as a data character string, and a data character string item is searched from these positional relationships.

JP 2003-22271 A JP 2008-204226 A

  However, when a specific electronic form is searched by limiting the search target to a specific item (heading) from a plurality of electronic forms recorded in the electronic form management system, item definition information is defined in advance. However, there is a problem that it takes time and effort.

  That is, when the item definition information is manually defined, for example, the administrator of the electronic form management system needs to define the item definition information in advance for all forms having various formats such as invoices. It will be very time consuming.

  In addition, even if the electronic form management system uses a conventional technique for searching for an item of a data character string using item definition information, depending on the positional relationship between the item name character string and the corresponding data character string, the data character string A column item cannot be searched with 100% accuracy, and as a result, an electronic form that matches the search request cannot be accurately searched. For this reason, it is necessary to visually check the item definition information including the item definition information itself in advance, which is troublesome.

  The present invention has been made in view of the above, and when searching for a specific electronic form by limiting a search target to a specific item (heading) from a plurality of electronic forms, it is troublesome in advance work. It is an object of the present invention to provide a form search device, a form search program, and a form search method that do not cost.

  In order to solve the above-described problems and achieve the object, the form retrieval apparatus stores a plurality of character strings in the form and drawing positions of the character strings for each form, and the storage means Extraction means for extracting a character string including a search keyword that forms a part of a condition for searching for a form and a drawing position of the character string from character strings in each form, and a character string extracted by the extraction means Search means for searching for a heading candidate corresponding to the character string from the form from which the character string has been extracted based on the drawing position; a heading candidate searched by the searching means; and a character string corresponding to the heading candidate Generating heading candidate information for associating with each form, outputting the generated heading candidate information, and a predetermined operation from the heading candidate information output by the output means. The form containing the string associated with the selected heading candidate by taking a structure and a form retrieval means for retrieving from the storage means.

  As described above, the form search device, the form search program, and the form search method do not require time and effort in advance when searching for a specific form by limiting the search target to a specific heading item from a plurality of forms. There is an effect.

FIG. 1 is a functional block diagram illustrating the configuration of the form retrieval apparatus according to the first embodiment. FIG. 2 is a diagram illustrating an example of a data structure of the form data information storage unit. FIG. 3 is a diagram illustrating a heading candidate search method. FIG. 4 is a diagram illustrating an example of heading candidate information. FIG. 5 is a diagram illustrating a modification of the heading candidate information. FIG. 6 is a diagram illustrating a result of selecting a headline by the user. FIG. 7 is a diagram illustrating an example when an error occurs in the search for a heading candidate. FIG. 8 is a flowchart illustrating the processing procedure of the form search apparatus according to the first embodiment. FIG. 9 is a functional block diagram illustrating the configuration of the form search apparatus according to the second embodiment. FIG. 10 is a flowchart illustrating the processing procedure of the form search apparatus according to the second embodiment. FIG. 11 is a functional block diagram illustrating the configuration of the form search apparatus according to the third embodiment. FIG. 12 is a diagram illustrating a case where the headline is corrected by the user. FIG. 13 is a diagram illustrating an example of the data structure of the headline correction information storage unit. FIG. 14 is a flowchart illustrating the processing procedure of the form search apparatus according to the third embodiment. FIG. 15 is a functional block diagram illustrating the configuration of the form search apparatus according to the fourth embodiment. FIG. 16 is a diagram illustrating an example of a data structure of the regular expression correspondence information storage unit. FIG. 17 is a diagram illustrating an example of heading candidate information. FIG. 18 is a flowchart illustrating the processing procedure of the form search apparatus according to the fourth embodiment. FIG. 19 is a functional block diagram illustrating the configuration of the form search apparatus according to the fifth embodiment.

  Hereinafter, embodiments of a form search apparatus, a form search program, and a form search method according to the present invention will be described in detail with reference to the drawings. In addition, this invention is not limited by the present Example.

  FIG. 1 is a functional block diagram illustrating the configuration of the form retrieval apparatus according to the first embodiment. As illustrated in FIG. 1, the form retrieval apparatus 1 includes an input unit 10, a control unit 20, a storage unit 30, and an output unit 40.

  Here, an outline of processing of the form retrieval apparatus 1 will be described. The form retrieval apparatus 1 is configured to search for a specific electronic form by limiting a search target to a specific heading item from all stored electronic forms without specifying item definition information in advance. Search electronic forms. Specifically, the form retrieval apparatus 1 extracts a character string including a search keyword that forms a part of a condition for retrieving an electronic form from character strings constituting each electronic form, and extracts the extracted character string. Based on the drawing position, a heading candidate corresponding to the character string is searched. Further, the form retrieval apparatus 1 generates heading candidate information in which the searched heading candidate is associated with the character string corresponding to the heading candidate, and displays and outputs the heading candidate information on the monitor. Further, the form retrieval apparatus 1 can retrieve an electronic form including a character string related to a heading candidate selected by a predetermined operation from the displayed heading candidate information.

  It should be noted that searching for a specific electronic form by limiting heading items to be searched from all electronic forms is referred to as “item limited search”. Further, the electronic form is hereinafter abbreviated as “form”. The processing performed by the form retrieval apparatus 1 will be described in detail sequentially.

  The input unit 10 is a functional unit that detects various requests, and includes a keyword input unit 11 and a headline selection unit 12.

  For example, when the keyword input unit 11 detects a search request including a search keyword input by the user, the keyword input unit 11 outputs the request to the keyword extraction unit 21 described later. Here, the search keyword is a keyword that forms part of the condition for the user to search for a form. For example, when a user performs a “search for a form including Kawasaki in the address”, the search condition is “include Kawasaki in the address”, the search range is an item “address”, and the search keyword is “Kawasaki” Become.

  When the heading selection unit 12 detects a heading candidate selected by a predetermined operation from the heading candidate information output by the output unit 40 described later, the heading selection unit 12 outputs the detected heading candidate to the form search unit 24 described later. . The predetermined operation is, for example, that the user presses the corresponding key on the keyboard, or touches the corresponding area on the touch panel when the monitor is a touch panel.

  The control unit 20 is a functional unit that controls the form search function, and includes a keyword extraction unit 21, a headline candidate search unit 22, a headline candidate list creation unit 23, and a form search unit 24.

  The storage unit 30 is a functional unit that stores information, and includes a form data information storage unit 31 and a form search result information storage unit 32.

  The keyword extraction unit 21 extracts a character string including a search keyword from character strings constituting each form data for all the form data stored in the storage unit 30. Specifically, when acquiring a search request including a search keyword from the keyword input unit 11, the keyword extraction unit 21 reads all the form data from a form data information storage unit 31 described later. In addition, the keyword extraction unit 21 extracts a character string including a search keyword from the character strings constituting each read form data, and obtains the drawing position of the extracted character string and the form data from which the character string is extracted. The data is output to the heading candidate search unit 22.

  Here, the structure of the form data stored in the form data information storage unit 31 will be described with reference to FIG. FIG. 2 is a diagram illustrating an example of the data structure of the form data information storage unit 31. As shown in FIG. 2, the form data information storage unit 31 stores form drawing data for each form.

  This drawing data refers to the drawing attribute of each element constituting the form. For example, when the element constituting the form is the character string 31c, the character code of the character and the drawing position thereof, for example, the drawing coordinates are the drawing attributes. When the element constituting the form is a straight line 31b, the drawing position of the start point and the end point of the straight line is a drawing attribute. When the element constituting the form is an arc 31d, the center position (center coordinate), radius, start angle, and end angle of the arc are drawing attributes. The drawing position may be a two-dimensional coordinate represented by the X coordinate and the Y coordinate when the upper left corner of the paper on which the form is printed is the origin, the right direction is the X coordinate, and the lower direction is the Y coordinate. good. When the form is in a table format, the drawing position may be a two-dimensional matrix represented by the number of rows and the number of columns.

  In the example of FIG. 2, form data of form A and form B is stored, and each form data includes a form identifier 31a for identifying the form and drawing data of the form. Note that the form A is a form described in FIG. 3A described later.

  Returning to FIG. 1, the headline candidate searching unit 22 searches for a headline candidate for a character string including a search keyword extracted from the form data. Specifically, when the headline candidate search unit 22 acquires the drawing position of the character string extracted by the keyword extraction unit 21 and the form data from which the character string is extracted, the heading candidate search unit 22 is based on the drawing position of the extracted character string. The heading candidate of the character string is searched from the form data from which the character string is extracted.

For example, the heading candidate search unit 22 is a character string arranged in the forward direction of the X-axis, and among the character strings that are on the same column as the character string extracted by the keyword extraction unit 21, the extracted character string Other character strings having a drawing position in front of the drawing position are searched for as heading candidates.
In addition, the heading candidate search unit 22 is a character string arranged in the forward direction of the Y-axis, and among the character strings that are on the same column as the character string extracted by the keyword extraction unit 21, the extracted character string Other character strings having a drawing position in front of the drawing position are searched for as heading candidates.

