JP4951407B2 - Content parts retrieval method and apparatus - Google Patents

Description

  The present invention relates to an information retrieval technique using a computer.

For example, as one of information retrieval technologies, there is a technology disclosed in Patent Document 1 (Japanese Patent Laid-Open No. 2003-15872). According to the technique disclosed in Patent Document 1, an object search device that efficiently searches each object of software developed in an object-oriented manner based on a search key is provided.
JP 2003-15872 A

  By the way, there are cases where it is necessary to design a screen content having text and input controls (for example, a text box, a radio button, or a check box). For example, in the electronic application system, it is necessary to design many input / output screens (for example, thousands of input / output screens if the electronic application system is large) as such screen content. In this case, if such a screen content is designed from scratch, the burden on the developer is large.

  As a method for supporting the design of such screen content, for example, a desired text is input as a search key, an existing screen content having a text matching the search key is searched, and the existing screen content is diverted. A way to do this is considered. However, it is desirable to reduce the burden on the developer rather than this method.

  Accordingly, an object of the present invention is to reduce the burden on a developer of screen content having text and input controls, compared to a method of searching for and using existing screen content using text as a search key.

  Other objects of the present invention will become clear from the following description.

  The storage unit stores one or more content component information groups for one screen content. The content component information group represents a content component including text and input controls, and includes one or more information packages. The information package includes a text information element representing text to be displayed and an input control type information element representing the type of input control. The search device receives an input of a search condition including a combination of a text information element and an input control type information element, and the similarity between the search condition and each content part information group in the storage unit is included in the search condition. It is calculated based on the text information element and the input control type information element that are included, and the text information element and the input control type information element that are included in the content component information group. Then, the search device displays information on the content part candidate for the search condition on the display screen based on the calculated similarity.

  In one embodiment, a content component search device is constructed by providing a search device with a search condition input unit, a component search unit, and a component candidate display unit. The search condition input unit accepts input of a search condition including a combination of a text information element representing text and an input control type information element representing a type of input control. The parts search unit determines the similarity between each content part information group in the storage unit (for example, storage resource) and the input search condition, the text information element, the input control type information element, and the content included in the search condition. Calculation is performed based on the text information element and the input control type information element included in the component information group. The component candidate display unit displays information on the content component candidate regarding the search condition on the display screen based on the calculated similarity.

  Here, the content component may be text or input control as a minimum unit, one combination of text and input control, a group including a plurality of the combinations, or screen content itself. Specifically, for example, for screen content having a hierarchical structure, when the hierarchy level is the highest, the content component is the screen content itself, and when the hierarchy level is the lowest, the content component is one If it is text, a single input control, or a combination thereof, and the hierarchy level is neither the highest nor the lowest, the content component may be a group having a plurality of combinations of text and input controls.

  In addition, the “display screen” may be one or a plurality of display screens as hardware included in one or a plurality of display devices, or as software displayed on the display screen as one or a plurality of hardware. It may be a display screen.

  Further, as “information regarding content component candidates”, content component identifier, similarity corresponding to the content component, or text information included in a certain information package of the content component information group corresponding to the content component Various information such as elements can be employed.

  Further, “displaying information on content component candidates based on the calculated similarity” may be, for example, displaying in descending order of similarity, or content having a similarity equal to or higher than a predetermined threshold. It is also possible to display information on the component candidates.

  In one embodiment, the similarity is calculated based on a first similarity between one or more text information elements included in the search condition and one or more text information elements included in the content part information group, and the search condition. Calculated based on the second similarity between one or more input control type information elements included in the content part information group and one or more input control type information elements included in the content part information group (for example, a predetermined value) A value calculated by substituting the first similarity and the second similarity into the calculation formula).

  In one embodiment, the degree of similarity is determined by one or more sets of text information elements and input control type information elements included in the search condition, and text information elements and input control type information included in the content part information group. It can be a similarity to one or more sets of elements.

  In one embodiment, the content component search device may further include a content analysis unit, a grouping unit, a package group generation unit, a text addition unit, and a component registration unit. The content analysis unit can specify a text information element, a text arrangement position, an input control type information element, and an input control arrangement position by analyzing a content file representing screen content (for example, an input form). Specifically, for example, the content analysis unit analyzes a content file, thereby combining a text / position combination that is a combination of a text information element and a text arrangement position information element, an input control type information element, and input control arrangement position information. An input control / position combination, which is a combination of elements, is specified, the specified text / position combination is stored in the text management information, and the specified input control / position combination is stored in the input control management information. The grouping unit identifies each input control arrangement positional relationship based on the analysis result of the content file (for example, by referring to the input control management information), and the identified positional relationship satisfies a predetermined first condition. When satisfied, a plurality of input controls that satisfy the condition can be extracted as a group. The package group generation unit can generate a package group having a plurality of information packages including a plurality of input control type information elements corresponding to a plurality of input controls constituting the extracted group. For example, based on the analysis result of the content file (for example, from the text management information), the text giving unit acquires the text information element corresponding to the input control type information element in the information package, and the acquired text information element is the information It can be included in the package (eg, the acquired text information element can be included as a label in the information package). The component registration unit can store a content information group, which is a package group having a plurality of information packages each including text information elements, in the storage unit in association with the analyzed content file.

  In one embodiment, the predetermined first condition may be that a distance between the arrangement positions of the input controls is equal to or less than a predetermined value.

  In one embodiment, the text information element included in the information package represents a text whose arrangement positional relationship with the input control corresponding to the input control type information element in the information package satisfies a predetermined second condition. It can be an information element.

  In one embodiment, the predetermined second condition may be that the distance between the text layout position and the input control layout position is less than or equal to a predetermined value.

  In one embodiment, the content information group is an information package tree in which an information package corresponding to the end of the lowest hierarchy level is a leaf and an information package corresponding to the head of a hierarchy level other than the lowest hierarchy level is a branch. be able to. The content part information group may be the content information group itself or a subset thereof.

  In one embodiment, the information related to the content component candidate may include a text information element in the information package at the top of the content component information group.

  In one embodiment, the content component candidate may be a content component corresponding to a content component information group whose calculated similarity is equal to or greater than a predetermined threshold.

  In one embodiment, the component candidate display unit can further display the number of content component candidates.

  In one embodiment, the component candidate display unit has content regarding a higher-level (for example, parent) search condition when there are a plurality of input search conditions and the search conditions are in a hierarchical relationship (for example, a parent-child relationship). Information regarding candidate parts can be preferentially displayed.

  In one embodiment, the component registration unit can create a component group that is a collection of identical or similar content component information groups. The component search unit can calculate the similarity between the input search condition and each component group. The component candidate display unit can display information regarding the component group as information regarding the content component candidate.

  Here, the similarity between the search condition and the part group may be a similarity between one content part information group belonging to the part group and the search condition, or a plurality of pieces of content part information belonging to the part group. It may be a value (for example, an average of a plurality of similarities) calculated by a predetermined calculation using a plurality of similarities between the group and the search condition.

  Further, the component group similar to the content component information group may be one (for example, the highest similarity) component group among the component groups whose similarity exceeds a predetermined threshold.

  In one embodiment, the component registration unit can further create a plurality of style groups for one component group. The style group may be a collection of a plurality of content component information groups having the same or similar style. The part candidate display unit can display information about the style group in addition to information about the part group.

  Here, the text style is, for example, the font size of the text, font type (eg, Mincho, Gothic, etc.), font style (eg, normal, bold, italic, etc.), font color, text length, etc. It can be defined by one or more text style attributes.

  A style is similar that a plurality of elements relating to the style (for example, text, text arrangement order, text font size, font type, etc.) match a predetermined ratio (for example, 80%) or more. Can be.

  In one embodiment, when the component registration unit registers a new content file in the storage unit, the content component information group for the new content file is changed to a component group that is the same as or similar to the content component information group. Can be classified.

  In one embodiment, when the component registration unit inputs that the search performance is more important than the registration performance, when the new content file is registered in the storage unit, the content for the new content file is stored. The component information group can be classified into a component group that is the same as or similar to the content component information group.

  In one embodiment, the component search unit can calculate the similarity between the content component information group that is not classified into any component group and the search condition. When a component part corresponding to a content component information group not classified into any component group is a content component candidate, the component registration unit can classify the content component information group into the corresponding component group .

  In one embodiment, when the component registration unit inputs that the registration performance is more important than the search performance, a content component corresponding to a content component information group that is not classified into any component group is a content component candidate. If it is determined that the content component information group is classified into a corresponding component group.

  In one embodiment, the plurality of content component information groups are classified into two or more component groups, and the component search unit calculates the similarity between the input search condition and each component group, and displays the component candidate display The section can display information on the component group as information on the content component candidate.

  In one embodiment, there is a plurality of style groups for one component group, the style group is a collection of a plurality of content component information groups having the same or similar style, and the component candidate display section relates to the component group. In addition to information, information about style groups can be displayed.

  In one embodiment, if an information package in an existing component subtree that belongs to a first component group belongs to a second component group that is different from the first component group, the second component group is The component candidate display unit can display information on the second component group included in the first component group.

  In one embodiment, the component candidate display unit can display the number of content component information groups corresponding to the style group.

  In one embodiment, the search condition may be definition information of a screen content having a hierarchical structure or a subset thereof.

  Two or more embodiments of the plurality of embodiments described above can be combined. Each of the above-described units (for example, a search condition input unit, a component search unit, a component candidate display unit, a content analysis unit, a grouping unit, a package group generation unit, a text addition unit, and a component registration unit) may be rephrased as each means. Good. Each unit can also be realized by hardware (for example, a circuit), a computer program, or a combination thereof (for example, a part is executed by a computer program and a part is executed by a hardware circuit). Each computer program can be read from a storage resource (for example, memory) provided in the computer machine. The storage resource can be installed via a recording medium such as a CD-ROM or DVD (Digital Versatile Disk), or can be downloaded via a communication network such as the Internet or a LAN.

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

  <First embodiment>.

  Web content can be taken as an example of screen content. The Web content may be digital content mainly composed of text excluding images and moving images, such as HTML (HyperText Markup Language) content, a unique format Web content, or an electronic document created by a document creation application. In the first embodiment, an HTML input form can be taken as an example of Web content. In the following description, Web content such as an input form is simply referred to as “content”.

  The processing performed in the first embodiment can be broadly divided into a component registration step and a component search step.

  In the component registration step, a grouping step for grouping elements using layout information such as position information of elements (text and input control) included in existing content, and a label is assigned to each group and element A label assignment step, a content structure tree generation step for generating a tree (hereinafter referred to as a content structure tree) representing a content structure from the results of the grouping step and the label assignment step, and a content structure tree associated with the content in the repository A content registration step for registration is performed.

