JP5870776B2 - Link granting device, link granting method, and program - Google Patents

Description

  The present invention relates to a link providing device that gives a link destination to an electronic book, a web page, or the like.

  Conventionally, there is a technique for linking information related to content to the content.

  For example, in Patent Literature 1, the link destination hypertext of the original anchor existing in the processing range in the original hypertext is acquired, the link destination anchor information is extracted from the link destination hypertext, and the effect of the original anchor on the original hypertext There has been proposed a technique for generating new anchor information that matches a character string existing in a range and a character string existing in link destination anchor information, and adding a new anchor.

  For example, in Patent Document 2, a verb and a noun that is a case element are extracted from a text as a key phrase, a link is created where a key phrase is common, and an important portion is extracted based on the expression of the text. Techniques have been proposed for determining the direction.

  For example, in Patent Document 3, a vocabulary of interest is detected based on designation of a display position on a Web page displayed on a display screen, and further peripheral vocabularies around the vocabulary of interest are detected. Techniques for searching Web pages using peripheral vocabulary have been proposed.

Japanese Patent No. 2008-129998 JP-A-8-329108 JP 2009-129098 A

  In Patent Document 1, when the character string notation is ambiguous, there is a problem that it is difficult to assign an appropriate link. In patent document 2 and patent document 3, since the appearance frequency of the extracted word is not taken into consideration, there is a case where it is influenced by a specific word and there is a problem that an incorrect link is given.

  The present invention has been made in view of the above-described problems, and the object of the present invention is to provide a link providing device capable of assigning an appropriate link destination to a keyword appearing in an electronic book or a web page. Is to provide.

In order to achieve the above-mentioned object, the first invention extracts a keyword from a link destination document and a link source document using the keyword list creation means for creating a keyword list from the headings of the link destination document and the keyword list. A keyword extracting unit that selects one synonym of the same notation from the keywords of the link source document and determines a selected word; and the selected word, peripheral words, and The appearance frequency of the keyword included in the linked document with respect to the peripheral word is acquired, and the appearance frequency of the keyword included in the keyword list is acquired in each linked document candidate, and the selected word and each linked document candidate are A similarity calculation means for calculating the similarity of the link destination document candidate having the maximum similarity and a link destination sentence for the selected word A selection word destination document determination unit that determines as a link application device, characterized in that it comprises a.
According to the first invention, not only the peripheral word but also the keyword included in the linked document with respect to the peripheral word is used for calculating the similarity, so that the influence of the specific word is reduced and the association accuracy is improved.

The similarity calculation means excludes the peripheral words whose link destination documents are not uniquely determined from the targets of similarity calculation processing.
As a result, it is possible to prevent noise information from being mixed into the similarity calculation process, thereby improving the accuracy.

The selected word determination means selects a unique word prior to the same notation among keywords of the link source document, and the selected word link destination document determination means follows the selection by the selection word determination means, The link destination document for the unique word is determined prior to the link destination document for the same notation synonym.
Thereby, in the similarity calculation process for the same notation different words, it is possible to increase the number of peripheral words to be processed, and thus the accuracy is improved.

2nd invention is the link provision method which a link provision apparatus performs, Comprising: The control part of the said link provision apparatus creates the keyword list from the headline of a link destination document, The said control part, The keyword extraction step of extracting a keyword from the link destination document and the link source document using the keyword list, and the control unit selects one synonym of the same notation from the keywords of the link source document, and the selected word The selected word determining step for determining the keyword, and the control unit obtains the selected word, the peripheral word, and the appearance frequency of the keyword included in the linked document for the peripheral word in the link source document, and each linked document In the candidate, the appearance frequency of the keyword included in the keyword list is acquired, and the selected word and each linked document candidate A similarity calculation step of calculating the similarity, the control unit, the similarity is the maximum landing document candidates, include a selection word destination document determining step of determining as a link destination document for the selected word A link providing method characterized by the above.
According to the second invention, not only the peripheral word but also the keyword included in the linked document with respect to the peripheral word is used for calculating the similarity, so that the influence of the specific word is reduced, and the association accuracy is improved.

According to a third aspect of the present invention, there is provided a keyword list creating means for creating a keyword list from a heading of a linked document, a keyword extracting means for extracting a keyword from a linked document and a linked document using the keyword list, and the linked source Included in the selected word, a neighboring word, and a linked document for the neighboring word in the linked source document, by selecting one synonym of the same notation from the keywords of the document and determining the selected word The similarity calculation is performed to acquire the appearance frequency of the keyword included in the keyword list and calculate the similarity between the selected word and each link destination document candidate. And a selected word link destination for determining a linked document candidate having the maximum similarity as a linked document for the selected word And writing determination means, a program for causing a computer to function as a link applying device having a.
By installing the program according to the third invention in the computer, the link providing device according to the first invention can be obtained.