  In the case of a tabular format, the headline candidate searching unit 22 has another character string having the drawing position in the forefront in the row direction from the drawing position of the character string extracted by the keyword extracting unit 21 or the top in the column direction. Another character string having a drawing position ahead may be searched for as a heading candidate. The headline candidate search unit 22 does not limit the headline candidate search method to these.

  Here, the headline candidate search method performed by the headline candidate search unit 22 will be described with reference to FIG. FIG. 3 is a diagram illustrating a heading candidate search method. (A) explains a method of searching for a character string arranged in the forward direction in the X-axis (horizontal) direction or in the front direction in the Y-axis (vertical) direction among the character strings arranged in the forward direction, and (B). A search method when the form is in a table format will be described. In the following description, it is assumed that the search keyword is “Kawasaki”.

  In (A), the headline candidate searching unit 22 searches forward in the X-axis (horizontal) direction from the drawing position of the character string “Kawasaki-shi, Kanagawa ...” including the search keyword “Kawasaki”, and the search result Based on the above, “address” is set as a headline candidate. Specifically, the headline candidate searching unit 22 uses the drawing data including the search keyword “Kawasaki” and uses the Y coordinate among the drawing coordinates of the character string “Kawasaki City, Kanagawa ...” including the search keyword “Kawasaki”. A character string “address” having the same Y-coordinate value as the value of X and the X-coordinate value of the forward coordinate on the X-axis is searched for as a heading candidate.

  Further, for the same character string “Kawasaki City, Kanagawa Prefecture” including the search keyword “Kawasaki”, the heading candidate search unit 22 starts from the drawing position of the character string including the search keyword “Kawasaki” in the Y-axis (vertical) direction. A forward search is performed, and “123-4567” indicating a zip code based on the search result is set as a heading candidate. Specifically, the headline candidate search unit 22 uses the drawing data including the search keyword “Kawasaki” to calculate an X coordinate value closer to the X coordinate value among the drawing coordinates of the character string including the search keyword “Kawasaki”. A character string (“123-4567” instead of “zip code”) having the Y coordinate value of the forward coordinate on the Y axis is searched for as a headline candidate.

  In (B), the headline candidate search unit 22 searches the forefront in the row direction from the drawing position (cell) where the character string “Jiro Kawasaki” including the search keyword “Kawasaki” is present, and based on the result of the search, The number “3” is set as a heading candidate. For the same character string “Jiro Kawasaki” including the search keyword “Kawasaki”, the heading candidate search unit 22 searches the forefront in the column direction from the drawing position (cell) where the character string including the search keyword “Kawasaki” is present. Based on the search result, “name” is set as a heading candidate.

  Further, the heading candidate searching unit 22 searches the forefront in the row direction from the drawing position (cell) having the character string “Kawasaki city, Kanagawa” including the search keyword “Kawasaki”, and the item number is based on the search result. Let “2” be a headline candidate. For the same character string “Kawasaki City, Kanagawa” including the search keyword “Kawasaki”, the heading candidate search unit 22 moves the forefront in the column direction from the drawing position (cell) where the character string including the search keyword “Kawasaki” exists. A search is performed, and “address” is set as a heading candidate based on the search result.

  As a result, headline candidates corresponding to the character string including the search keyword are limited. In the above-described heading candidate search method, there may be a plurality of heading candidates corresponding to the character string including the search keyword, but the heading candidates that match the search request are further limited by the subsequent processing. become.

  Returning to FIG. 1, the heading candidate search unit 22 sets the searched heading candidate and the character string including the search keyword for each form data as a source of extraction, and outputs the pair to the heading candidate list creation unit 23.

  The heading candidate list creation unit 23 outputs the heading candidates of the character string including the search keyword as list information. Specifically, when the heading candidate list creation unit 23 obtains a heading candidate grouped for each form data from the heading candidate search unit 22 and a character string including a search keyword, the form identifier of the form data, the heading candidate, A heading candidate list in which character strings including the search keywords are included in one list is created.

  Further, the heading candidate list creation unit 23 edits the heading candidate list into heading candidate information provided with a check column for each heading candidate so that the user can select a heading that matches the search request. Further, the heading candidate list creation unit 23 outputs the edited heading candidate information to the output unit 40 and outputs the heading candidate list that is the editing source of the heading candidate information to the form search unit 24.

  Here, an example of the heading candidate information edited by the heading candidate list creation unit 23 will be described with reference to FIG. FIG. 4 is a diagram illustrating an example of heading candidate information. In the following description, it is assumed that the heading candidate information is related to a heading candidate searched from a form including “Kawasaki” in the search keyword.

  In the example of FIG. 4, the search keyword “Kawasaki” is included in the form A, the form B, and the form C. As a candidate for the search keyword “Kawasaki”, “address” is searched for in form A, “name” is searched for in form B, and “address” and “Heisei” are searched in form C.

  Under such circumstances, the headline candidate list creating unit 23 adds a headline candidate “address” and a character string “Kawasaki City, Kanagawa,... In addition, the heading candidate list creation unit 23 adds the heading candidate “name” and the character string “Jiro Kawasaki” including the search keyword to the heading candidate list for the form B. Further, the heading candidate list creating unit 23 adds the heading candidates “address” and “Heisei” and the character string “Nakahara-ku, Kawasaki-shi ...” including the search keyword to the heading candidate list for the form C.

  Then, the heading candidate list creation unit 23 edits the heading candidate list, and creates heading candidate information 23a in which a check column is provided for each heading candidate. The heading candidate information 23a includes the number of forms having the search keyword “Kawasaki” (three), and the number of forms in which the search keyword “Kawasaki” has been searched for each heading candidate. The heading candidate information 23a may include a heading candidate of “heading unknown” when a heading candidate of a character string including the search keyword “Kawasaki” is not searched.

  Thereby, when the heading candidate information 23a is output by the output unit 40 described later, the heading candidate corresponding to the search keyword “Kawasaki” can be confirmed by the user.

  A modification of the heading candidate information will be described with reference to FIG. FIG. 5 is a diagram illustrating a modification of the heading candidate information. As shown in FIG. 5, the heading candidate information 23b includes a headline candidate and a reduced image (thumbnail) of the form data in which the heading candidate exists.

  As a result, even if the user forgets whether “the form includes Kawasaki in the address” or “the form includes Kawasaki in the name”, the heading candidate information 23b is output by the output unit 40 described later. When this is done, it can be recalled which case it was, and headline candidates that match the search request can be quickly confirmed.

  The form search unit 24 searches for a form corresponding to the heading candidate. Specifically, when the form search unit 24 obtains a heading candidate from the heading selection unit 12, it extracts all form identifiers corresponding to the heading candidate from the heading candidate list output from the heading candidate list creation unit 23. To do. In addition, the form search unit 24 searches the form data information storage unit 31 for each form data corresponding to the extracted all form identifiers. Further, the form search unit 24 collects each searched form data and stores it in the form search result information storage unit 32 as a search result. As a result, the user can find a form that matches his search request.

  The output unit 40 displays and outputs the headline candidate information. Specifically, when acquiring the heading candidate information from the heading candidate list creation unit 23, the output unit 40 displays and outputs the acquired heading candidate information on the monitor. As a result, the user can check the heading candidate that matches the search request, and even if there is an incorrect heading candidate, by selecting the correct heading candidate, the form that matches the search request. Can be searched.

  Here, the result of the headline candidate being selected by the user from the headline candidate information displayed and output on the monitor will be described with reference to FIG. FIG. 6 is a diagram illustrating a result of selecting a headline candidate by the user. The heading candidate information is assumed to be a case where the search keyword is “Kawasaki”.

  In the example of FIG. 6, the heading candidate information indicates that there are three forms including the search keyword “Kawasaki”. These forms indicate that “address”, “name”, and “Heisei” have been searched for in addition to “heading unknown” as the heading candidates for the character string including the search keyword “Kawasaki”. Here, when the user's search request is “search for a form including Kawasaki in the address”, the user selects a headline candidate that matches the search request by checking the headline candidate “address”.

  As a result, even if a plurality of headline candidates are searched for the search keyword, the user restricts headline candidates that match the search request of the user and performs “item limited search” regarding the form that matches the search request. "It can be performed.

  By the way, when the heading candidate search unit 22 searches for a heading candidate of a character string including a search keyword from a plurality of forms, a character string inappropriate as a heading may be erroneously searched as a heading candidate. Even in such a case, if it is recognized that the character string erroneously searched as a headline candidate is included in the form that matches the user's search request, the character string is regarded as a headline candidate. You may consider it.

  For example, FIG. 7 is a diagram illustrating an example when an error occurs in the search for a heading candidate. In the example of FIG. 7, it is assumed that the search keyword is “Kawasaki”. As shown in FIG. 7, in the form A, the heading candidate search unit 22 erroneously searches for “Kanagawa Prefecture” as a heading candidate for the search keyword “Kawasaki”. In this case, the headline candidate information includes “Kanagawa Prefecture”, which is originally inappropriate as a headline, as a headline candidate.

  However, in the case where the user recognizes that “Kanagawa Prefecture” is a character string that originally belongs to an “address” appropriate as a headline, “Kanagawa Prefecture” may be regarded as a headline candidate. As a result, the user can search for a form that matches the user's own search request by selecting the headline candidate “Kanagawa”.