  In the component search step, the subtree of definition information for new content having a hierarchical structure (hereinafter referred to as a query tree) is similar to the subtree of each content configuration tree stored in the repository (hereinafter referred to as an existing component subtree). A component search step for calculating a degree and a component candidate output step for outputting an existing component subtree having a high similarity to the query tree (for example, a predetermined threshold or more) as a component candidate for the query tree are performed.

  This will be described in more detail below.

  FIG. 1 shows an example of the overall configuration of a content parts search system according to the first embodiment of the present invention.

  The content component search system can be composed of one or a plurality of computers. The content component search system includes, for example, a CPU 100, a magnetic disk device 101, a main memory 102, a floppy disk drive (hereinafter referred to as an FDD 103), and a bus 104 connecting them. The content component search system can be connected to other devices via the network 105.

  The magnetic disk device 101 is one of secondary storage devices, and stores a content file 160 including a content source and a content configuration tree 161. Information stored in the floppy disk 106 is read into the magnetic disk device 101 or the main memory 102 via the FDD 103.

  The main memory 102 includes a system control processing unit 110, a component registration control processing unit 111, a component search control processing unit 112, a content acquisition processing unit 121, an in-content information acquisition processing unit 122, a content configuration tree generation processing unit 123, and a grouping process. , 130, label addition processing unit 131, component registration processing unit 126, input / output definition information acquisition processing unit 140, input / output definition information analysis processing unit 141, component search processing unit 142, and component candidate output processing unit 143 are stored. Area 150 is secured. In this embodiment, system control, component registration control, component search control, content acquisition, content information acquisition, content configuration tree generation, grouping, label assignment, component registration, input / output definition information acquisition, input / output definition information analysis, Each process of component search and component candidate output includes a system control processing unit 110, a component registration control processing unit 111, a component search control processing unit 112, a content acquisition processing unit 121, an in-content information acquisition processing unit 122, and a content configuration tree generation process. Unit 123, grouping processing unit 130, label addition processing unit 131, component registration processing unit 126, input / output definition information acquisition processing unit 140, input / output definition information analysis processing unit 141, component search processing unit 142, and component candidate output processing unit 143. Is executed by the CPU 100, but each process is performed. It may be implemented in hardware, for example, by an integrated circuit as a program. In order to simplify the description below, each program implemented by the CPU 100 executing various programs will be described as the subject of each process. When each processing unit is realized by hardware, each processing unit mainly performs each process.

  The system control processing unit 110 controls the component registration control processing unit 111 and the search control processing unit 121.

  The component registration control processing unit 111 controls the content acquisition processing unit 121, the in-content information acquisition processing unit 122, the content configuration tree generation processing unit 123, the grouping processing unit 130, the label addition processing unit 131, and the component registration processing unit 126. .

  The content configuration tree generation processing unit 123 controls the grouping processing unit 130 and the label addition processing unit 131.

  The component search control processing unit 112 controls the input / output definition information acquisition processing unit 140, the input / output definition information analysis processing unit 141, the component search processing unit 142, and the component candidate output processing unit 143.

  In this embodiment, these processing units are stored in the main memory 102, but a storage medium (not shown in FIG. 1) such as a magnetic disk device 101, a floppy disk 106, an MO, a CD-ROM, or a DVD. Can be stored in the main memory 102 via the driving device and executed by the CPU 100.

  In the present embodiment, the content file 160 and the content configuration tree 161 are stored in the magnetic disk device 101, but a floppy disk (registered trademark) 106, an MO (Magneto-Optical disk), a CD-ROM, It may be stored in a storage medium such as a DVD (not shown in FIG. 1), or may be stored in a storage device (not shown in FIG. 1) connected to another system via the network 105. It may be stored. Further, it may be stored in a storage medium (not shown in FIG. 1) directly connected to the network 105.

  Hereinafter, an outline of processing performed in the content component search system according to the present embodiment will be described.

  The system control processing unit 110 analyzes a command input from the keyboard 101 (which may be another type of input device). If the result is analyzed as a registration execution command, the system control processing unit 110 activates the component registration control processing unit 111, and the component registration control processing unit 111 executes the content file 160 and the content configuration tree. 161 is registered in the magnetic disk device 101. Further, when it is analyzed that the command is a search execution command, the system control processing unit 110 activates the component search control processing unit 112, and the component search control processing unit 112 performs the specified query tree 171 (FIG. 11). The part search process is performed based on the reference).

  Hereinafter, the flow of processing performed by the component search control processing unit 112 will be described.

  First, the component search control processing unit 112 activates the input / output definition information acquisition processing unit 140, and the input / output definition information acquisition processing unit 140 defines definition information of newly created content (hereinafter referred to as “input / output definition information”). 170 (see FIG. 11) is acquired, and the acquired input / output definition information 170 is stored in the work area 150.

  Next, the part search control processing unit 112 activates the input / output definition information analysis processing unit 141, and the input / output definition information analysis processing unit 141 is designated by the user (for example, developer) from the input / output definition information 170. The query tree 171 is acquired, and the acquired query tree 171 is stored in the work area 150.

  Next, the component search control processing unit 112 activates the component search processing unit 142, and the component search processing unit 142 performs processing for all components included in each content file 160 (hereinafter referred to as an existing component subtree 174). The degree of similarity with the query tree 171 is calculated, and the calculation result is stored in the work area 150 as the evaluation result 183 (see FIG. 12).

  Finally, the component search control processing unit 112 activates the component candidate output processing unit 143, and the component candidate output processing unit 143 converts the existing component subtree 174 having a high similarity to the query tree 171 into the component for the query tree 171. Output as a candidate.

  The input / output definition information acquisition processing unit 140 stores the input / output definition information 170 input by the user in the work area 150. The input / output definition information 170 can be expressed in an XML (Extensible Markup Language) format, for example, as shown in FIG. The structure information of the input item of the newly created content is expressed by the XML structure, the label of the input item (the label substantially means the input item name) is expressed by the XML tag name, and the type of the input control is expressed. , And can be expressed by the type attribute of each element. In this embodiment, the data format of the input / output definition information 170 is the XML format. However, any data format can be used as long as it can define the input item structure information, the input item label, and the input control type. May be. The input / output definition information 170 is input from the user, but a storage medium (not shown in FIG. 1) such as a floppy disk 106, MO, CD-ROM, or DVD via a drive such as the FDD 103. Stored in a storage device (not shown in FIG. 1) connected to another system via the network 105, or a storage medium directly connected to the network 105 ( It may be input from those stored in (not shown in FIG. 1).

  The input / output definition information analysis processing unit 141 uses the structure information (hereinafter referred to as “user-specified structure information”) 172 specified by the user from the input / output definition information 170 acquired by the input / output definition information acquisition processing unit 140. 11 is obtained, and the obtained query tree 171 is stored in the work area 150. The query tree 171 is a subset of the input / output definition information 170.

  The flow of processing performed by the component search processing unit 142 will be described with reference to FIGS.

  First, the component search processing unit 142 repeats Steps 2010 to 2011 for all the query trees 171 acquired by the input / output definition information analysis processing unit 141 (Step 2000).

  Next, the part search processing unit 142 acquires the in-part information 177A (information including the label information 175A and the input control information 176A) from the query tree 171 and stores the acquired in-part information 177A in the work area 150. Store (step 2010).

  Next, the component search processing unit 142 repeats Step 2012 for all the content files 160 registered in the magnetic disk device 101 (Step 2011).

  Next, the component search processing unit 142 repeats Steps 2113 to 2015 for all the existing component subtrees 174 included in the content configuration tree 161 corresponding to the content file 160 (Step 2012).

  Next, the part search processing unit 142 acquires the in-part information 177B (information including the label information 175B and the input control information 176B) of the existing part subtree 174, and uses the acquired in-part information 177B as the work area 150. (Step 2013).

  Next, the part search processing unit 142 uses the in-part information 177A of the query tree 171 and the in-part information 177B of the existing part subtree 174 to obtain vector space 182 and vector information 182 in a table format expressing vectors. Label information 178, input control information 179, query tree vector information 180, and information including existing parts subtree vector information 181) are generated, and the generated vector information 182 is stored in the work area 150 (step 2014).

  Finally, the part search processing unit 142 uses the vector information 182 to calculate the similarity of the existing part subtree 174 with respect to the query tree 171 using (Equation 1) (step 2015), and calculates the similarity calculation result as follows: An evaluation result 183 (information including content ID information 1831, node number information 1832, and similarity information 1833) of each existing component subtree 174 for the query tree 171 is stored in the work area 150. The similarity is calculated from the vector information 182 by a cosine scale according to the vector space method.

  In step 2015, the vector space in the similarity calculation formula is calculated using a space (dimension) including only the query tree for calculating the similarity and the existing component subtree. It is good also as what calculates in the common space (dimension) regarding the content 160 of this, and a query tree.

  First, the component candidate output processing unit 143 obtains information on candidate components corresponding to the existing component subtree 174 in descending order of similarity for each query tree 171 from the evaluation result 183 calculated by the component search processing unit 142. A list (hereinafter referred to as “part candidate list”) is displayed on the part candidate list display screen 187 (see FIG. 13). In the component candidate list, each component candidate is stored in the component ID (for example, the identifier of the component expressed by “content ID_content configuration tree node number”), the similarity, and the root node of the existing component subtree. The label is included.

  Next, the component candidate output processing unit 143 acquires a code corresponding to the selected component from the content file 160 for the component candidate selected by the user from the component candidate list, and the acquired code is used as the work area. 150.

  Finally, the component candidate output processing unit 143 displays the component selected by the user on the preview screen 184 (see FIG. 13) using the code acquired above.

  The processing procedure of the component candidate output processing unit 143 has been described above.

  Note that the component candidate is displayed on the preview screen 184 when the user selects a component candidate from the component candidate list, but the component candidate output processing unit 143 displays the query candidate in the query tree 171 without any user operation. On the other hand, the part corresponding to the existing part subtree having the highest similarity may be displayed on the preview screen 184.

  When there are a plurality of query trees designated by the user, the query tree information screen 186 (see FIG. 13) shows the relationship between the plurality of query trees as query tree information (hereinafter referred to as “query tree information”). Display).

  In the component candidate list, an existing component subtree that does not have a similarity greater than or equal to a certain threshold may not be included as a component candidate. At this time, the number of parts searched for each query tree (the number of existing part subtrees) may be displayed on the query tree information display screen 186.