  According to the present invention, it is possible to provide a link assigning device, a link assigning method, and a program capable of assigning an appropriate link destination to a keyword appearing in an electronic book or a web page.

It is a figure which shows the structural example of the link provision apparatus which concerns on embodiment of this invention. It is a figure which shows the database memorize | stored in the memory | storage part of a link provision apparatus. It is a figure which shows an example of the information memorize | stored in linked document DB. It is a figure which shows an example of the information memorize | stored in link original document DB. It is a figure which shows an example of a keyword list file. It is a figure which shows an example of the information memorize | stored in link information DB of a selection word and a link destination document. It is a figure which shows an example of the information memorize | stored in link source document DB after link provision. It is a flowchart explaining a link provision process. It is a flowchart explaining the detail of the keyword extraction process of step S2 of FIG. It is a figure for demonstrating the specific processing content of FIG. It is a flowchart explaining the detail of the related process of the same notation different term of step S8 of FIG. It is a figure for demonstrating the specific processing content of FIG. It is a figure which shows the structural example of the electronic book provision system which provides the electronic book after link provision. It is a figure which shows an example of the evaluation result of the method (Example) of this invention, and the conventional method (comparative example). It is a figure which shows an example of the evaluation result at the time of calculating the number of correct answers for every number of surrounding words.

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

[Embodiments of the present invention]
FIG. 1 is a diagram illustrating a configuration example of a link providing device 1 according to an embodiment of the present invention. Note that the configuration in FIG. 1 is an example, and various configurations can be adopted depending on the application and purpose.

  The computer that realizes the link providing device 1 includes a control unit 11, a storage unit 12, a media input / output unit 13, a communication control unit 14, an input unit 15, a display unit 16, a peripheral device I / F (interface) unit 17, and the like. Connection is made via a bus 18.

  The control unit 11 includes a CPU, a ROM, a RAM, and the like. The CPU calls a program stored in the storage unit 12, ROM, recording medium or the like to a work memory area on the RAM, executes it, controls the drive of each device connected via the bus 18, and the link providing device 1 The process to be described later is realized. The ROM is a non-volatile memory, and permanently stores programs such as computer boot programs, BIOS (Basic Input / Output System), data, and the like. The RAM is a volatile memory, and temporarily stores a program, data, and the like loaded from the storage unit 12, ROM, recording medium, and the like, and includes a work area used by the control unit 11 to perform various processes.

  The storage unit 12 is an HDD, and stores a program executed by the control unit 11, data necessary for program execution, an OS (operating system), and the like. With respect to the program, a control program corresponding to an OS (operating system) and an application program for causing a computer to execute processing described later are stored. Each of these program codes is read by the control unit 11 as necessary, transferred to the RAM, read by the CPU, and executed as various means.

  The media input / output unit 13 (drive device) inputs / outputs data, for example, media such as a CD drive (-ROM, -R, -RW, etc.), DVD drive (-ROM, -R, -RW, etc.) Has input / output devices. The communication control unit 14 includes a communication control device, a communication port, and the like, and is a communication interface that mediates communication between a computer and a network, and performs communication control between other computers via the network.

  The input unit 15 inputs data and includes, for example, a keyboard, a pointing device such as a mouse, and an input device such as a numeric keypad. An operation instruction, an operation instruction, data input, and the like can be performed on the computer via the input unit 15.

  The display unit 16 includes a display device such as a liquid crystal panel, and a logic circuit or the like (video adapter or the like) for realizing a video function of the computer in cooperation with the display device.

The peripheral device I / F unit 17 is a port for connecting a peripheral device to the computer, and the computer transmits and receives data to and from the peripheral device via the peripheral device I / F unit 17. The peripheral device I / F unit 17 is a USB (Universal Serial Bus) or IEEE (The Institute of Electrical and
Electronics Engineers) 1394, RS (Recommended Standard) -232C, etc., and usually has a plurality of peripheral devices I / F. The connection form with the peripheral device may be wired or wireless. The bus 18 is a path that mediates transmission / reception of control signals, data signals, and the like between the devices.

  The link assigning device 1 of the present invention stores information related to a link source document and a link destination document in a database in advance, and includes keywords, peripheral words, and keywords included in the link destination document for the peripheral words selected in the electronic book by the user. Is compared with the appearance frequency of the keyword in each link destination document candidate, the link destination document candidate that maximizes the degree of similarity is determined, and can be assigned to the keyword.