  Next, a processing procedure of the form retrieval apparatus 1 according to the first embodiment will be described with reference to FIG. FIG. 8 is a flowchart illustrating the processing procedure of the form search apparatus according to the first embodiment.

  First, when the keyword input unit 11 detects a search request including a search keyword input by the user (step S11), the keyword input unit 11 outputs the detected search keyword to the keyword extraction unit 21.

  Next, the keyword extraction unit 21 reads all the form data from the form data information storage unit 31, and selects one form data from all the read form data (step S12).

  Then, the keyword extracting unit 21 extracts a character string including the search keyword from the character strings constituting the selected form data (step S13).

  Further, based on the drawing position of the character string extracted by the keyword extracting unit 21, the headline candidate searching unit 22 searches the form data for a headline candidate corresponding to the character string (step S14). For example, the heading candidate search unit 22 is a character string arranged in the forward direction of the X-axis, and among the character strings that are on the same column as the character string extracted by the keyword extraction unit 21, the extracted character string Other character strings having a drawing position in front of the drawing position are searched for as heading candidates. In addition, the heading candidate search unit 22 is a character string arranged in the forward direction of the Y-axis, and among the character strings that are on the same column as the character string extracted by the keyword extraction unit 21, the extracted character string Other character strings having a drawing position in front of the drawing position are searched for as heading candidates. The headline candidate search unit 22 does not limit the headline candidate search method to this.

  Thereafter, the heading candidate search unit 22 determines whether or not all the form data has been processed (step S15). If the headline candidate search unit 22 has not processed all the form data (No in step S15), the process proceeds to step S12 in order to process unprocessed form data.

  On the other hand, if the headline candidate search unit 22 has processed all the form data (step S15 Yes), the headline candidate list creation unit 23 creates list information of the headline candidates searched from all the form data, A display is output to the output unit 40 (step S16).

  Specifically, the heading candidate list creation unit 23 creates a heading candidate list in which form identifiers of the form data, heading candidates, and character strings including search keywords are listed, and a heading candidate having a check column for each heading candidate. The information is edited into information, and the edited heading candidate information is displayed and output on the output unit 40.

  Subsequently, the headline selection unit 12 detects the headline candidate selected from the headline candidate information (step S17), and outputs it to the form search unit 24.

  Then, the form retrieval unit 24 extracts the form identifier corresponding to the heading candidate acquired from the heading selection unit 12 from the heading candidate list created by the heading candidate list creation unit 23. Then, the form search unit 24 searches the form data information storage unit 31 for form data corresponding to the extracted form identifier, and stores the searched form data in the form search result information storage unit 32 as a search result. (Step S18).

  As described above, according to the first embodiment, the form retrieval apparatus 1 includes the form data information storage unit 31 that stores a plurality of character strings in the form data and the drawing positions of the character strings for each form. From a character string in each form data stored in the data information storage unit 31, a character string including a search keyword that forms a part of a condition for searching for a form and a drawing position of the character string are extracted. Then, the form search device 1 searches for headline candidates corresponding to the character string from the form data from which the character string is extracted based on the drawing position of the extracted character string. Then, the form retrieval apparatus 1 generates heading candidate information in which the searched heading candidate and a character string corresponding to the heading candidate are associated with each other for the form data, and displays and outputs the generated heading candidate information. Further, the form search device 1 searches the form data information storage unit 31 for form data including a character string related to a headline candidate selected by a predetermined operation from among the displayed headline candidate information.

  According to such a configuration, the form retrieval apparatus 1 makes a search request even if there is no item definition information indicating the relationship between the character string of the heading and the character string corresponding to the plurality of character strings in the form data. The matched form data can be searched, and there is an effect that it does not take time to generate item definition information in advance. In particular, in the form data, for example, there are generally a large number of headlines that are swayed in the notation such as “address”, “home”, etc., so there is no need to define these items in the item definition information in advance, and this is done in advance. Work to be reduced.

  In addition, even if the search for the headline candidate corresponding to the search keyword is wrong, the form search device 1 selects a headline candidate that matches the search request once, and selects a form that matches the search request. Compared with the case where the item definition information is generated in advance, not only can the work be remarkably reduced, but also the search accuracy of the form can be improved.

  Furthermore, the form retrieval apparatus 1 uses the item definition information to find out in advance whether or not to search for a headline character string as requested by the search request and whether or not to retrieve a form as requested by the search request. This eliminates the work that needs to be done in advance.

  By the way, in the first embodiment, the case where the form search apparatus 1 searches for a headline candidate corresponding to the search keyword when the headline candidate corresponding to the search keyword exists in the form has been described. The present invention is not limited to this, and the form search device 2 estimates a virtual headline candidate corresponding to the search keyword even if the headline candidate corresponding to the search keyword does not exist in the form. Also good.

  Therefore, in the second embodiment, a case will be described in which the form search device 2 estimates a virtual headline candidate corresponding to a search keyword even when the headline candidate corresponding to the search keyword does not exist in the form.

  First, the configuration of the form retrieval apparatus 2 according to the second embodiment will be described with reference to FIG. FIG. 9 is a functional block diagram illustrating the configuration of the form search apparatus according to the second embodiment. Note that the same components as those in the form retrieval apparatus 1 shown in FIG. 1 are denoted by the same reference numerals, and the description of the overlapping configuration and operation is omitted. The difference between the first embodiment and the second embodiment is that a heading candidate estimation unit 41 is added to the control unit 20.

  The heading candidate estimation unit 41 estimates a virtual heading candidate corresponding to the character string when there is no heading candidate for the character string including the search keyword extracted from the form data. Specifically, the headline candidate estimation unit 41 includes a search keyword acquired from the headline candidate search unit 22 using a relation table in which a relationship between a character string drawing position and a virtual headline candidate is defined in advance. A heading candidate corresponding to the drawing position of the character string is estimated.

  For example, if the drawing position of the character string including the search keyword is the center in the X-axis (horizontal) direction and the forefront in the Y-axis (vertical) direction, the headline candidate estimation unit 41 selects “form title”. Presumed to be a headline candidate. Further, the headline candidate estimation unit 41 selects “form issuer” when the drawing position of the character string including the search keyword is behind the X axis (horizontal) direction and forward of the Y axis (vertical) direction of the form. Presumed to be a headline candidate. Further, the headline candidate estimation unit 41 estimates “form serial number” as a headline candidate when the drawing position of the character string including the search keyword is the last position in the Y-axis (vertical) direction of the form.

  As another example, the headline candidate estimation unit 41 has the drawing position of the character string including the search keyword at the front 1/4 in the vertical direction (Y axis) of the form, and the font of the character string forms the form. If it is larger than the font of the other character string, the “form title” is estimated as a heading candidate.

  In addition, based on the drawing position of the character string including the search keyword, the heading candidate estimation unit 41 estimates “heading unknown” as a heading candidate when a virtual heading candidate of the character string cannot be estimated.

  Further, the heading candidate search unit 41 sets the estimated heading candidate and the character string including the search keyword for each form data from which the search keyword is extracted, and outputs the set to the heading candidate list creation unit 23.

  Next, the processing procedure of the form retrieval apparatus 2 according to the second embodiment will be described with reference to FIG. FIG. 10 is a flowchart illustrating the processing procedure of the form search apparatus 2 according to the second embodiment. In FIG. 10, the same parts as those in the processing procedure (FIG. 8) of the form search apparatus 1 according to the first embodiment are denoted by the same reference numerals and detailed description thereof is omitted.

  First, when the keyword input unit 11 detects a search request including a search keyword input by the user (step S11), the keyword input unit 11 outputs the detected search keyword to the keyword extraction unit 21.

  Next, the keyword extraction unit 21 reads all the form data from the form data information storage unit 31, and selects one form data from all the read form data (step S12).

  Then, the keyword extracting unit 21 extracts a character string including the search keyword from the character strings constituting the selected form data (step S13).

  Further, based on the drawing position of the character string extracted by the keyword extracting unit 21, the headline candidate searching unit 22 searches the form data for a headline candidate corresponding to the character string (step S14).

  Then, the headline candidate search unit 22 determines whether there is a headline candidate for the character string extracted by the keyword extraction unit 21 (step S21). If the headline candidate search unit 22 determines that there is a headline candidate for the character string extracted by the keyword extraction unit 21 (Yes in step S21), the process proceeds to step S15.

  On the other hand, when the headline candidate search unit 22 determines that there is no headline candidate for the character string extracted by the keyword extraction unit 21 (No in step S21), the headline candidate estimation unit 41 estimates the headline candidate for the character string. (Step S22).

  Specifically, the headline candidate estimation unit 41 uses the relation table in which the relationship between the character string drawing position and the virtual heading candidate is defined in advance, and the character string drawing position extracted by the keyword extraction unit 21. The heading candidate corresponding to is estimated.