  In addition, when there is a hierarchical relationship (for example, a parent-child relationship) between a plurality of query trees specified by the user (including a case where all subtrees included in the input / output definition information 170 are used as the query tree 171), a higher level (for example, It may be preferentially displayed on the part candidate list display screen 187 from the candidate parts of the parent) query tree. For example, even if the similarity of the part candidate of the upper query tree is lower than the similarity of the part candidate of the lower query tree, the part candidate of the upper query tree is displayed in the lower query. It means that it is positioned higher in descending order than the tree part candidates.

  In addition, the component candidate output processing unit 143 uses the query tree in the input / output definition information displayed on the input / output definition information display screen 185 so that the contents of the query tree 171 for the currently displayed component candidate list can be easily determined. The portion 171 may be highlighted (for example, shaded).

  This completes the description of the processing performed by the component search control processing unit 112.

  Next, processing performed by the component registration control processing unit 111 will be described.

  First, the component registration control processing unit 111 activates the content acquisition processing unit 121, and the content acquisition processing unit 121 reads the content file 160 stored in the floppy disk 108 via the FDD 104.

  Next, the component registration control processing unit 111 activates the in-content information acquisition processing unit 122, and the in-content information acquisition processing unit 122 performs text and input control (for example, a text box or radio) included in the source of the content file 160. Button, etc.) and position information and sizes (for example, font size for text, area size for input control) of these elements in the content.

  Next, the component registration control processing unit 111 activates the content configuration tree generation processing unit 123, and the content configuration tree generation processing unit 123 uses the element information in the content extracted by the in-content information acquisition processing unit 122. The content composition tree 161 is generated by grouping between elements and assigning labels to the elements and groups.

  Finally, the component registration control processing unit 111 activates the component registration processing unit 126, and the component registration processing unit 126 registers the content file 160 and the content configuration tree 161 corresponding to the content file 160 in the magnetic disk device 101.

  The content acquisition processing unit 121 reads the content file 160 stored in the floppy disk 108 via the FDD 104 and stores it in the work area 150. Although the content file 160 is stored in the floppy disk 108, it may be stored in a storage medium (not shown in FIG. 1) such as MO, CD-ROM, DVD, or network. The data may be stored in a storage medium (not shown in FIG. 1) connected to another system via 105.

  First, the in-content information acquisition processing unit 122 first selects a text information element or input control from the content file 160 acquired by the content acquisition processing unit 121 based on the rules stored in the magnetic disk device 101 or the main memory 102. Type information elements (hereinafter, these may be referred to as “content information”) are extracted. Text list information 162 in which all the extracted text information elements are recorded and input control list information 163 in which all the extracted input control type information elements are recorded are stored in the work area 150. “Text” here refers to a character string that is actually displayed on the screen, not a tag that is not displayed on the screen, for example, when the content file is an HTML file. The “input control” here refers to an element that accepts designation from the user such as an INPUT tag, a SELECT tag, and a TEXTAREA tag when the content file is an HTML file, for example.

  The text list information 162 and the input control list information 163 are stored in the work area 150, but may be stored in the magnetic disk device 101. Information indicating what is extracted from the content file as text and information indicating what is extracted from the content file as input control are stored in the main memory 102 or the magnetic disk device 101. Stored.

  Next, the in-content information acquisition processing unit 122 extracts the arrangement position information of the extracted in-content element from the content file. Here, the arrangement position information is information representing the definition of the range occupied by the element in the content and the position in the content. Specifically, for example, text layout position information includes reference position information and font size information, and input control layout position information includes reference position information and area size information. Text reference position information and font size information are recorded in the text list information 162, and input control reference position information and area size information are recorded in the input control list information 163. The text list information 162 is, for example, a table as illustrated in FIG. 5, and for each text detected from the content file, a text number 700 (for example, a serial number), a text information element 710, and reference position information 720 (for example, for example). The coordinates of the reference position when the predetermined position of the content is the origin) and the font size 730 are recorded. Further, as illustrated in FIG. 6, the input control list information 163 is, for example, a table, and for each input control detected from the content file, an input control number 800 (for example, a serial number) and an input control type information element 810 (for example, for example). Information with a tag), reference position information 820, and region size 830 (for example, width and height sizes centered on the reference position) are recorded.

  The content configuration tree generation processing unit 123 first activates the grouping processing unit 130, and the grouping processing unit 130 uses the input control list information 163 extracted by the in-content information acquisition processing unit 122 to be included in the content file 160. The input control is grouped, and the grouping result (see, for example, FIG. 8) is stored in the work area 150.

  Next, the content configuration tree generation processing unit 123 activates the label addition processing unit 131, and the label addition processing unit 131 performs text list information 162 and an input control list on the grouping result generated by the grouping processing unit 130. By using the arrangement position information stored in the information 163, the name of each input control or the name of a group generated by grouping (hereinafter collectively referred to as a label) is given, so that the final content configuration tree 161 is obtained. And the generated content configuration tree 161 is stored in the work area 150.

  Note that the content configuration tree 161 has, as node attribute information, each node has a node number, a label, a reference point, a parent node number (the most recent upper node number), and a control type. FIG. 9 shows a conceptual diagram of a content structure tree and an example of storing attribute information of each node. FIG. 9 shows the upper part of the content structure tree, and the lower part is not shown. The content configuration tree 161 is stored in the work area 150, but may be stored in the magnetic disk device 101.

  The component registration processing unit 126 first assigns a content ID to each content file 160. The content ID is generated and assigned based on the rules stored in the main memory 102 or the magnetic disk device 101.

  Next, the component registration processing unit 126 associates the content file 160 and the content configuration tree 161 in the same content, and registers them in the magnetic disk device 101.

  FIG. 2 shows an example of the flow of processing performed by the grouping processing unit 130.

  First, the grouping processing unit 130 performs steps 301 to 302 for n input controls included in the input control list information 163 extracted by the in-content information acquisition processing unit 122, with n = 1 to (N− Repeat until 1) (step 300).

  Next, the grouping processing unit 130 performs step 302 from m = 2 to (M−1) for M hierarchical levels based on the value M stored in the magnetic disk device 101 or the main memory 102. Repeatedly (step 301). Note that the hierarchical level m represents the entire content, and the higher the m, the more detailed the grouping is performed, and at the lowest hierarchical level (that is, m = M-1), 1 group 1 input control is performed. .

  Next, the grouping processing unit 130 determines whether or not the nth and n + 1th input controls belong to the same group with respect to the hierarchy level m as the grouping condition at the hierarchy level m. (Step 302). When the nth and n + 1th input controls are determined to be the same group at the hierarchical level m, and the n + 1th and n + 2th input controls are determined to be the same group, the nth to n + 2th input controls. Are considered the same group at hierarchy level m. That is, if there are multiple input control pairs that are determined to be the same group, and one input control in the input control pair is the other input control in another input control pair, the input control in the multiple input control pairs Are considered to be the same group. The nth input control represents the nth input control (for example, n is the input control number 800 in FIG. 6) input in the content.

  Finally, the grouping processing unit 130 stores the parent node number as attribute information in each node in the content configuration tree 161 in order to manage the result of grouping by hierarchical level in step 302 in a tree structure (step 303). The content configuration tree 161 includes, as attribute information of each node, a node number, a label, a reference point of a corresponding area (a range occupied by the input control) in the content, a node number of the parent node, and an input control type (control). However, the label and the reference point are not stored when the processing of the grouping processing unit 130 is completed.

  Here, the grouping method between the nth and n + 1th input controls performed in step 302 will be described in detail.

  The grouping method between the nth and n + 1th input controls is determined by determining the determination rule (hereinafter referred to as a grouping condition) in the following eight cases depending on the arrangement positional relationship between the nth and n + 1th input controls. To do. The reference point that appears in the following description represents the upper left of the input control. Further, the rightmost end, the leftmost end, the uppermost end, and the lowermost end represent the endmost points of the input control determined from the reference point and the region size 830.

  First, when the reference point of the (n + 1) th input control is located right to the reference point of the nth input control (hereinafter referred to as the first case), the rightmost end of the nth input control and the (n + 1) th input control Nth and n + 1th input controls are considered to belong to the same group at the hierarchy level m when the horizontal distance from the leftmost edge of the input control satisfies the grouping condition at the hierarchy level m in the first case.

  When the reference point of the (n + 1) th input control is located directly below the reference point of the nth input control (hereinafter referred to as the second case), the lowest end of the nth input control and the (n + 1) th input control When the vertical distance from the topmost edge satisfies the grouping condition at hierarchical level m in the second case, the nth and n + 1th input controls are considered to belong to the same group at hierarchical level m.

  When the reference point of the (n + 1) th input control is located in the lower right with respect to the reference point of the nth input control (hereinafter referred to as the third case), the rightmost end of the nth input control and the (n + 1) th input The horizontal distance from the leftmost edge of the control satisfies the grouping condition at the hierarchical level m in the third case, and the vertical distance between the lowest edge of the nth input control and the highest edge of the (n + 1) th input control is 3rd. When the grouping condition at hierarchical level m in the case is met, the nth and n + 1th input controls are considered to belong to the same group at hierarchical level m.

  When the reference point of the (n + 1) th input control is located in the upper right with respect to the reference point of the nth input control (hereinafter referred to as the fourth case), the rightmost end of the nth input control and the (n + 1) th input control When the horizontal distance from the leftmost edge of the second pixel satisfies the grouping condition at the hierarchical level m in the fourth case, the vertical distance between the uppermost edge of the nth input control and the lowermost edge of the (n + 1) th input control is the fourth distance. When the grouping condition at hierarchical level m in the case is met, the nth and n + 1th input controls are considered to belong to the same group at hierarchical level m.

  When the reference point of the (n + 1) th input control is located at the lower left with respect to the reference point of the nth input control (hereinafter referred to as the fifth case), the lowest end of the nth input control and the (n + 1) th input control Nth and n + 1th input controls are considered to belong to the same group at hierarchy level m when the vertical distance to the topmost edge satisfies the grouping condition at hierarchy level m in the fifth case.

  In addition, when the reference point of the (n + 1) th input control is located to the left of the reference point of the nth input control (hereinafter referred to as the sixth case), the reference point of the nth input control When the reference point of the (n + 1) th input control is located directly above (hereinafter referred to as the seventh case), the reference point of the (n + 1) th input control is located at the upper left with respect to the reference point of the nth input control. In the case (hereinafter referred to as the eighth case), the nth and n + 1th input controls are considered not to belong to the same group at all hierarchy levels except hierarchy level 1.

  FIG. 7 is a diagram illustrating an example of grouping condition list information 900 that stores grouping conditions 930. The grouping condition information 900 is, for example, a table. For each case, the case number 910, information indicating the arrangement positional relationship of the (n + 1) th input control with respect to the nth input control (hereinafter referred to as “positional relationship information”) 920, And a grouping condition 930. For example, in the case of the first case, the case number 910 is “1”, the positional relationship information 920 is “true right”, and the grouping condition 930 is “the distance between the nth rightmost edge and the (n + 1) th leftmost edge. Information such as “less than 100 px” is stored.