  Hereinafter, terms will be defined in the description of the present embodiment. The “selected word” is a character string that is a link source for providing a link among keywords included in an electronic book or a web page. A “peripheral word” is a character string existing around a selected word of an electronic book or a web page. A “same synonym” is a character string that cannot uniquely identify a phrase, that is, has a different meaning or content. A “unique word” is a character string that can uniquely identify a phrase.

  FIG. 2 is a diagram illustrating a database stored in the storage unit 12 of the link providing device 1. As shown in FIG. 2, the storage unit 12 of the link assigning apparatus 1 includes a link destination document database (DB) 21, a link source document database (DB) 22, a keyword list file 23, and a link between the selected word and the link destination document. An information database (DB) 24 and a post-link link source document database (DB) 25 are stored.

  For example, as shown in FIG. 3, the linked document DB 21 stores at least a normal heading 31, a unique heading 32, document contents 33, and a document location 34 as linked document information for each No.

  The normal headline 31 indicates a word (keyword) that summarizes the main points of the contents of a linked document (for example, an online dictionary or Wikipedia). The unique heading 32 is a word for uniquely identifying the normal heading 31 having the same notation. The document content 33 indicates the content of the link destination document. The document location 34 indicates the location (address) of the linked document.

  In the case of the example of FIG. 3, the linked document information of No. 1 includes a normal heading 31 of “Hashiba Hideyoshi”, a unique heading 32 of “Hashiba Hideyoshi”, a document content 33 of “From the Sengoku period ...”, “ ... / hashibahideyoshi.txt "is linked to the document location 34. The linked document information of No. 2 includes a normal heading 31 of “Takamatsu Castle”, a unique heading 32 of “Takamatsu Castle (Kagawa Prefecture)”, a document content 33 of “To Takamatsu ... in Kagawa Prefecture”, “・ The document location 34 of "/takamatsujyo-okayamaken.txt" is linked. The linked document information of No. 3 includes a normal heading 31 of “Takamatsu Castle”, a unique heading 32 of “Takamatsu Castle (Okayama)”, a document content 33 of “Bi-Chinese Takamatsu ...”, “・ The document location 34 of "/takamatsujyo-okayamaken.txt" is linked. The link destination document information of No. 4 includes a normal headline 31 of “Marugame Kaoru”, a unique heading 32 of “Marugame Kaoru”, a document content 33 of “...”, and a document of “... / marugamehan.txt” Location 34 is associated. The link destination document information of No. 5 includes a normal headline 31 of “Tokugawa Tsunakata”, a unique heading 32 of “Tokugawa Tsunakata”, a document content 33 of “...”, “... / tokugawatsunakata.txt”. The document locations 34 are associated with each other. The linked document information of No. 6 includes a normal headline 31 of “Honnoji no Hen”, a unique heading 32 of “Honnoji Hen”, a document content 33 of “...”, “... / honnnoujinohen.txt” The document location 34 of No. 7 is linked, and the link destination document information of No. 7 includes a normal headline 31 of “Thailand”, a unique headline 32 of “Thailand (country name)”, a document content 33 of “. ... / tai-kunimei.txt "is linked to the document location 34.

  For example, as shown in FIG. 4, the link source document DB 22 stores at least the document name 41 and the document content 42 as link source document information for each No.

  The document name 41 indicates the name of the link source document (electronic book, web page, etc.). The document content 42 indicates the content of a sentence (electronic book, web page, etc.).

  In the case of the example in FIG. 4, the document document number 41 of “History of Kagawa Prefecture” is included in the No. 1 link source document information, “Marugame Kaoru participates in Kintohira Shrine… Hideyoshi Hashiba is Nagahama's governing policy .. ”Is linked. The link source document information of No. 2 is associated with a document name 41 of “Sengoku Busho Nobunaga Oda” and a document content 42 of “Nobunaga must follow its style ... .

  For example, as shown in FIG. 5, the keyword list file 23 is a list in which normal headings 31 of the link destination document DB 21 are registered as keywords 51. Note that only one keyword with the same notation is registered, such as “Takamatsu Castle (Kagawa Prefecture)” and “Takamatsu Castle (Okayama Prefecture)”.

  In the case of the example in FIG. 5, the keywords 51 of “Hideshihide Hashiba”, “Takamatsu Castle”, “Marugame Kaoru”, “Tokugawa Tsunakata”, “Honnoji Temple” and “Thailand” are registered in No1 to No6, respectively. ing.