  For example, when the drawing position of the character string extracted by the keyword extracting unit 21 is the center in the X-axis (horizontal) direction and the forefront in the Y-axis (vertical) direction of the form, The “form title” is estimated as a headline candidate. In addition, the headline candidate estimation unit 41 determines that “form” is used when the drawing position of the character string extracted by the keyword extraction unit 21 is behind the X axis (horizontal) direction and forward of the Y axis (vertical) direction. "Publisher" is estimated as a headline candidate. Further, the headline candidate estimation unit 41 estimates “form serial number” as a headline candidate when the drawing position of the character string extracted by the keyword extraction unit 21 is the last position in the Y-axis (vertical) direction of the form. . The headline candidate estimation unit 41 is not limited to the headline candidate estimation method.

  Thereafter, the heading candidate search unit 22 determines whether or not all the form data has been processed (step S15). If the headline candidate search unit 22 has not processed all the form data (No in step S15), the process proceeds to step S12 in order to process unprocessed form data.

  On the other hand, if the headline candidate search unit 22 has processed all the form data (step S15 Yes), the headline candidate list creation unit 23 creates list information of the headline candidates searched from all the form data, A display is output to the output unit 40 (step S16).

  Specifically, the heading candidate list creation unit 23 displays a heading candidate list in which character strings including form identifiers, heading candidates, and search keywords of the form data are listed, and heading candidate information including a check column for each heading candidate. The edited heading candidate information is displayed and output to the output unit 40.

  Subsequently, the headline selection unit 12 detects the headline candidate selected from the headline candidate information (step S17), and outputs it to the form search unit 24.

  Then, the form retrieval unit 24 extracts the form identifier corresponding to the heading candidate acquired from the heading selection unit 12 from the heading candidate list created by the heading candidate list creation unit 23. Then, the form search unit 24 searches the form data information storage unit 31 for form data corresponding to the extracted form identifier, and stores the searched form data in the form search result information storage unit 32 as a search result. (Step S18).

  As described above, according to the second embodiment, the form search device 2 includes the form data information storage unit 31 that stores a plurality of character strings in the form data and the drawing positions of the character strings for each form data. A character string including a search keyword that forms a part of a condition for searching for a form is extracted from character strings in each form data stored in the form data information storage unit 31. Then, the form search device 2 searches for a headline candidate corresponding to the character string from the form data from which the character string is extracted based on the drawing position of the extracted character string. At this time, when the headline candidate cannot be searched, the form search device 2 estimates a virtual headline candidate corresponding to the character string. Then, the form search device 2 generates heading candidate information that associates the searched or estimated heading candidate and the character string corresponding to the heading candidate for each form data, and displays and outputs the generated heading candidate information. Further, the form search device 1 searches the information data information 31 for form data including a character string associated with a heading candidate selected by a predetermined operation from the output heading candidate information.

  According to such a configuration, the form retrieval apparatus 2 estimates a virtual headline candidate even when there is no headline candidate corresponding to the search keyword in the form data. Thus, the form search process can be executed, and the form that matches the search request can be searched without omission.

  By the way, in the first embodiment described above, the form retrieval apparatus 1 searches the character string heading candidate including the search keyword and, as a result, does not correct the heading candidate even if there is a heading candidate that has failed in the search. explained. The present invention is not limited to this, and the form retrieval apparatus 3 corrects the headline candidate that has failed in the search as a correct headline candidate as a result of searching for the headline candidate of the character string including the search keyword. Also good.

  Therefore, in the third embodiment, a case will be described in which the form search device 3 corrects a headline candidate that has failed to be searched as a result of searching for a headline candidate of a character string including the search keyword.

  First, the configuration of the form retrieval apparatus 3 according to the third embodiment will be described with reference to FIG. FIG. 11 is a functional block diagram illustrating the configuration of the form search apparatus according to the third embodiment. Note that the same components as those in the form retrieval apparatus 1 shown in FIG. 1 are denoted by the same reference numerals, and the description of the overlapping configuration and operation is omitted. The difference between the first embodiment and the third embodiment is that a headline correction unit 51 is added to the input unit 10, a headline correction information creation unit 52 is added to the control unit 20, and a headline correction information storage unit 53 is added to the storage unit 30.

  The headline correction unit 51 corrects a headline candidate that has failed to be searched out of the headline candidate information displayed and output by the output unit 40, to a plurality of drawing positions related to the headline candidate and a character string corresponding to the headline candidate. A plurality of related drawing positions are detected and output to the headline correction information creation unit 52.

  Specifically, the headline correction unit 51 detects a headline candidate selected by the user and failed in the search from the headline candidate information displayed and output by the output unit 40. Then, the headline correction unit 51 outputs the headline candidate to the headline correction information creation unit 52 in order to cause the output unit 40 to display and output a form related to the detected headline candidate. Further, the headline correction unit 51 detects the rectangular information in the data area of each of the correct headline candidate and the character string corresponding to the headline candidate selected by the user from the form displayed and output by the output unit 40 to correct the headline. The data is output to the information creation unit 52. The rectangle information refers to the drawing position of the vertex of the rectangle when the data area is a rectangle.

  Here, a case where the headline is corrected by the user from the headline candidate information 23d output to the output unit 40 will be described with reference to FIG. FIG. 12 is a diagram illustrating a case where the headline is corrected by the user. The heading candidate information 23d is assumed to be a case where the search keyword is “Kawasaki”.

  In the example of FIG. 12, in the form A, “Kanagawa Prefecture”, which is originally inappropriate as a heading candidate, is erroneously searched as a heading candidate for the search keyword “Kawasaki”. In this case, the user selects (for example, clicks) “Kanagawa prefecture 1 case” (d1) including the headline candidate searched by mistake.

  Then, the form A including the selected headline candidate and the corresponding character string is displayed and output on the monitor. Then, the user selects each of the character string “Nakahara-ku, Kawasaki-shi, Kanagawa Prefecture” (d3) and the corresponding correct heading candidate “address” (d2) arranged on the form A displayed and output on the monitor. Specify the data area. For example, the user designates the upper left and lower right vertices of a rectangle including each character string by clicking the mouse.

  Returning to FIG. 11, the heading candidate list creation unit 23 displays and outputs heading candidate information obtained by editing the heading candidate list so that the user can select a heading that matches the search request, and outputs the heading candidate list. The table is output to the headline correction information creation unit 52 and the form search unit 24.

  The headline correction information creation unit 52 displays and outputs on the output unit 40 a form corresponding to the headline candidate for which the search has failed. Specifically, when the headline correction information creation unit 52 acquires a headline candidate that has failed to be searched by the headline correction unit 51, the headline correction information generation unit 52 extracts and extracts the form identifier of the form data corresponding to the heading candidate from the headline candidate list. The form data information storage unit 31 is searched for form data corresponding to the form identifier. Further, the headline correction information creation unit 52 displays and outputs a form developed based on each drawing data included in the retrieved form data on the output unit 40.

  In addition, the headline correction information creation unit 52 stores the correct headline candidate and the rectangular information of each data area of the corresponding character string in the headline correction information storage unit 53 as form correction data.

  Here, the structure of the correction data stored in the headline correction information storage unit 53 will be described with reference to FIG. FIG. 13 is a diagram illustrating an example of the data structure of the headline correction information storage unit 53. As shown in FIG. 13, the headline correction information storage unit 53 stores headline correction data for each form.

  The headline correction data includes a form identifier 53a, headline character string rectangle information 53b, and character string rectangle information 53c corresponding to the headline. The form identifier 53a is an identifier of the form including the corrected heading candidate. The rectangular information 53b of the heading character string is rectangular information of the corrected heading candidate data area, and indicates the drawing positions of the upper left and lower right rectangular vertices of the heading candidate character string arranged on the form. The rectangular information 53c of the character string corresponding to the headline is rectangular information of the data area of the character string corresponding to the corrected heading candidate, and the upper left and lower right rectangular vertices of the character string arranged on the form are drawn. Indicates the position.

  Although the headline information creating unit 52 has been described as storing the correction data of the form in the headline correction information storage unit 53, it may be additionally stored in the form data information storage unit 31.

  The heading candidate search unit 22 searches for a heading candidate for a character string including a search keyword extracted from the form data. At this time, the heading candidate search unit 22 uses the corrected heading candidate as a search result when the heading candidate of the character string including the search keyword has been corrected in the past.

  Specifically, when the headline candidate search unit 22 obtains the drawing position of the character string extracted by the keyword extraction unit 21 and the form data from which the character string is extracted, the headline candidate search unit 22 refers to the headline correction information storage unit 53, It is determined whether there is heading correction data of the form data.

  Further, when it is determined that there is heading correction data of the form data, the heading candidate search unit 22 determines that the drawing position of the character string extracted by the keyword extraction unit 21 corresponds to “heading” of the heading correction data. It is determined whether or not the character string rectangle information 53c exists.

  In addition, the headline candidate search unit 22, when the drawing position of the character string extracted by the keyword extraction unit 21 exists within the “rectangular information of the character string corresponding to the headline” 53 c of the headline correction data, A headline candidate is searched from the “rectangular information of the character string corresponding to the heading” 53c corresponding to the “rectangular information of the character string corresponding to the heading” 53c. For example, the heading candidate search unit 22 searches the drawing data of the character string existing in the “rectangular information of the heading character string” 53b included in the heading correction information storage unit 53, and searches for the drawing position. The selected character string is set as a heading candidate.