  The grouping condition 920 is set for each hierarchical level. For this reason, for example, when there is one grouping condition list information 900 for each hierarchical level, each grouping condition list information 900 corresponding to each hierarchical level is prepared. The grouping condition list information 900 is stored in the main memory 102 or the magnetic disk device 101, for example.

  Note that the grouping condition 930 illustrated in FIG. 7 may be set, for example, based on a general input form (content), in which a threshold regarding the relative distance between the nth and n + 1th input controls is set. it can. However, the present invention is not limited to this. For example, after the relative distance between the nth and n + 1th input controls is actually extracted from the content file 160, the relative distance in the grouping condition 930 is dynamically calculated based on the average value. May be set. Specifically, for example, in the case of the second case, the grouping condition is “the vertical distance between the nth lowest end and the (n + 1) th highest end is [average interval between adjacent input controls in the vertical direction]” It may be “within”. Such dynamic threshold setting (condition setting) makes it easy to set an optimal grouping condition and can absorb individual differences among developers in the creation criteria between contents.

  In step 302, the grouping is performed using the position information of the input control, but the font size information of the text list information 162 generated by the in-content information acquisition processing unit 122 is used to set a font size greater than a certain threshold. A method may be used in which the boundary of the group between the nth and n + 1th input controls at the hierarchical level m is regarded as a group boundary.

  In addition, a method of using the document structure information such as H1 of the text list information 162 generated by the in-content information acquisition processing unit 122 to regard a group boundary between the nth and n + 1th input controls at the hierarchical level m. It may be taken. By performing the grouping process using the font size and document structure information together with the position information, the grouping accuracy is improved.

  FIG. 3 shows an example of the flow of processing performed by the label addition processing unit 131.

  First, the label addition processing unit 131 repeats Steps 410 to 411 and 415 for M hierarchical levels of the content structure tree 161 until m = M-1 (Step 400).

  Next, the label assignment processing unit 131 determines whether or not the hierarchy level is the lowest hierarchy level (whether m = M) (step 410).

  As a result of the determination in step 410, when the hierarchical level is the lowest hierarchical level (m = M), the label assignment processing unit 131 performs steps 412 to 414 for N input controls included in the entire content. = 1 to N are repeatedly executed (step 411).

  Next, the label assignment processing unit 131 sets the reference point of the nth input control as the upper left of the area and stores it in the work area 150 (step 412).

  Next, the label addition processing unit 131 moves upward and leftward from the text in the content stored in the text list information 162 with respect to the reference point of the nth input control (from the left if the distance is the same). The text at the closest distance is extracted as the label of the nth input control, and the label is stored in the work area 150 as attribute information of the nth node at the hierarchical level M in the content structure tree 161. (Step 413).

  Next, the label assignment processing unit 131 sets the reference point of the nth input control including the label as the upper left of the label area, and sets the reference point as the attribute information of the nth node at the hierarchical level M in the content configuration tree 161. Is stored in the work area 150 (step 414). If the label cannot be extracted in step 413, the label addition processing unit 131 sets the upper left of the input control set in step 412 as a reference point.

  Next, if the result of determination in step 410 is that the hierarchy level is not the lowest hierarchy level, the label assignment processing unit 131 performs steps 415 to 418 for n groups included in the hierarchy level m, where n = 1 to 1. Repeat until N (step 415).

  Next, the label addition processing unit 131 selects the reference point of the element having the reference point of the upper left (e.g., priority over the left when the distance is the same) among the elements (child nodes) included in the nth group. Is stored in the work area 150 as a reference point of the nth group (step 416).

  Next, the label addition processing unit 131 performs the upward and leftward (distances) from the text in the content stored in the text list information 162 with respect to the reference point of the nth group set in step 416. If it is the same, the text that is closest to the left) and has not been extracted as a label yet is extracted as the label of the nth group, and the label of the node in the content composition tree 161 is extracted. The attribute information is stored in the work area 150 (step 417).

  If the text that has not yet been extracted as a label belongs to a group that is not the nth group, or if there is no text that has not yet been extracted as a label, the label assignment processing unit 131 determines that the nth group. Let the label of the node of the reference point be the label of the nth group.

  Next, the label assignment processing unit 131 sets the reference position of the nth group including the label as the upper left of the label area, and stores the reference position in the work area 150 as attribute information of the node in the content configuration tree 161. (Step 418).

  The label addition processing unit 131 can sequentially apply labels in a bottom-up manner from the leaf of the node tree representing the grouping result to the root.

  In step 413, the label of the nth input control is selected from the text in the content stored in the text list information 162 in the upward direction and the left direction (if the distance is the same). The text is the closest distance in the top priority (from left to top), but the radio buttons and check boxes are closest to the reference point of the input control in the right and down directions (if the distance is the same, the right priority is given to the bottom) It may be text at a distance.

  In addition, regarding a group in which adjacent input controls are arranged without an interval, it is regarded as a table structure of n rows × m columns, and labels given to the columns of the first row are assigned to the columns after the nth row. Also give the same label. For example, the label given to the input control in the first column of the first row is also used as the label given to the input control in the first column of the second row.

  Also, the group label extraction method according to the present embodiment is the closest distance in the upward direction and the left direction from the text in the content stored in the text list information 162 with respect to the reference position of the group (node). Although the text is extracted as a label, a threshold related to distance may be set as an extraction condition. This reduces false labeling.

  Also, the group label extraction method in this embodiment is the closest distance in the upward and left directions from the text in the content stored in the text list information 162 with respect to the reference point of the group (node). Although text is extracted as a label, it may be extracted as a representative value from the labels of elements (child nodes) included in the group, or it may be labeled in training data by training data from other content. You may use the label provided to the group similar to the group to give. Thereby, even when the text applicable to a label does not exist in content, a label can be provided to each content element or group.

  The above is the description of the processing performed by the component registration control processing unit 111.

  Hereinafter, processing performed by the component search control processing unit 112 and the component registration control processing unit 111 will be described in more detail. First, specific processing of the component search control processing unit 112 will be described.

  FIG. 11 is a diagram illustrating specific processing of the input / output definition information analysis processing unit 141.

  The input / output definition information analysis processing unit 141 extracts a subtree corresponding to the user-specified structure information 172 specified by the user from the input / output definition information 170 acquired by the input / output definition information acquisition processing unit 140, and the query tree 171. To get. In the example of this figure, “Applicant Information” is specified as the user-specified structure information 172, and the subtree of “Applicant Information” is acquired as the query tree 171 from the input / output definition information 170.

  FIG. 12 is a diagram for explaining specific processing of the component search processing unit 142.

  The component search processing unit 142 calculates the similarity with one query tree 171 for all the existing component subtrees 174 included in all the content files 160 registered in the magnetic disk device 101. The existing part subtree 174 is a subset of the content structure tree 161, and may be a leaf of the content structure tree 161 as a minimum unit.

  In the example of FIG. 12, the query tree 171 of the “Applicant Information” portion and the existing component subtree 174 (content configuration tree shown in FIG. 9) of the content configuration tree 161 in the content 500 (content indicated by the content file 160) shown in FIG. A detailed specific example of similarity calculation processing of a subtree of node number 2 of 1100) is shown.

  First, the part search processing unit 142 acquires the in-part information 177A (information including the label information 175A and the input control information 174A) from the query tree 171, and similarly, the in-part information 177B (label) from the existing part subtree 174. Information 175B and input control information 174B). In the example of this figure, seven label information elements (text information elements) and input control type information elements are acquired from the query tree 171, and eight label information elements and input control types are acquired from the existing part subtree 174. An information element is obtained.

  Next, the component search processing unit 142 generates, as vector information 182, label information 178, input control information 179, query tree vector, in order to generate a vector space and a vector necessary for similarity calculation from the in-component information 177. 180 and an existing part subtree vector 181 are generated. In the example of this figure, as the dimension representing the vector space, in other words, using the in-part information 177A and 177B as the label information 178 and the input control information 179, (Name · text, Address · text, Telephone Number · text, Company Name / text, Year / select, Month / select, Day / select, First / text, Last / text, Email Address / text) are generated (in the above parentheses, label information element / input control type information element, ) In addition, as a query tree vector 180 in the generated vector space (1, 1, 1, 1, 1, 1, 1, 0, 0, 0) and as an existing part subtree vector 181 (0, 1, 1, 0, 1, 1, 1, 1, 1, 1) has been generated. The element of each vector is, for example, “1” when the combination of the label “Name” and the input control type “text” exists, and “0” when the combination does not exist.

  Next, the component search processing unit 142 calculates the similarity of the existing component subtree 174 with respect to the query tree 171 from the generated vector information 182 by (Equation 1), and the content ID information 1831, node number information 1832, and similarity An evaluation result 183 including information 1833 is generated and stored in the work area 150.

  In the example of this figure, “0.67” is calculated as the similarity of the existing component subtree 174 with respect to the query tree 171, and the calculation result is “0001” as the content ID information 1831 of the evaluation result 183 and the node number information 1832. “0.67” is stored as “2” and similarity information 1833.

  FIG. 13 is a diagram illustrating specific processing of the component candidate output processing unit 143.

  The component candidate output processing unit 143 outputs a component candidate list display 187 from the similarity calculation result (evaluation result 183) generated by the component search processing unit 143. In the example of this figure, a list of candidate parts for the query tree 171 is displayed on the candidate part list display screen 187 in descending order of similarity with respect to the parts included in the contents of the content IDs 0001 to 0003. According to the list, the first candidate component is a component having a component ID “0001_2”, a similarity “0.67”, and a label “Applicant Information”.

  Further, in the example of this figure, a preview button for previewing a component is displayed for each component on the component candidate list display screen 187. When the user presses the preview button of the first candidate component (component ID: 0001_2), the candidate component output processing unit 143 displays the code corresponding to the first candidate component of the content including the candidate component. The part image of the first candidate part is displayed on the preview screen 184 based on the acquired code.

  Further, in the example of this figure, an input / output definition information display screen 185 is displayed. In the input / output definition information display screen 185, the query tree 171 portion in the input / output definition information 170 corresponding to the currently output component candidate list is displayed. Is shown by shading.

  In the example of this figure, “Applicant Information”, “Date of Birth”, and “Service” are designated as the plurality of query trees 171, and the relationship (for example, hierarchical relationship) between these query trees is the query. It is displayed on the tree information display screen 186. It can be seen from the query tree information display screen 186 that there is a parent-child relationship in which the query tree “Applicant Information” is a parent and the query tree “Date of Birth” is a child. Further, among these three query trees 171, the candidate parts for the query tree “Applicant Information”, which has the candidate parts and is the highest hierarchical level, are preferentially displayed in the candidate part list display 187.