  Next, a processing procedure of the form retrieval apparatus 3 according to the third embodiment will be described with reference to FIG. FIG. 14 is a flowchart illustrating the processing procedure of the form search apparatus according to the third embodiment. In FIG. 14, the same parts as those in the processing procedure (FIG. 8) of the form retrieval apparatus 1 according to the first embodiment are denoted by the same reference numerals, and detailed description thereof is omitted.

  First, the keyword input unit 11 detects a search request including a search keyword input by the user (step S11), and outputs the detected search keyword to the keyword extraction unit 21.

  Next, the keyword extraction unit 21 reads all the form data from the form data information storage unit 31, and selects form data corresponding to one form identifier from all the read form data (step S12).

  Then, the keyword extracting unit 21 extracts a character string including the search keyword from the character strings constituting the selected form data (step S13).

  Furthermore, the headline candidate search unit 22 uses the form identifier of the form data selected by the keyword extraction unit 21 as the headline correction information storage unit in order to check whether or not the headline candidate of the character string including the search keyword has been corrected in the past. It is determined whether or not it is stored in 53 (step S31).

  When the form identifier of the selected form data is stored in the headline correction information storage unit 53 (step S31 Yes), the headline candidate search unit 22 indicates that the drawing position of the character string including the extracted search keyword is the headline. It is determined whether or not it exists inside the rectangular information 53c of the character string corresponding to (step S32).

  When the drawing position of the character string including the extracted search keyword exists inside the rectangle information 53c of the character string corresponding to the headline (Yes in step S32), the headline candidate search unit 22 includes the headline correction information storage unit 53. A headline candidate is searched for from the rectangular information 53b of the headline character string included in (step S33).

  On the other hand, when the selected form identifier is not stored in the headline correction information storage unit 53 (No in step S31), or the drawing position of the character string including the extracted search keyword is the character string rectangle information 53c corresponding to the headline. If it does not exist inside (No in step S32), the headline candidate search unit 22 searches the form data for a headline candidate corresponding to the character string based on the drawing position of the character string extracted by the keyword extraction unit 21. (Step S14).

  Thereafter, the heading candidate search unit 22 determines whether or not all the form data has been processed (step S15). If the headline candidate search unit 22 has not processed all the form data (No in step S15), the process proceeds to step S12 in order to process unprocessed form data.

  On the other hand, if the headline candidate search unit 22 has processed all the form data (step S15 Yes), the headline candidate list creation unit 23 creates list information of the headline candidates searched from all the form data, A display is output to the output unit 40 (step S16).

  When the headline correction information creation unit 52 acquires from the headline correction unit 51 a headline candidate that has failed to be searched in the headline candidate list information, the headline correction information creation unit 52 outputs a form developed based on the form data corresponding to the headline candidate. 40 is displayed and output (step S34).

  Then, the headline correction information creation unit 52 detects the rectangular information of the corrected heading candidate and the data area of each of the character strings corresponding thereto from the headline correction unit 51, so that the rectangular information is headed as the headline correction data of the form. The correction information is stored in the correction information storage unit 53 (step S35).

  Subsequently, the headline selection unit 12 detects the headline candidate selected from the headline candidate information (step S17), and outputs it to the form search unit 24.

  Then, the form retrieval unit 24 extracts the form identifier corresponding to the heading candidate acquired from the heading selection unit 12 from the heading candidate list created by the heading candidate list creation unit 23. Then, the form search unit 24 searches the form data information storage unit 31 for form data corresponding to the extracted form identifier, and stores the searched form data in the form search result information storage unit 32 as a search result. (Step S18).

  As described above, according to the third embodiment, the form retrieval apparatus 3 includes the form data information storage unit 31 that stores a plurality of character strings in the form data and the drawing positions of the character strings for each form data. A character string including a search keyword that forms a part of a condition for searching for a form is extracted from character strings in each form data stored in the form data information storage unit 31. Then, the form search device 3 searches for a headline candidate corresponding to the character string from the form data from which the character string is extracted based on the drawing position of the extracted character string. Then, the form search device 3 generates heading candidate information in which the searched heading candidate is associated with the character string corresponding to the heading candidate, and outputs the generated heading candidate information. Furthermore, in order to correct the headline candidate having an error in the output heading candidate information, the form search device 3 uses a plurality of drawing positions related to the character string of the heading candidate and the character string corresponding to the heading candidate. The user is made to select a plurality of drawing positions involved and is stored in the headline correction information storage unit 53. Then, the form search device 3 searches the form data information storage unit 31 for form data including a character string associated with a heading candidate selected by a predetermined operation from the output heading candidate information.

  According to such a configuration, the form retrieval apparatus 3 stores the correct headline candidate in the headline correction information storage unit 53 even when the headline candidate information includes an incorrect headline candidate. In response to a search request for the same search keyword from the next time, a correct heading candidate can be searched with high accuracy.

  By the way, in the first embodiment described above, the form retrieval apparatus 1 has been described as a result of searching for a heading candidate for a character string including a search keyword, and as a result, does not evaluate the validity of the searched heading candidate. The present invention is not limited to this, and the form search device 4 may evaluate the validity of the search for the searched headline candidate as a result of searching for the headline candidate of the character string including the search keyword.

  Thus, in the fourth embodiment, a case will be described in which the form search device 4 evaluates the validity of searching for a searched heading candidate as a result of searching for a heading candidate for a character string including a search keyword.

  First, the configuration of the form retrieval apparatus 4 according to the fourth embodiment will be described with reference to FIG. FIG. 15 is a functional block diagram illustrating the configuration of the form search apparatus according to the fourth embodiment. Note that the same components as those in the form retrieval apparatus 1 shown in FIG. 1 are denoted by the same reference numerals, and the description of the overlapping configuration and operation is omitted. The difference between the first embodiment and the fourth embodiment is that a headline candidate evaluation unit 61 is added to the control unit 20 and a regular expression correspondence information storage unit 62 is added to the storage unit 30.

  The headline candidate evaluation unit 61 evaluates the validity of the search for the headline candidate searched by the headline candidate search unit 22. Note that the validity of the search is the certainty that there was no error in the search.

  Specifically, as a method for evaluating the validity of a search for a headline candidate, the headline candidate evaluation unit 61 acquires a headline candidate grouped for each form data from the headline candidate search unit 22 and a character string including a search keyword. Then, it is determined whether or not the same character string as the obtained character string of the heading candidate exists at a position other than the drawing position of the heading candidate.

  When the headline candidate evaluation unit 61 determines that the same character string as the acquired character string of the heading candidate exists at a position other than the drawing position of the heading candidate, the true heading candidate is different from the acquired heading candidate. Since there is a possibility of a character string existing at the drawing position, it is assumed that the legitimacy of the search for the acquired heading candidate is low. On the other hand, if the headline candidate evaluation unit 61 determines that there is no character string that is the same as the acquired character string of the heading candidate other than the drawing position of the heading candidate, the legitimacy of the search for the acquired heading candidate is high. Shall.

  As another method for evaluating the likelihood of searching, the headline candidate evaluation unit 61 obtains a headline candidate grouped for each form data and a character string including the search keyword from the headline candidate search unit 22. It is determined whether or not a character string including the character string is represented by a regular expression corresponding to the heading candidate paired therewith. For example, the headline candidate evaluation unit 61 uses the regular expression correspondence information storage unit 62 that stores in advance a regular expression of data corresponding to the headline character string, and the character string including the search keyword is paired with the headline candidate. It is determined whether or not the corresponding regular expression is satisfied.

  Here, the data structure stored in the regular expression correspondence information storage unit 62 will be described with reference to FIG. FIG. 16 is a diagram illustrating an example of a data structure of the regular expression correspondence information storage unit 62. As shown in FIG. 16, the regular expression correspondence information storage unit 62 includes a heading character string 62a and a regular expression 62b of the corresponding data, and is stored in pairs for each heading character string 62a.

  For example, when the heading character string 62a is “date”, the regular expression of the data corresponding to “date” is “1999 [1-12] month [1-31] day”. Here, “9” in the regular expression represents an arbitrary single digit, [1-12] represents a number from 1 to 12, and [1-31] represents an arbitrary number from 1 to 31.

  When the heading character string 62a is “order number”, the regular expression of the data corresponding to “order number” is “ZZZ99999-999-Z”. Here, “Z” in the regular expression represents an arbitrary one-digit alphabet, and “9” represents an arbitrary one-digit number.

  When the heading character string 62a is “telephone number”, the regular expression of the data corresponding to “telephone number” is “0999999999”. Here, “9” in the regular expression represents an arbitrary single-digit number.

  Furthermore, when the heading character string 62a is “classification”, the regular expression of the data corresponding to “classification” is “normal”, “current”, “general”, “periodic”, or “funded”.

  Returning to FIG. 15, if the headline candidate evaluation unit 61 determines that the character string including the keyword does not satisfy the regular expression corresponding to the headline candidate paired therewith, the headline candidate evaluation unit 61 may be a true headline candidate. It is determined that the search for headline candidates is low. On the other hand, if the headline candidate evaluation unit 61 determines that the character string including the keyword satisfies the regular expression corresponding to the headline candidate paired with the keyword, the headline candidate evaluation unit 61 determines that the possibility of being a true headline candidate is high. Therefore, it is assumed that the legitimacy of searching for a headline candidate is high.

  In addition, the headline candidate evaluation unit 61 combines a headline candidate including a search legitimacy evaluation result and a character string including a search keyword for each form data as a source of extraction, and generates a headline candidate list creation unit. To 23. Note that the headline candidate evaluation unit 61 may exclude a headline candidate having a low search legitimacy and output a group related to the headline candidate to the headline candidate list creation unit 23.

  As a result, the heading candidate list creation unit 23 outputs the heading candidate information including the evaluation result of the search validity to the output unit 40.

  Here, an example of the heading candidate information 23e edited by the heading candidate list creation unit 23 will be described with reference to FIG. FIG. 17 is a diagram illustrating an example of heading candidate information. As shown in FIG. 17, the heading candidate information 23e outputs a warning message e1 informing that there is a heading candidate “address” that has a low legitimacy of search.

  In the example of FIG. 17, the heading candidate information 23e outputs a warning message informing that there is a heading candidate with a low search validity. However, the form candidate including a heading candidate with a low search validity is output. The form identifier may be output at the same time.

  Thereby, the heading candidate list creation unit 23 can notify the user of heading candidates whose search validity is low, and can prompt confirmation of a form including the heading candidate.

  Next, the processing procedure of the form retrieval apparatus 4 according to the fourth embodiment will be described with reference to FIG. FIG. 18 is a flowchart illustrating the processing procedure of the form search apparatus according to the fourth embodiment. In FIG. 18, the same parts as those in the processing procedure (FIG. 8) of the form retrieval apparatus 1 according to the first embodiment are denoted by the same reference numerals, and detailed description thereof is omitted.

  First, when the keyword input unit 11 detects a search request including a search keyword input by the user (step S11), the keyword input unit 11 outputs the detected search keyword to the keyword extraction unit 21.

  Next, the keyword extraction unit 21 reads all the form data from the form data information storage unit 31, and selects one form data from all the read form data (step S12).

  Then, the keyword extracting unit 21 extracts a character string including the search keyword from the character strings constituting the selected form data (step S13).

  Further, based on the drawing position of the character string extracted by the keyword extracting unit 21, the headline candidate searching unit 22 searches the form data for a headline candidate corresponding to the character string (step S14).

  Then, the headline candidate evaluation unit 61 evaluates whether or not the searched headline candidate has failed in the search, that is, the validity of the search for the headline candidate (step S41). For example, the heading candidate evaluation unit 61 determines whether the same character string as the searched character string of the heading candidate exists at a position other than the drawing position of the heading candidate in the same form data.

  When there is a possibility that the searched headline candidate has failed in the search (Yes in step S42), the headline candidate evaluation unit 61 determines that the headline candidate is a headline candidate having a low search validity, and is visually confirmed. It is determined that the form data is necessary. Then, the headline candidate evaluation unit 61 sets, for example, the form data, a headline candidate including an evaluation result indicating that the search validity is low, and a character string including a search keyword corresponding to the headline candidate in the storage unit 30. Store (step S43).

  On the other hand, when there is no possibility that the searched headline candidate has failed in the search (No in step S42), the headline candidate evaluation unit 61 determines that the headline candidate is a headline candidate with high search validity, Judge as form data that does not require visual confirmation. Then, the headline candidate evaluation unit 61 sets, for example, the form data, a headline candidate including an evaluation result indicating that the search validity is high, and a character string including a search keyword corresponding to the headline candidate in the storage unit 30. Store.

  Thereafter, the heading candidate search unit 22 determines whether or not all the form data has been processed (step S15). If the headline candidate search unit 22 has not processed all the form data (No in step S15), the process proceeds to step S12 in order to process unprocessed form data.

  On the other hand, if the headline candidate search unit 22 has processed all the form data (Yes in step S15), the headline candidate list creation unit 23 includes the headline including the probability evaluation results searched from all the form data. Candidate list information is created and displayed on the output unit 40 (step S16).

  Subsequently, the headline selection unit 12 detects the headline candidate selected from the headline candidate information (step S17), and outputs it to the form search unit 24.

  Then, the form retrieval unit 24 extracts the form identifier corresponding to the heading candidate acquired from the heading selection unit 12 from the heading candidate list created by the heading candidate list creation unit 23. Then, the form search unit 24 searches the form data information storage unit 31 for form data corresponding to the extracted form identifier, and stores the searched form data in the form search result information storage unit 32 as a search result. (Step S18).

  As described above, according to the fourth embodiment, the form retrieval apparatus 4 includes the form data information storage unit 31 that stores a plurality of character strings and drawing positions in the form data for each form data. A character string including a search keyword that forms a part of the condition for searching for a form is extracted from the character strings in each form data stored in the unit 31. Then, the form search device 3 searches for a headline candidate corresponding to the character string from the form data from which the character string is extracted based on the drawing position of the extracted character string. Then, the form retrieval device 4 evaluates the validity of the search for the searched headline candidate, and if it is evaluated that the validity of the search for the headline candidate is low, the headline candidate information corresponding to the headline candidate Display and output a warning message. Further, the form search device 4 searches the information data information 31 for form data including a character string related to a heading candidate selected by a predetermined operation from the displayed heading candidate information.

  According to such a configuration, after searching for a heading candidate corresponding to the search keyword, the form search device 4 can notify the user of a heading candidate with low validity of the search, and confirm the form including the heading candidate. Can be urged.

  By the way, in the first embodiment, the form retrieval apparatus 1 uses the form data information storage unit 31 in which drawing data including character strings and drawing positions of data constituting the form is stored in advance. Explained the case. The present invention is not limited to this, and the form retrieval apparatus 5 may use a form data information storage unit 31 that stores drawing data created from image data obtained by scanning a paper form with a document scanner.

  Thus, in the fifth embodiment, a case will be described in which the form retrieval apparatus 5 uses a form data information storage unit 31 that stores drawing data created from image data obtained by scanning a paper form with a document scanner.

  First, the configuration of the form retrieval apparatus 5 according to the fifth embodiment will be described with reference to FIG. FIG. 19 is a functional block diagram illustrating the configuration of the form search apparatus according to the fifth embodiment. Note that the same components as those in the form retrieval apparatus 1 shown in FIG. 1 are denoted by the same reference numerals, and the description of the overlapping configuration and operation is omitted. The difference between the first embodiment and the fifth embodiment is that a form data information creation unit 71 is added to the control unit 20 and a form image information storage unit 72 is added to the storage unit 30.

  The form data information creation unit 71 extracts a plurality of drawing data from the image data of each form included in the form image information storage unit 72 using, for example, OCR technology, and stores each extracted drawing data in the form data information storage. The data is converted into the format of the section 31 and form data is created. Here, the drawing data indicates, for example, the character code and drawing position of the character string, and the drawing position of the start point and end point of the ruled line.

  In addition, the form data information creation unit 71 stores the created form data in the form data information storage unit 31.

  The form image information storage unit 72 stores image data obtained by scanning a paper form with a document scanner for each form. For example, the image data is a form represented by JPEG (Joint Photographic Experts Group), bitmap, or PDF (Portable Document Format).

  As described above, according to the fifth embodiment, the form retrieval apparatus 5 includes the form data information storage unit 31 that stores a plurality of character strings in the form and the drawing positions of the character strings for each form, and uses the OCR technique. Using the image data stored in the form image information storage unit 72 to create drawing information including a plurality of character strings and drawing positions of the character strings, the created drawing information is stored in the form data information storage unit 31 for each form. To store. Then, the form retrieval apparatus 5 extracts a character string including a search keyword that forms a part of the condition for retrieving the form from the character strings in each form data stored in the information data information 31. Then, the form search device 5 searches for the heading candidate corresponding to the character string from the form data from which the character string is extracted based on the drawing position of the extracted character string. The form retrieval apparatus 5 generates heading candidate information in which the searched heading candidate is associated with the character string corresponding to the heading candidate, and displays and outputs the generated heading candidate information. Further, the form search device 5 searches the information data information 31 for form data including a character string related to a heading candidate selected by a predetermined operation from the displayed heading candidate information.

  According to such a configuration, the form search device 5 stores the drawing information created from the image data in the form data information storage unit 31 even if the form file format is image data such as PDF. A form search process can be executed in the same manner as when the form file format is text, and a form that matches the search request can be searched without omission.

  Further, the form retrieval device 5 does not have item definition information indicating the relationship between the character string of the heading and the corresponding character string for a plurality of character strings in the form data, and the form data that matches the search request. This is advantageous in that it does not take time to generate item definition information in advance. In particular, in the form data, for example, there are generally a large number of headlines that are swayed in the notation such as “address”, “home”, etc., so there is no need to define these items in the item definition information in advance, and this is done in advance. Work to be reduced.