  Further, in the example of this figure, the similarity threshold is set to “0.50”, and four parts with similarity of 0.50 or more are displayed as part candidates on the part candidate list display screen 187. Has been. That is, the component candidate output processing unit 143 does not display a component having a similarity threshold value less than “0.50” as a component candidate. Note that instead of or in addition to the similarity threshold, only the upper predetermined number of components from the components with a high similarity may be displayed.

  In addition, the number of candidate parts for each query tree 171 is displayed on the query tree information display screen 186. From the screen 186, it can be seen that there are four candidate parts for the query tree “Applicant Information”. Further, it can be seen that there is no case for the query tree “Service”, that is, there is no component candidate.

  Moreover, according to FIG. 13, one display screen is comprised by the combination of the input / output definition information display screen 185, the preview screen 184, the query tree information display screen 186, and the component candidate list display screen 187. May be separated from each other or overlap.

  The above is the specific processing of the component search control processing unit 112. The developer can create a new content by selecting a desired component candidate from the component candidate list and editing the component candidate as it is or editing it. The developer can register a content file representing the new content in the magnetic disk device 101.

  Next, specific processing of the component registration control processing unit 111 will be described.

  Specific processing examples of the content acquisition processing unit 121 and the in-content information acquisition processing unit 122 will be described with reference to FIGS. 4, 5, and 6.

  FIG. 4 shows an example of a conference application form for HTML content accessed from a client terminal as an example of the content 500.

  First, the in-content information acquisition processing unit 122 reads text information elements and arrangement position information (for example, reference position information and font size) of the text from the HTML source (content file 160) of the content 500 acquired by the content acquisition processing unit 121. ) And the text number 700, text information 710, reference position information 720, and font size 730 are recorded in the text list information 162. A specific example of the text list information 162 is shown in FIG. In the example of FIG. 5, a total of 13 text information elements are recorded in the text list information 162 (for example, “XX Conference Form” is recorded as the first text information element 710, and the reference position information "X: 177 px Y: 35 px" is recorded as 720, and "+4" is recorded as the font size 730).

  Next, the in-content information acquisition processing unit 122, from the HTML source of the content 500 acquired by the content acquisition processing unit 121, an INPUT tag and a SELECT tag (information element indicating the type of input control), and the arrangement of these input controls Position information (for example, reference position information and area size) is extracted, and the input control number 800, the input control type information element 810, the reference position information 820, and the area size 830 are stored in the input control list information 163. A specific example of the input control list information 163 is shown in FIG. In the example of FIG. 6, a total of 15 input control type information elements are recorded in the input control list information 163 (eg, “<INPUT type =“ text ”> as the first input control type information element 810) "Is recorded," X: 315 px Y: 161 px "is recorded as the reference position information 820, and" width: 73 px height: 20 px "is recorded as the area size 830).

  A specific processing flow of the content configuration tree generation processing unit 123 when generating a content configuration tree having five hierarchical levels for the content 500 shown in FIG. 4 will be described with reference to FIGS.

  The grouping processing unit 130 reads the input control list information 163 shown in FIG. 6 and determines whether or not the input controls adjacent to each other in the input order belong to the same group for each hierarchical level (2 to M−1). The determination is made based on the grouping condition list 900 stored in the main memory 102 or the magnetic disk device 101.

  FIG. 7 is an example of the grouping condition list 900 at the hierarchical level 2. In the example of FIGS. 5 and 6, regarding the grouping process at hierarchical level 2, first, the input control list information 163 shown in FIG. 6 indicates that the positional relationship of the second input control with respect to the first input control is “true right”. For this reason, the first case in the grouping condition list 900 shown in FIG. 7 is used.

  From the reference position information and area size shown in the input control list information 163, the horizontal distance between the rightmost end of the first input control and the leftmost end of the second input control is 52 px. The positional relationship between the second input control and the second input control is that the horizontal distance between the rightmost end of the nth input control and the leftmost end of the (n + 1) th input control is the first case condition in the grouping condition 900. Satisfies “less than 100 px”. As a result, the first input control and the second input control are regarded as the same group at the hierarchical level 2.

  Next, from the input control list information 163, since the positional relationship of the third input control with respect to the second input control is “lower left”, the fifth case in the grouping condition 900 is used. Since the vertical distance between the lowermost end of the second input control and the uppermost end of the third input control is 15 px from the reference position information and area size indicated in the input control list information 163, the second The positional relationship between the input control and the third input control is “the vertical distance between the lowest end of the nth input control and the highest end of the (n + 1) th input control, which is the condition of the fifth case in the grouping condition 900. Satisfies “less than 20 px”. As a result, the second input control and the third input control are regarded as the same group at the hierarchical level 2.

  Next, according to the input control list information 163, the positional relationship of the 10th input control with respect to the 9th input control is “lower left”, so the fifth case in the grouping condition 900 is used. Based on the reference point and area size information indicated in the input control list information 163, the vertical distance between the bottom end of the ninth input control and the top end of the tenth input control is 108 px. The arrangement position relationship between the tenth input control and the tenth input control is “the vertical distance between the lowest end of the nth input control and the highest end of the (n + 1) th input control, which is the condition of the fifth case in the grouping condition 900. Does not satisfy “less than 20 px”. As a result, the ninth input control and the tenth input control are regarded as different groups in the hierarchy level 2, that is, the group boundaries.

  As a result, the first to ninth input controls are regarded as one group in the hierarchy level 2, and the tenth input control is set as the head of the next group.

  In this way, grouping is determined at each hierarchical level as described above for all input controls that are adjacent in input order, and groups are generated. The grouping result is managed in a tree structure as shown in FIG. 8 having the node number, label, reference point, parent node number, and control type as attribute information of each node, and when the processing of the grouping processing unit 130 ends. A content structure tree 1000 is generated.

  In FIG. 8, the first to ninth input controls are nodes having node numbers 5 to 9 and node numbers 13 to 17, and a group in which these are bundled is a branch node and node number 2. In addition, the three branch nodes having node numbers 2 to 4 are bundled into a root node (node number 1) at a hierarchical level 1 representing the entire content. As the data structure of the content structure tree, for example, a pointer to a data storage area of a node one level higher in order from the leaf to the root is designated to form the tree structure. It is also possible to form a tree structure by designating a pointer to the data storage area of a node one level lower in order from the root to the leaf. A node can also have pointers in both the root-to-leaf direction and the leaf-to-root direction.

  As shown in FIG. 8, the labels and reference points that are the attribute information of each node of the content structure tree 1000 are not yet stored when the execution of the grouping processing unit 130 is completed.

  In this embodiment, the grouping is performed only for the input control. However, not only the input control, but generally, text, an image, or the like may be the target of the grouping process for the Web content.

  In this embodiment, the content configuration tree generation processing unit 123 activates the grouping processing unit 130, and the grouping processing unit 130 generates a group and the content configuration tree 1000. May only generate a group, and the content configuration tree generation processing unit 123 may generate the content configuration tree 1000.

  The label assignment processing unit 131 extracts labels and sets reference points in each node of the content configuration tree 1000 shown in FIG. 8 in order from the hierarchy level M (the lowest hierarchy level) to the hierarchy level 1.

  When the hierarchical level is a leaf node of the hierarchical level M (the lowest hierarchical level), the label assignment processing unit 131 sets the reference position of each input control as the upper left of the area, and then the text list information 162 and the input control list information 163. Is used to extract the text closest to the position of the nth input control as the label of the nth input control in the up or left direction (when the distance is the same, the top is given priority over the left). Then, the reference point of the nth input control including the label is reset.

  In the example of FIGS. 5 and 6, the reference point “X: 285 px Y: 196 px” of the third input control from the input control list information 163, and the reference point “X: 100 px Y of the sixth text from the text information 162. Since 192 px ”is closest in position, the sixth text“ Address ”is extracted as the label of the third input control, and the extraction result is the content configuration tree 1100 ( Is stored as attribute information in the node of node number 6 in FIG. Further, the reference position of the third input control (node number 6) including the label is reset as the reference position “X: 100px Y: 192px” of the sixth text, and the node number 6 of the content structure tree 1100 is set. It is stored as attribute information in the node.

  As described above, the same processing as described above is performed on all the input controls included in the content 500, so that the label name and the reference position of each node at the hierarchical level M (the lowest hierarchical level) Stored as attribute information of each node of the tree 1100.

  In this specific example, the label of the nth input control is selected from the text in the content stored in the text list information 162 in the upward direction and the left direction (when the distance is the same). Is the closest text to the left, but for radio buttons and check boxes, the most right and lower direction (if the distance is the same, the right priority is given to the bottom) The text may be a close distance.

  Next, when the hierarchy level is a branch node other than the hierarchy level M (the lowest layer level), the label assignment processing unit 131 uses the text list information 162 and the input control list information 163 as the reference point of the node number n. On the other hand, the text that is closest in position in the upward direction or the left direction (when the distance is the same, priority is given to the upper left) and has not yet been extracted as a label, is extracted as the label of the node number n and includes the label. The reference point of node number n is reset.

  In the example of FIG. 5 and FIG. 6, regarding the node number 2 of the content structure tree 1000 shown in FIG. 8, among the elements (child nodes) belonging to the node number 2, the upper left (when the distance is the same, the upper priority is given to the left) The reference point “X: 100 px Y: 158 px” of the node number 5 having the reference position is set as the reference position of the node number 2. The reference position of the text closest to the reference point of the node number 2 is the fourth text information “First” in the text list information 162, which is already used for the label of the node number: 5. Yes. Therefore, since the reference position “X: 90 px Y: 118 px” of the second text in the text list information 162 is not yet extracted as a label and is the closest text information element in position, The second text “Applicant Information” is extracted, and the extraction result is stored as attribute information in the node of node number 2 in the content structure tree 1100. Further, the reference position of the node number 2 including the label is reset as the reference position “X: 90 px Y: 118 px” of the label (second text), and the node of the node number 2 in the content configuration tree 1100 shown in FIG. Stored as attribute information.

  In this way, the same processing as described above is performed on all nodes other than the hierarchy level M (the lowest hierarchy level), so that the hierarchy level is other than the hierarchy level M (the lowest hierarchy level). The label and reference position of each node are stored as attribute information of each node of the content configuration tree 1100 (note that a part of the content configuration tree in FIG. 9 is not shown).

  In the grouping processing unit 130 in this specific example, the reference position information of each input control is used as the grouping method. However, the font size information of the text list information 162 shown in FIG. A method may be used in which a position of text having a font size of 2 is regarded as a group boundary between input controls at hierarchical level m. For example, in the example of FIGS. 5 and 6, when the grouping condition that the immediately preceding position of the text having the font size “+3” or more is regarded as the group boundary at the hierarchical level 2 is set, Since the second text “Applicant Information”, the 13th text “Elective Conference”, and the 26th text “Payment” have a font size of “+3”, the first and second of the input control list information 163, 9 The 10th, 10th, and 12th and 13th input controls are considered as group boundaries in the hierarchy level 2, and the 1st to 9th, 10th to 12th and 13th to 15th input controls are the hierarchy level. 2 is regarded as one group.

  The above is a specific processing flow of the component registration control processing unit 111.

  According to the first embodiment described above, the content is divided into a plurality of parts, and a plurality of existing part subtrees representing the plurality of parts are also stored in the repository (for example, the magnetic disk device 101), and each existing part subtree is also searched. It becomes the object of. As a result, it is possible to search for a component as content closer to the content that the user intends to create than to search only the content file itself, thereby reducing the burden on the user.

  Further, according to the first embodiment described above, the query tree and the existing part subtree include the input control type information element in addition to the text information element, and the input control type information element is not only the text information element. Based on this, the similarity between the query tree and the existing part subtree is calculated. As a result, compared to a search method in which only text is input as a search key, it is possible to search for a component as a content closer to the content that the user intends to create. Can be reduced.

  In addition, according to the first embodiment described above, it is possible to store existing component subtrees representing content components in the repository simply by registering content files in the repository, and to manually store the stored existing component subtrees. Even without adding meta information in advance, the developer can search for a desired content component.

  Further, according to the first embodiment described above, it is possible to unify the design of content (for example, an input form) even in the same computer system or different computer systems (for example, an electronic application system).

  <Second embodiment>.

  Next, a second embodiment of the present invention will be described. In the following, differences from the first embodiment will be mainly described, and description of common points with the first embodiment will be omitted or simplified.

  In the second embodiment, all existing parts subtrees registered in the repository are grouped by function and style before parts search (specifically, when parts are registered). Information on candidate parts is presented by style. As a result, information regarding a plurality of component candidates can be displayed in units of functions and styles.

  The difference from the first embodiment is that, as shown in FIG. 14, a function-specific grouping processing unit 127 and a style-specific grouping 128 are added to the component registration processing unit 126, a component search processing unit 142, and a component The processing of the candidate output processing unit 143 is partially changed (the component registration processing unit 126, the component search processing unit 142, and the component candidate output processing unit 143 after the change are replaced by the component registration processing unit 126c and the component (Referred to as group search processing unit 142c and component candidate output processing unit 143c).

  In the following, the processing performed in the second embodiment will be described mainly with respect to differences from the processing performed in the first embodiment.

  FIG. 15 shows an example of the flow of processing performed by the functional grouping processing unit 127.

  First, the functional grouping processing unit 127 repeats Step 2511 and Steps 2517 to 2519 for all content files 160 that have not been subjected to this processing registered in the magnetic disk device 101 or the work area 150 (Step 2511). 2510).

  Next, the functional grouping processing unit 127 repeats Steps 2512 to 2513 for all subtrees (existing part subtrees 174) included in the content configuration tree 161 corresponding to the content file 160 (Step 2511).

  Next, the functional grouping processing unit 127 acquires in-part information 177B (information including label information 175B and input control information 176B) of the existing part subtree, and stores the acquired in-part information 177B in the work area 150. (Step 2512).

  Next, the function-specific grouping processing unit 127 repeats steps 2514 to 2516 for all the function groups 188 (step 2513). A function group is a collection of existing component subtrees corresponding to components having the same or similar functions.

  Next, the function-specific grouping processing unit 127 acquires in-part information 177C (information including label information 175C and input control information 176C) of the function group 188, and the acquired in-part information 177C (see FIG. 19). Is stored in the work area 150 (step 2514).

  Next, the functional grouping processing unit 127 uses a table format representing a vector space and vectors for calculating similarity from the in-part information 177B of the existing part subtree 174 and the in-part information 177C of the function group 188. Vector information 182 is generated, and the generated vector information 182 is stored in the work area 150 (step 2515).

  Next, the function-specific grouping processing unit 127 uses the vector information 182 to calculate the similarity of the function group with respect to the existing component subtree using Equation 2. The similarity is calculated from the vector information 182 by the cosine scale (Formula 2) by the vector space method, and includes the similarity calculation result 189 (functional group ID information 1891 and similarity information 1892) illustrated in FIG. Information) is stored in the work area 150 (step 2516).

  Next, the function-specific grouping processing unit 127 determines whether or not there is a function group 188 having a similarity greater than or equal to a threshold for the existing subtree 174 (step 2517).

  If the result of determination in step 2517 is that there is a function group 188 having a similarity greater than or equal to the threshold, the function-specific grouping processing unit 127 sets the existing component subtree 174 as a member of the function group 188 having the highest similarity. The function group ID is stored in the node 195 (see FIG. 22) of the content structure tree 161 corresponding to the existing part subtree 174 (step 2518).

  If the result of determination in step 2517 is that there is no function group 188 having a similarity greater than or equal to the threshold, the function-specific grouping processing unit 127 generates a new function group whose members are the existing component subtree, and the function group ID (Step 2519), the existing component subtree is registered in the magnetic disk device 101 or the work area 150. The function group ID can be generated and assigned based on the rules stored in the main memory 102 or the magnetic disk device 101. Further, for one function group, the existing component subtree that is the last member of the function group can be used as a representative existing component subtree. The representative existing part subtree can be identified by including, for example, a code representing the representative (hereinafter, representative code) in the root node of the existing part subtree (the existing existing part subtree). If the subtree changes, the representative code can be deleted from the root node of the existing part subtree of the existing representative).

  FIG. 16 shows an example of the flow of processing performed by the style-specific grouping processing unit 128.

  First, the style-specific grouping processing unit 128 repeats Step 2611 for all function groups 188 registered in the magnetic disk device 101 or the work area 150 and not executing this processing (Step 2610).

  Next, the style-specific grouping processing unit 128 repeats Steps 2612 to 2613 for all the existing component subtrees included in the function group (Step 2611).

  Next, the style-specific grouping processing unit 128 acquires the style information of the part corresponding to the existing part subtree from the source of the content file 160 corresponding to the existing part subtree (step 2612). Store in work area 150. When the number of texts between parts to be compared for similarity is two or more, the style information 193 illustrated in FIG. 20 (the order of arrangement of two or more texts and the contents of the two or more texts (in FIG. Left is first)) is acquired. Further, when the number of texts between parts to be compared for similarity is 1, style information 194 (label information 1941, type information 1942, size information 1943, font-size information 1944, color information 1945, and maxlength illustrated in FIG. Information including information 1946) is acquired. The style information 193 and 194 can be constructed from attribute information or the like of an HTML source (content file), but may be constructed from other information elements that can be extracted from the source. The information element included in the style information 194 may be any information element that represents an attribute related to one text. Further, even if the number of texts between parts to be compared for similarity is two or more, style information 194 may be created for each text and the style information 194 may be compared.

  Next, the style grouping processing unit 128 repeats Steps 2614 to 2617 for all the style groups 190 (Step 2613). A style group is a collection of existing component subtrees having the same or similar style displayed on the screen.

  Next, the grouping processor by style 128 acquires the style information 193 or 194 from the style group, and stores the acquired style information 193 or 194 in the work area 150 (step 2614).

  Next, the style-specific grouping processing unit 128 compares the style information 193 or 194 of the existing part subtree with the style information 193 or 194 of the style group 190 and determines whether the style information completely matches (step 2615). ).

  As a result of the determination in step 2615, when the style information completely matches, the style grouping processing unit 128 makes the part a member of the style group 190 and the content structure tree 161 corresponding to the existing part subtree 174. The style group ID is stored in the node 195 (see FIG. 22) (step 2516).

  As a result of the determination in step 2615, if the style information does not completely match, the style-specific grouping processing unit 128 generates a new style group having the existing component subtree as a member, and assigns the style group ID. (Step 2617), the data is registered in the magnetic disk device 101 or work area 150. The style group ID can be assigned by generating a function group ID based on the rules stored in the main memory 102 or the magnetic disk device 101. In addition, the existing component subtree that is the last member for one style group can be used as a representative existing component subtree.

  In the similarity determination process in step 2615, it is assumed that the existing part subtree and the style information of the style group 190 are not similar if they do not completely match each other. May be considered similar to absorb. Specifically, for example, in the comparison between the style information 193, if the text is completely matched, the arrangement order may be different, or the existing part subtree to be compared may be a member of the style group to be compared. Alternatively, if the arrangement order is completely the same, even if the predetermined number of texts are different, the existing part subtree to be compared may be a member of the style group to be compared. For example, in the comparison between the style information 194, if the information elements of a predetermined ratio or more match each other, the existing part subtree to be compared may be a member of the style group to be compared.

  Note that the processing results of the function-specific grouping processing unit 127 and the style-specific grouping processing unit 128 are the existing component grouping information 196 illustrated in FIG. 22, that is, the function group ID information 1961, the style group ID information 1962, and the component ID information 1963 ( Information including the ID of a component corresponding to an existing component subtree included in the style group and the appearance number information 1964 (the number of existing component subtrees included in the style group) may be used. The existing part grouping information 196 may be stored in the magnetic disk device 101 or the work area 150.

  FIG. 17 shows an example of the flow of processing performed by the component group search processing unit 142c.

  First, the component group search processing unit 142c repeats steps 2710 to 2711 for all the query trees 171 acquired by the input / output definition information analysis processing unit 141 (step 2700).

  Next, the part group search processing unit 142c acquires the in-part information 177A of the query tree 171 and stores the acquired in-part information 177A in the work area 150 (step 2710).

  Next, the parts group search processing unit 142 c repeats steps 2712 to 2713 for all the function groups 188 registered in the magnetic disk device 101 or the work area 150.

  Next, the part group search processing unit 142c acquires the in-part information 177C of the function group 188, and stores the acquired in-part information 177C in the work area 150 (step 2712).

  Lastly, the part group search processing unit 142c calculates the similarity of the function group 188 with respect to the query tree 171, and calculates the similarity by function group calculation result 189 (function group ID information 1891 and similarity information) illustrated in FIG. (Information including 1892) is stored in the work area 150 (step 2713). In step 2713, substantially the same processing as in step 2516 in the processing flow of the function-specific grouping processing unit 127 can be performed (for example, the similarity is calculated by (Expression 2)).