  Moreover, even if the search for the headline candidate corresponding to the search keyword is erroneous, the form search device 5 selects a headline candidate that matches the search request once, and selects a form that matches the search request. Compared with the case where the item definition information is generated in advance, not only can the work be remarkably reduced, but also the search accuracy of the form can be improved.

  Furthermore, the form retrieval apparatus 5 uses the item definition information to find out in advance whether or not to search for a headline character string as requested by the search and whether or not to retrieve a form as requested by the search. This eliminates the work that needs to be done in advance.

  In the above embodiment, the headline candidate search unit 22 searches for a headline candidate based on the drawing position of the character string including the search keyword. However, the headline character string list showing character string groups that can be headline candidates. May be stored in the storage unit 30 in advance. Examples of character strings that can be headline candidates include “address”, “current address”, “home”, “name”, “name”, and the like. In this case, the heading candidate search unit 22 sets a character string included in the heading character string list among the character strings existing around the drawing position of the character string including the search keyword as a heading candidate.

  In the above embodiment, the headline candidate search unit 22 searches for headline candidates based on the drawing position of the character string including the search keyword. However, the headline candidate search unit 22 includes a dictionary such as a personal name dictionary, an address dictionary, and a product name dictionary. The character strings of the heading candidates corresponding to these dictionaries may be stored in the storage unit 30 in advance. In this case, when the search keyword includes a word registered in those dictionaries, the heading candidate search unit 22 searches for the heading character string corresponding to the word from the vicinity of the drawing position of the character string including the search keyword. Then, the searched character string is used as a heading candidate.

  As mentioned above, although the Example of this invention was described, the range of the technical idea of this invention is not limited by this Example, and unless it deviates from the range of the technical scope described in the claim, it is various. It goes without saying that various embodiments can be implemented. Moreover, the effect described in the present Example is not limited to this.

  Each component of the illustrated form search apparatus 1 is functionally described, and is not necessarily physically configured as illustrated. The specific form of the form search apparatus 1 is as follows. Needless to say, the present invention is not limited to the illustrated one.

  It should be noted that all or some of the processing functions performed in the form retrieval apparatuses 1 to 5 are micro-processors such as a CPU (Central Processing Unit) (or MPU (Micro Processing Unit), MCU (Micro Controller Unit), etc. -It is implement | achieved by the program analyzed and executed by the said computer (or microcomputers, such as MPU and MCU), or may be implement | achieved as hardware by a wired logic.

  The processing functions performed by the form search apparatuses 1 to 5 are all or any part of the processing functions such as CPU (Central Processing Unit) (or MPU (Micro Processing Unit), MCU (Micro Controller Unit), etc. The computer may be recorded on a computer-readable recording medium so as to be analyzed and executed by the CPU (or a microcomputer such as an MPU or MCU). Needless to say, similar effects can be obtained.

  The following additional remarks are disclosed regarding the embodiment according to the above example.

(Supplementary note 1) Storage means for storing a plurality of character strings in a form and the drawing positions of the character strings for each form,
Extraction means for extracting a character string including a search keyword that forms part of a condition for searching for a form from among character strings in each form stored in the storage means, and a drawing position of the character string;
Search means for searching for a heading candidate corresponding to the character string from the form from which the character string is extracted based on the drawing position of the character string extracted by the extracting means;
Output means for generating heading candidate information for associating the heading candidate searched by the search means and the character string corresponding to the heading candidate for each form, and outputting the generated heading candidate information;
A form search means for searching for a form including a character string related to a headline candidate selected by a predetermined operation from the heading candidate information output by the output means, from the storage means;
A form retrieval apparatus characterized by comprising:

(Additional remark 2) The said search means is a character string arranged in the predetermined forward direction, Comprising: From the character string on the same line as the character string extracted by the said extraction means, from the drawing position of the said extracted character string The form search device according to appendix 1, wherein another character string arranged in front is searched for as a heading candidate.

(Supplementary note 3) The search means includes
When the drawing position of the form is represented by two-dimensional rows and columns, the character at the foremost drawing position in the row direction or the foremost drawing position in the column direction from the drawing position of the character string extracted by the extraction means The form search device according to appendix 1, wherein a column is searched as a heading candidate.

(Appendix 4) The predetermined operation is:
4. The form search apparatus according to any one of appendix 1 to appendix 3, wherein the form search operation is a selection operation for selecting a headline candidate from the heading candidate information output by the output unit.

(Additional remark 5) It has the headline candidate evaluation means which evaluates the correctness of the search in the headline candidate searched by the said search means,
The output means includes
As a result of the evaluation by the heading candidate evaluation means, if the heading candidate searched by the searching means is evaluated as having low validity, the heading candidate information corresponding to the heading candidate includes a warning message and is output. The form search device according to Supplementary Note 1, wherein:

(Appendix 6) The headline candidate evaluation means
It is determined whether the same character string as the heading candidate exists in a form including the heading candidate other than the drawing position of the heading candidate, and the same character string as the heading candidate is 6. The form search device according to appendix 5, wherein when it is determined that the image exists at a position other than the drawing position, the validity of the search for the heading candidate is evaluated to be low.

(Appendix 7) The headline candidate evaluation means
It is determined whether the character string including the search keyword is a predetermined regular expression corresponding to the heading candidate, and the character string including the search keyword is not a predetermined regular expression corresponding to the heading candidate. The form search device according to appendix 5, wherein, when determined, the validity of the search for the heading candidate is evaluated to be low.

(Supplementary note 8) In order to correct a heading candidate having an error in the heading candidate information output by the output means, a plurality of drawing positions related to the character string of the heading candidate and a character string corresponding to the heading candidate Correction selection means for selecting a plurality of drawing positions related to
Correction information storage means for storing a plurality of drawing positions related to a character string of a heading candidate selected by the correction selection means in association with a plurality of drawing positions related to a character string corresponding to the heading candidate. The form search device according to Supplementary Note 1, which is a feature.

(Supplementary note 9) The search means includes:
When the drawing position of the character string extracted by the extracting means is included in a closed region represented by a plurality of drawing positions related to the character string corresponding to the heading candidate stored in the correction storage means, The form search device according to appendix 8, wherein a headline candidate associated with the character string is searched.

(Supplementary Note 10) Drawing information generation that generates drawing information including a plurality of character strings and drawing positions of the character strings from the image data of the form using the OCR technique, and stores the generated drawing information in the storage unit The form retrieval device according to appendix 1, further comprising: means.

(Supplementary Note 11) The output means includes:
The form search device according to appendix 1, wherein the heading candidate information includes a reduced image of the form associated with each heading candidate information.

(Supplementary Note 12) A storage procedure for storing a plurality of character strings in a form and a drawing position of the character string in a storage medium for each form,
An extraction procedure for extracting a character string including a search keyword that forms part of a condition for searching for a form and a drawing position of the character string from character strings in each form stored in the storage procedure;
Based on the drawing position of the character string extracted by the extraction procedure, a search procedure for searching for a heading candidate corresponding to the character string from the form from which the character string has been extracted;
An output procedure for generating heading candidate information for associating the heading candidate searched by the search procedure and the character string corresponding to the heading candidate for each form, and outputting the generated heading candidate information;
A form search procedure for searching a form including a character string related to a headline candidate selected by a predetermined operation from the heading candidate information output by the output procedure from the storage medium stored by the storage procedure; ,
A form search program characterized by causing a computer to execute.

(Supplementary note 13) The search procedure is as follows.
Among the character strings arranged in a predetermined forward direction and on the same string as the character string extracted by the extraction unit, other characters arranged forward from the drawing position of the extracted character string The form search program according to appendix 12, wherein a column is searched as a heading candidate.

(Supplementary Note 14) The search procedure is as follows.
When the drawing position of the form is represented by two-dimensional rows and columns, the character at the foremost drawing position in the row direction or the foremost drawing position in the column direction from the drawing position of the character string extracted by the extraction procedure The form search program according to appendix 12, wherein a column is searched as a heading candidate.

(Supplementary Note 15) The predetermined operation is:
15. The form search program according to any one of Supplementary Note 12 to Supplementary Note 14, which is a selection operation for selecting a headline candidate from the heading candidate information output by the output procedure.

(Additional remark 16) The headline candidate evaluation procedure which evaluates the correctness of the search in the headline candidate searched by the said search procedure,
The output procedure is as follows:
As a result of the evaluation by the heading candidate evaluation means, when it is evaluated that the search legitimacy of the heading candidate searched by the search procedure is low, the heading candidate information corresponding to the heading candidate is output including a warning message. The form search program according to supplementary note 12, characterized by:

(Supplementary Note 17) The headline candidate evaluation procedure is as follows.
It is determined whether the same character string as the heading candidate exists in a form including the heading candidate other than the drawing position of the heading candidate, and the same character string as the heading candidate is The form search program according to supplementary note 16, wherein when it is determined that it exists at a position other than the drawing position, the validity of the search for the heading candidate is evaluated to be low.

(Supplementary Note 18) The headline candidate evaluation procedure is as follows:
It is determined whether the character string including the search keyword is a predetermined regular expression corresponding to the heading candidate, and the character string including the search keyword is not a predetermined regular expression corresponding to the heading candidate. The form search program according to supplementary note 16, wherein if judged, the validity of the search for the heading candidate is evaluated to be low.