  The component candidate output processing unit 143c includes the function group similarity calculation processing result 189 generated by the component group search processing unit 142c, and the existing component grouping information generated by the function grouping processing unit 127 and the style grouping processing unit 128. 196 can be used to display a list of information related to the function group 188 for the query tree 171 and information related to the style groups in the function group 188 (hereinafter, a part group candidate list) on the part group candidate list display screen 197. . This screen 197 is, for example, a screen prepared instead of the component candidate list display screen 187 in the first embodiment.

  The part group candidate list includes the function group ID, the similarity of the function group 188 to the query tree 171, the ID of the sub-function group included in the function group, the ID of the style group included in the function group 188, and the style group. The number of parts included in is included. In addition, a preview button for previewing a representative part of each style group is displayed. The ID of the sub function group included in the function group is the function group ID recorded in the node 195 in the existing part subtree 174 included in the function group. That is, if the ID of the second function group different from the first function group is recorded in the node in the existing component subtree belonging to the first function group, the second function group is This is a sub-function group of the function group.

  Hereinafter, specific processing of the component registration processing unit 126c will be described with reference to FIGS.

  FIG. 18 is an explanatory diagram of specific processing of the functional grouping processing unit 127.

  The functional grouping processing unit 127 calculates the similarity with the registered function group 188 for all the existing component subtrees 174 included in all the content files 160 registered in the magnetic disk device 101. The functional group to which the part corresponding to each existing part subtree 174 belongs is determined.

  In the example of FIG. 18, in order to determine which functional group of the three function groups 188 (ID “001” to “003”) the existing part subtree 174 to be grouped belongs to. The similarity of each function group 188 is calculated. Specifically, the similarity “0.50” is calculated for the function group ID “001”, the similarity “0.67” is calculated for the function group ID “002”, and the function group ID “003”. The similarity “0.42” is calculated, and the function group ID information 1891 and the similarity information 1892 are stored in the function group similarity calculation result 189, respectively. As a result, for example, when the similarity threshold is “0.50”, the existing component subtree 174 is added to the member of the function group with ID “002” having the highest similarity. Further, when the similarity threshold is “0.70”, it is assumed that the parts in the existing part subtree 174 do not belong to any function group, and a new function group (a member of the parts in the existing part subtree 174 as members) ID “004”) is generated.

  FIG. 19 is an explanatory diagram of specific processing of similarity calculation in the functional grouping processing unit 127.

  First, the functional grouping processing unit 127 acquires the in-part information 177B from the existing part subtree 174, and acquires the in-part information 177C from the function group 188. In the example of FIG. 19, the in-part information 177B has the same contents as the in-part information 177A in FIG. 12, and the in-part information 177C has the same contents as the in-part information 177B in FIG. For this reason, the similarity calculated using the above (Formula 2) with reference to the vector information 182 created based on the in-part information 177B and 177C is 0.67. The functional grouping processing unit 127 records the functional group ID information 1891 “002” and the similarity information 1892 “0.67” in the functional group similarity calculation result 189.

  In the similarity calculation between the existing part subtree 174 to be grouped and the function group 188, the representative existing part subtree in the function group 188 and the existing part subtree 174 to be grouped may be compared. Alternatively, the grouping target existing part subtree 174 is compared with two or more (for example, all) existing part subtrees 174 in the function group 188, and the calculated average of the two or more similarities is the grouping target existing part subtree 174. And the function group 188 may be similar to each other.

  As described above, the function group and the existing part subtree 174 to be grouped

  FIG. 20 is an explanatory diagram of specific processing of the style-specific grouping processing unit 128 when the number of texts between parts to be compared for similarity is two or more.

  First, the style-specific grouping processing unit 128 acquires the style information 193 of the existing part subtree to be grouped from the content file 160 (source). The style information 193 is information expressing the text (for example, input item names) and the order. In the example of this figure, three input item names such as “name”, “address”, and “phone number” are “name”. , “Address”, “phone number” in the order of content.

  Next, the style grouping processing unit 128 also acquires the style information 193 from the representative existing part tree of the style group 190 for the style group 190. In the example of this figure, “sex” name address telephone number ”is acquired as style information 193 for style group ID“ 001 ”, and“ name telephone number address ”is used as style information 193 for style group ID“ 002 ”. Is acquired, and “name, address, telephone number” is acquired as the style information 193 for the style group ID “003”.

  Next, the style-specific grouping processing unit 128 compares the style information 193 of the grouping target existing part subtree 174 with the style information 193 of each style group 190 to determine whether or not they are completely matched. In the example of this figure, among the three style groups 190, the input item names are different for the style group ID “001”, the order of the input item names is different for the style group ID “002”, and the style group ID “003”. Only the style information 193 corresponding to is completely matched with the style information 193 of the existing part subtree 174 to be grouped. Therefore, the style-specific similarity determination result 192 that is information indicating the determination result indicates that the existing part subtree 174 to be grouped becomes a member of the style group 190 with the style group ID “003”.

  FIG. 21 is an explanatory diagram of specific processing of the style-specific grouping processing unit 128 when the number of texts between parts to be compared for similarity is 1.

  First, the style-specific grouping processing unit 128 acquires the style information 194 of the existing part tree 174 to be grouped from the content file 160 (source). The style information 194 includes information elements indicating attributes relating to the style of one input item name (text). In the example of this figure, label information 1941 “name”, type information 1942 “text”, and size information 1943 “ 10 ”, font-style information 1944“ normal ”, font-size information 1945“ medium ”, color information 1946“ black ”, and maxlength information 1947“ 10 ”are included.

  The style grouping processing unit 128 also acquires the style information 194 from the representative existing component subtree of the style group 190 for each style group 190. In the example of this figure, the style information 194 corresponding to the style group ID “001” includes label information 1941 “name”, type information 1942 “text”, size information 1943 “10”, and font-style information 1944 “normal”. , Font-size information 1945 “medium”, color information 1946 “black”, and maxlength information 1947 “10”. Hereinafter, the style information 194 is similarly acquired for the style group IDs “002” and “003”.

  Next, the grouping processing unit 128 by style compares the style information 194 of the existing part subtree 174 to be grouped with the style information 194 of the style group 190 and determines whether or not they completely match each other. In the example of this figure, the style group ID “001” has different label information 1941, the style group ID “002” has different font-size information 1944, and only the style information 194 of the style group ID “003” is the grouping target. The part style information 194 matches all the information. The style-specific grouping processing unit 128 detects this and records in the style-specific similarity determination result 192 that the existing part tree 174 to be grouped becomes a member of the style group with ID “003”.

  Although not shown in this figure, if the style information of the existing part subtree to be grouped does not match the style information 193 or 194 of any style group, the existing part subtree to be grouped is a member. A new style group can be generated, and a style group ID can be assigned and registered in the magnetic disk device 101 or the work area 150. In this case, the existing part subtree to be grouped becomes the representative existing part subtree of the newly generated style group.

  Further, the grouping results of the function-specific grouping processing unit 127 and the style-specific grouping processing unit 128 are stored as a function group ID and a style group ID in the node 195 of the content configuration tree 161 corresponding to the existing part subtree to be grouped. . In the example of FIG. 22, the function group ID “002” and the style group ID “003” are stored in the node 195 of the content configuration tree 161, respectively.

  Further, as described above, the grouping results of the functional grouping processing unit 127 and the style-specific grouping processing unit 128 are obtained from the existing component grouping information 196 (functional group) in order to improve the information acquisition performance in the component candidate output processing unit 143c. ID information 1961, style group ID information 1962, component ID information 1963, and information including appearance number information 1964) may be stored in the magnetic disk device 101 or the work area 150. In the example of this figure, as the second record information of the existing component grouping information 196, function group ID information 1961 “001”, style group ID information 1962 “002”, component ID information 1963 “0031_3, 0065_2” and appearance number information 1964 “2” is stored. In addition, the component ID stored at the top of the component ID information 1963 (indicated by an underline) indicates that it is a representative existing component subtree in the style group. As a method of indicating the representative existing component subtree, other methods such as an underline can be adopted instead of storing the component ID corresponding to the representative existing component subtree at the head.

  The above is the specific processing of the component registration processing unit 126c.

  The component search processing unit 142 calculates the similarity to all the functional groups 188 registered in the magnetic disk device 101 or the work area 150 with respect to the query tree 171 specified by the user. For the similarity calculation of the function group 188 with respect to the query tree 171, the process of acquiring the in-part information 177 B from the grouping target existing part subtree illustrated in FIG. 19 is changed to the process of acquiring the in-part information 177 A from the query tree 171. Except for the above, the processing example shown in FIG. 19 is the same.

  FIG. 23 is an explanatory diagram of specific processing of the component candidate output processing unit 143c.

  The component candidate output processing unit 143c includes a function group similarity calculation result 189 generated by the component group search processing unit 142c, and existing component grouping information 196 generated by the function grouping processing unit 127 and style grouping processing unit 128. Then, a list of information regarding the function group 188 for the query tree 171 and information regarding the style group 190 within the function group 188 is displayed on the part group candidate list display screen 197. In the example of FIG. 23, as the first function group, the function group ID “002”, the similarity “0.67” of the function group 188 with respect to the query tree 171, and the ID “004” of the sub function group included in the function group. 005 "is displayed. The sub function group ID is a function group ID other than the ID of the function group in the node 195 in the existing component subtree 174 included in the function group. In addition, style group IDs “001”, “002”, and “003” are displayed as a list of style groups included in the first functional group (ID: 002), and are included in each style group. “30”, “2”, and “1” are respectively displayed as the number of parts (number of appearances) to be displayed.

  In the example of this figure, a preview button is provided for previewing the part represented by the representative existing part subtree of each style group. When a certain preview button is pressed, the component candidate output processing unit 143c acquires the code represented by the representative existing component subtree in the style group corresponding to the preview button from the source, and based on the acquired code, the representative existing An image of a part corresponding to the part subtree can be displayed.

  The above is the description of the second embodiment of the present invention.

  According to the second embodiment described above, by grouping the existing part subtree for each function and style before the search, the part candidates for the query tree can have different functions without duplication in terms of function and style. It is possible to display a list for each part of style and style. As a result, the user can efficiently search for a desired component from the component candidate list display.

  <Third embodiment>.

  Next, a third embodiment of the present invention will be described.

  In the third embodiment, after narrowing a set of existing component subtrees to some extent by the component search method in the first embodiment, each existing component subtree in the set is classified according to the function described in the second embodiment. And grouping processing by style is performed.

  This embodiment is different from the first embodiment in that, as shown in FIG. 24, the component candidate output processing unit 143 includes a functional grouping processing unit 127, a style grouping 128, a component group search processing unit 142c, and a component candidate. An output processing unit 143c is added (the component candidate output processing unit 143 after the change is referred to as a component candidate output control processing unit 144).

  Hereinafter, processing different from the first embodiment and the second embodiment will be mainly described.

  The component candidate output control processing unit 144 first compares the similarity of each existing component subtree calculated by the component search processing unit 142 with the similarity threshold stored in the magnetic disk device 101 or the work area 150. Then, an existing part subtree having a similarity greater than or equal to the threshold is extracted. Hereinafter, the set of extracted existing component subtrees is referred to as a “primary search result component set”.

  Next, the component candidate output control processing unit 144 executes the function-specific grouping processing unit 127, and the function-specific grouping processing unit 127 performs grouping by function for each existing component subtree in the primary search result component set. The processing procedure of the functional grouping processing unit 127 is the same as that of the second embodiment, except that the processing target is an existing part subtree in the primary search result part set.

  Next, the component candidate output control processing unit 144 executes the style-specific grouping processing unit 128, and the style-specific grouping processing unit 128 groups all the registered function groups 188 by style. The processing procedure of the style grouping processing unit 128 is the same as that of the second embodiment except that the processing target is each function group 188 grouped by function with respect to the primary search result component set.

  Next, the part group search processing unit 142c is executed, and the similarity with the query subtree 171 is calculated for all the function groups 188 registered in step 3410. The processing procedure of the component group search processing unit 142c is the same as that of the second embodiment except that the processing target is each function group 188 grouped by function with respect to the primary search result component set.

  Next, the component candidate output control processing unit 144 executes the component candidate output processing unit 143c, and the component candidate output processing unit 143c performs the generated function group similarity calculation result 189, the function grouping processing unit 127, and the like. A part group candidate list for the query tree 171 is displayed on the part group candidate list display screen 197 from the existing part grouping information 196 generated by the style grouping processing unit 128. The processing procedure of the part candidate output processing unit 143c is the same as that of the second embodiment except that the part candidate output target is the primary search result part set.

  The processing performed in the third embodiment is substantially the same as in the second embodiment, except that the processing target for grouping by function and style is the primary search result component set.

  As described above, according to the third embodiment, the processing target of grouping by function and style is newly created by making a part set narrowed down to some extent by the part search method according to the first embodiment. Only parts groups used in the content field can be output. As a result, the user can efficiently search for a desired component from the component group candidate list.

  <Fourth embodiment>.

In the fourth embodiment, the similarity of the text and the similarity of the input control are calculated separately, and the similarity between the query tree and the existing part subtree (or function group) is calculated based on the similarity. be able to. Specifically, for example, the following (formula 3)
Similarity between query tree and existing parts subtree (or function group) = α (text similarity) + β (input control similarity) (Equation 3)
Can be calculated. α and β can be obtained using a cosine scale, respectively.

  For example, there are five texts (input item names) in the query tree, such as name, address, telephone number, and e-mail address, and there are five text boxes, checklists, radio buttons, and pull-downs as eight input controls. . On the other hand, it is assumed that the existing part subtree to be compared includes name, sex, name, mail address, occupation, address, and telephone number as seven texts, and five text boxes and pull-down as six input controls. For this reason, between the query tree and the existing part subtree to be compared, the name, address, telephone number, and e-mail address match for the text, and the five text boxes and pull-downs for the input control. Match.

  Of the eight sets of text and input controls, the query tree has three matching sets, the comparison target existing part subtree has seven matching sets, and the query tree and the comparison target existing part subtree. , There are three sets that match each other.

In this case, according to the first to third embodiments, the similarity is calculated as 0.57. Because, in the cosine scale ((Equation 1) and (Equation 2)), the numerator is 1 ² +1 ² +1 ² = 3, and the denominator is {square root of ((1 ² +1 ² +1 ² +1 ² )} × {( This is because the square root of 1 ² +1 ² +1 ² +1 ² +1 ² +1 ² +1 ² )}.

However, in the fourth embodiment, the similarity is obtained using the above (Equation 3). Specifically, for example, α is 0.76 and β is 0.96. Therefore, the similarity is 1.72. The reason for α being 0.76 is that, on the cosine scale, the numerator is 1 ² +1 ² +1 ² +1 ² = 4, and the denominator is the square root of {(1 ² +1 ² +1 ² +1 ² )} × {(1 ² + 1 ² +1 ² +1 ² +1 ² +1 ² +1 ² )}. The reason for β being 0.96 is that, on the cosine scale, the numerator is 5 ² +1 ² = 26, and the denominator is {square root of (5 ² +1 ² +1 ² +1 ² )} × {(5 ² +1 ² ) Because the square root of.

  Note that α and β are calculated using a cosine scale, but may be calculated using other methods.

  <Fifth embodiment>.

  The fifth embodiment is a combination of the second and third embodiments, and a function for accepting an input indicating whether registration performance is important or search performance is important is provided in the content component search system. In the content parts search system, when a developer inputs importance on registration performance, as described in the third embodiment, a grouping process for each function and style is performed on the primary search result parts set. If importance is placed on search performance from the developer, as described in the second embodiment, for each existing part subtree representing each part of newly registered content, by function and style Another grouping process is performed.

  As mentioned above, although several embodiment of this invention was described, these embodiment is only the illustration for description of this invention, and is not the meaning which limits the scope of the present invention only to those embodiment. The present invention can be implemented in various other modes without departing from the gist thereof. For example, the storage destination of the existing component subtree and the evaluation result is not limited to the main memory or the magnetic disk device, but may be other types of storage resources.

FIG. 1 shows an example of the overall configuration of a content parts search system according to the first embodiment of the present invention. FIG. 2 shows an example of the flow of processing performed by the grouping processing unit. FIG. 3 shows an example of the flow of processing performed by the label addition processing unit. FIG. 4 shows an example of content display. FIG. 5 shows an example of text list information. FIG. 6 shows an example of input control information. FIG. 7 shows an example of the grouping condition list. FIG. 8 shows an example of a content configuration tree before labeling by the labeling processing unit. FIG. 9 shows an example of a content structure tree after labeling by the labeling processing unit. FIG. 10 shows an example of the flow of processing performed by the component search processing unit. FIG. 11 is an explanatory diagram of obtaining a query tree from the input / output definition information. FIG. 12 is an explanatory diagram of an example of processing performed by the component search processing unit. FIG. 13 is an explanatory diagram of an example of processing performed by the component candidate output processing unit. FIG. 14 shows a configuration example of the component registration processing unit and the component search control processing unit in the second embodiment of the present invention. FIG. 15 shows an example of the flow of processing performed by the functional grouping processing unit. FIG. 16 shows an example of the flow of processing performed by the style-specific grouping processing unit. FIG. 17 shows an example of the flow of processing performed by the component group search processing unit in the second embodiment of the present invention. FIG. 18 is an explanatory diagram of an outline of processing performed by the functional grouping processing unit. FIG. 19 is an explanatory diagram of similarity calculation processing of the functional grouping processing unit. FIG. 20 is an explanatory diagram of specific processing of the grouping processing unit by style when the number of texts between parts to be compared for similarity is two or more. It is. FIG. 21 is an explanatory diagram of specific processing of the grouping processing unit by style when the number of texts between parts to be compared for similarity is 1. FIG. 22 shows an example of nodes in the content structure tree and existing part grouping information. FIG. 23 is an explanatory diagram of processing performed by the component candidate output processing unit according to the second embodiment of the present invention. FIG. 24 shows a configuration example of the component search control processing unit in the third embodiment of the present invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 ... Central processing unit (CPU) 101 ... Magnetic disk apparatus 102 ... Main memory 103 ... Floppy disk drive (FDD) 104 ... Bus 105 ... Network 106 ... Floppy disk 110 ... System control processing part 111 ... Component registration control processing part 112 ... part search control processing part 112c ... part search control processing part 121 ... content acquisition processing part 122 ... content element information acquisition processing part 123 ... content configuration tree generation processing part 126 ... part registration processing part 126c ... part registration processing part 127 ... Grouping processing unit by function 128 ... Grouping processing unit by style 130 ... Grouping processing unit 131 ... Label assignment processing unit 140 ... Input / output definition information acquisition processing unit 141 ... Input / output definition information analysis processing unit 142 ... Component search processing unit 142c ... Part group search processing unit 143 ... Part candidate output processing unit 143c ... Part candidate output processing unit 144 ... Part candidate output control processing unit 150 ... Work area 160 ... Content file 161 ... Content configuration tree

Claims (4)

In an apparatus for searching for a content part as a part of the web content for creating the web content,
A search condition input unit that receives an input of a search condition that includes a combination of input control type information indicating the type of text information and input control that represents the text,
A plurality of information packages including a plurality of content parts composed of combinations of text corresponding to the text information and input controls corresponding to the input control type information, and including a plurality of combinations of the text information and the input control type information are stored. A storage unit;
The information package searched from the storage unit using text information and input control type information included in the input search condition, and the similarity of the information package searched by the search condition to the search condition A similarity calculation unit for calculating
A content component search apparatus comprising: a component candidate display unit that displays a plurality of the information packages searched under the search condition on a display screen based on each of the calculated similarities. The similarity, a first similarity between the at least one text information is included in at least one of the information package and the text information contained in the search condition, it included in the retrieval condition and has at least one input control type information as a value calculated on the basis of the second degree of similarity between at least one input control type information is included in the information package,
The content parts retrieval apparatus according to claim 1.   In a method in an apparatus for searching for a content part as a part of a web content for creating the web content,
  Accepts search condition input that includes a combination of text information representing text and input control type information representing the type of input control,
  A plurality of information packages including a plurality of content parts including a combination of text corresponding to the text information and an input control corresponding to the input control type information and a plurality of combinations of the text information and the input control type information are stored. Searching the information package from the storage unit using text information and input control type information included in the input search condition, and the similarity of the information package searched by the search condition to the search condition To calculate
  Displaying a plurality of the information packages searched under the search condition on a display screen based on the calculated similarity,
Content part retrieval method characterized by the above.   In a program for causing a computer to function as an apparatus for searching for a content part as a part of the Web content for creating the Web content,
  A search condition step for accepting input of a search condition including a combination of text information representing text and input control type information representing a type of input control;
  A plurality of information packages including a plurality of content parts composed of combinations of text corresponding to the text information and input controls corresponding to the input control type information, and including a plurality of combinations of the text information and the input control type information are stored. Searching the information package from the storage unit using text information and input control type information included in the input search condition, and the similarity of the information package searched by the search condition to the search condition A similarity calculation step for calculating
  A candidate component display step for displaying a plurality of the information packages searched under the search condition on a display screen based on the calculated similarities;
A content parts search program for causing the computer to execute.