(Supplementary note 19) In order to correct a heading candidate having an error in the heading candidate information output by the output procedure, a plurality of drawing positions related to the character string of the heading candidate and a character string corresponding to the heading candidate Correction selection procedure for selecting a plurality of drawing positions related to
A correction information storage procedure for storing a plurality of drawing positions related to a character string of a heading candidate selected by the correction selection procedure in association with a plurality of drawing positions related to a character string corresponding to the heading candidate in a storage medium; The form search program according to appendix 12, characterized in that it includes:

(Supplementary note 20) The search procedure is as follows.
The drawing position of the character string extracted by the extraction procedure is included in a closed region represented by a plurality of drawing positions related to the character string corresponding to the heading candidate stored in the storage medium stored by the correction storage procedure. If it is, the form search program according to appendix 19, wherein a headline candidate associated with the character string is searched.

(Appendix 21) Using OCR technology from image data of a form, drawing information including a plurality of character strings and drawing positions of the character strings is generated, and the generated drawing information is stored in the storage procedure. 13. The form search program according to appendix 12, including a drawing information generation procedure to be stored in a storage medium.

(Appendix 22) The output procedure is as follows:
13. The form search program according to appendix 12, wherein the heading candidate information includes a reduced image of the form associated with each heading candidate information.

(Supplementary note 23) A form retrieval method in which a form retrieval apparatus retrieves a form,
A storage step of storing a plurality of character strings in the form and a drawing position of the character string in a storage medium for each form;
An extraction step for extracting a character string including a search keyword that forms part of a condition for searching for a form and a drawing position of the character string from character strings in each form stored in the storage step;
Based on the drawing position of the character string extracted by the extraction step, a search step for searching for a heading candidate corresponding to the character string from the form from which the character string has been extracted;
Generating heading candidate information that associates the heading candidate searched by the searching step and the character string corresponding to the heading candidate for each form, and outputting the generated heading candidate information;
A form search step for searching a form including a character string related to a headline candidate selected by a predetermined operation from the heading candidate information output by the output step from the storage medium stored in the storage step; ,
A form search method characterized by including:

(Supplementary Note 24) The search step includes:
Among the character strings arranged in a predetermined forward direction and on the same string as the character string extracted by the extraction unit, other characters arranged forward from the drawing position of the extracted character string The form search method according to appendix 23, wherein the column is searched as a heading candidate.

(Supplementary Note 25) The search step includes:
When the drawing position of the form is represented by two-dimensional rows and columns, the character at the foremost drawing position in the row direction or the foremost drawing position in the column direction from the drawing position of the character string extracted by the extraction step The form search method according to appendix 23, wherein the column is searched as a heading candidate.

(Supplementary Note 26) The predetermined operation is:
26. The form search method according to any one of appendix 23 to appendix 25, wherein the form search operation is a selection operation for selecting a headline candidate from the headline candidate information output in the output step.

(Additional remark 27) The headline candidate evaluation step which evaluates the correctness of the search in the headline candidate searched by the said search step,
The output step includes
As a result of the evaluation of the heading candidate evaluation means, if it is evaluated that the search candidate in the heading candidate searched by the searching step is low, the heading candidate information corresponding to the heading candidate includes a warning message and is output. The form search method according to supplementary note 23, wherein

(Supplementary Note 28) The heading candidate evaluation step includes:
It is determined whether the same character string as the heading candidate exists in a form including the heading candidate other than the drawing position of the heading candidate, and the same character string as the heading candidate is 28. The form search method according to appendix 27, wherein, when it is determined that it exists at a position other than the drawing position, the validity of the search for the heading candidate is evaluated to be low.

(Supplementary Note 29) The heading candidate evaluation step includes:
It is determined whether the character string including the search keyword is a predetermined regular expression corresponding to the heading candidate, and the character string including the search keyword is not a predetermined regular expression corresponding to the heading candidate. 28. The form search method according to appendix 27, wherein if it is determined, the validity of the search for the heading candidate is evaluated to be low.

(Supplementary Note 30) In order to correct a heading candidate having an error in the heading candidate information output by the output step, a plurality of drawing positions related to the character string of the heading candidate and a character string corresponding to the heading candidate A correction selection step for selecting a plurality of drawing positions related to
A correction information storing step of storing a plurality of drawing positions related to the character string of the heading candidate selected in the correction selection step in association with a plurality of drawing positions related to the character string corresponding to the heading candidate. The form search method according to supplementary note 23, which is a feature.

(Supplementary Note 31) The search step includes:
The drawing position of the character string extracted by the extraction step is included in a closed region represented by a plurality of drawing positions related to the character string corresponding to the heading candidate stored in the storage medium stored in the correction storage step. If it is, the form search method according to Supplementary Note 30, wherein a heading candidate associated with the character string is searched.

(Supplementary Note 32) Using OCR technology from image data of a form, drawing information including a plurality of character strings and drawing positions of the character strings is generated, and the generated drawing information is stored in the storage step. 24. The form search method according to appendix 23, further comprising a drawing information generation step of storing in a storage medium.

(Supplementary Note 33) The output step includes:
The form search method according to appendix 23, wherein the heading candidate information includes a reduced image of the form associated with each heading candidate information.

DESCRIPTION OF SYMBOLS 1 Form search device 10 Input part 11 Keyword input part 12 Headline selection part 20 Control part 21 Keyword extraction part 22 Headline candidate search part 23 Headline candidate list creation part 24 Form search part 30 Storage part 31 Form data information storage part 32 Form search result Information storage unit 41 Heading candidate estimation unit 51 Heading correction unit 52 Heading correction information creation unit 53 Heading correction information storage unit 61 Heading candidate evaluation unit 62 Regular expression correspondence information storage unit 71 Form data information creation unit 72 Form image information storage unit

Claims (7)

Storage means for storing a plurality of character strings in a form and the drawing position of the character string for each form;
Extraction means for extracting a character string including a search keyword that forms part of a condition for searching for a form from among character strings in each form stored in the storage means, and a drawing position of the character string;
Search means for searching for a heading candidate corresponding to the character string from the form from which the character string is extracted based on the drawing position of the character string extracted by the extracting means;
Output means for generating heading candidate information for associating the heading candidate searched by the search means and the character string corresponding to the heading candidate for each form, and outputting the generated heading candidate information;
A form search means for searching for a form including a character string related to a headline candidate selected by a predetermined operation from the heading candidate information output by the output means, from the storage means;
A form retrieval apparatus characterized by comprising: The search means is a character string arranged in a predetermined forward direction and is arranged forward from a drawing position of the extracted character string among character strings on the same column as the character string extracted by the extraction means. The form search device according to claim 1, wherein the other character string is searched as a heading candidate. The search means includes
When the drawing position of the form is represented by two-dimensional rows and columns, the character at the foremost drawing position in the row direction or the foremost drawing position in the column direction from the drawing position of the character string extracted by the extraction means The form search device according to claim 1, wherein the column is searched as a heading candidate. The predetermined operation is:
The form search apparatus according to claim 1, wherein the form search device is a selection operation for selecting a headline candidate from the heading candidate information output by the output unit. Headline candidate evaluation means for evaluating the validity of the search in the headline candidate searched by the search means,
The output means includes
As a result of the evaluation by the heading candidate evaluation means, if the heading candidate searched by the searching means is evaluated as having low validity, the heading candidate information corresponding to the heading candidate includes a warning message and is output. The form retrieval apparatus according to claim 1, wherein: A storage procedure for storing a plurality of character strings in a form and a drawing position of the character string in a storage medium for each form;
An extraction procedure for extracting a character string including a search keyword that forms part of a condition for searching for a form and a drawing position of the character string from character strings in each form stored in the storage procedure;
Based on the drawing position of the character string extracted by the extraction procedure, a search procedure for searching for a heading candidate corresponding to the character string from the form from which the character string has been extracted;
An output procedure for generating heading candidate information for associating the heading candidate searched by the search procedure and the character string corresponding to the heading candidate for each form, and outputting the generated heading candidate information;
A form search procedure for searching a form including a character string related to a headline candidate selected by a predetermined operation from the heading candidate information output by the output procedure from the storage medium stored by the storage procedure; ,
A form search program characterized by causing a computer to execute. A form search method in which a form search device searches for a form,
The form search device
A storage step of storing a plurality of character strings in the form and a drawing position of the character string in a storage medium for each form;
An extraction step for extracting a character string including a search keyword that forms part of a condition for searching for a form and a drawing position of the character string from character strings in each form stored in the storage step;
Based on the drawing position of the character string extracted by the extraction step, a search step for searching for a heading candidate corresponding to the character string from the form from which the character string has been extracted;
Generating heading candidate information that associates the heading candidate searched by the searching step and the character string corresponding to the heading candidate for each form, and outputting the generated heading candidate information;
A form search step for searching a form including a character string related to a headline candidate selected by a predetermined operation from the heading candidate information output by the output step from the storage medium stored in the storage step; ,
A form search method characterized by including:

Images