  For example, as shown in FIG. 6, the link information DB 24 of the selected word and the linked document stores at least the document name 61, the selected word 62, and the linked document identification information 63 as link information for each No. Yes.

  The document name 61 indicates the name of the link source document (electronic book, web page, etc.). The selected word 62 indicates one selected from the keywords selected from the link source document. The link destination document identification information 63 is a word for uniquely identifying the same notation word, and is linked to the unique heading 32 of the link destination document DB 21.

  In the case of the example in FIG. 6, the link information of No. 1 is linked with the document name 61 of “Kagawa Prefecture History”, the selection word 62 of “Marugame Kaoru”, and the linked document identification information 63 of “Marugame Kaoru”. Yes. The link information of No. 2 is associated with the document name 61 of “History of Kagawa Prefecture”, the selection word 62 of “Tokukawa Tsunakata”, and the linked document identification information 63 of “Tokukawa Tsunakata”. The link information of No. 3 is associated with a document name 61 of “History of Kagawa Prefecture”, a selection word 62 of “Hideshiba Hideyoshi”, and link destination document identification information 63 of “Hideshiba Hashiba”. The link information of No. 4 is associated with the document name 61 of “History of Kagawa Prefecture”, the selected word 62 of “Takamatsu Castle”, and the linked document identification information 63 of “Takamatsu Castle (Kagawa Prefecture)”. The link information of No. 5 is associated with the document name 61 of “Sengoku Busho Oda Nobunaga”, the selection word 62 of “Honnoji no Hen”, and the link destination document identification information 63 of “Honnoji no Hen”.

  In the example of FIG. 6, the link destination document identification information 63 is linked to the unique heading 32 of the link destination document DB 21. However, the present invention is not limited to this, and the document location 34, the identification number, or the like Anything can be used as long as the link destination can be uniquely identified.

  For example, as shown in FIG. 7, the post-link link source document DB 25 stores at least the document name 71 and the document content 72 as post-link link source document information for each No.

  The document name 71 indicates the name of the link source document (electronic book, web page, etc.), and is linked to the document name 61 of the link information DB 24 between the selected word and the link destination document. The document content 72 indicates, for example, the content of a sentence (electronic book, web page, etc.) described in XML (Extensible Markup Language) format.

The link source document information after the link of No. 1 includes the document name 71 of “History of Kagawa Prefecture”, “<a
The document content 72 of "href =" http: // ... ~ Hideyoshi Hashiba </a> is Nagahama's governing policy ... "is linked. The link source document information after the No. 2 link is attached includes the document name 71 of “Sengoku Busho Nobunaga Oda” and “<a href =” http: // ... Document content 72 is linked.

  Of the database group as described above, the linked document DB 21 is created in advance by the administrator of the link assigning apparatus 1 based on the linked document acquired from an online dictionary, Wikipedia, or the like. The link source document DB 22 is created in advance by an administrator of the link providing device 1 based on an electronic book acquired (purchased) from an electronic book sales site or the like (not shown) or a web page acquired from the web. On the other hand, the keyword list file 23 to the post-link-attached link source document DB 25 are registered by a link assignment process described later with reference to the flowchart of FIG.

  FIG. 8 is a flowchart for explaining the link provision process executed by the link provision apparatus 1.

  In step S1, the control unit 11 of the link assignment device 1 creates a keyword list file 23 (FIG. 5) from the normal headline 31 (FIG. 3) in the link destination document DB 21.

  In step S2, the control unit 11 of the link providing apparatus 1 compares the document stored in the link destination document DB 21 (or the link source document DB 22) with the keyword list file 23 created in the process of step S1, and performs keyword extraction processing. I do.

  Here, the details of the keyword extraction process in step S2 of FIG. 8 will be described with reference to the flowchart of FIG. In the description of FIG. 9, the specific processing contents will also be described with reference to FIG.

  In step S21, the control unit 11 of the link assignment device 1 and the document (the document content 33 in FIG. 3 or the document content 42 in FIG. 4) stored in the link destination document DB 21 (or the link source document DB 22) and the step S1. The keyword list file 23 created by the processing is compared, and a character string whose character codes are completely matched is extracted as a matched character string. The character encoding method applied to the documents stored in the link destination document DB 21 (or the link source document DB 22) and the character encoding method applied to the keywords included in the keyword list file 23 are the same method.

  For example, the keyword list of “Nounaga may go through the style ... by the change of Honnoji” in the document content 42 linked to the link source document information No. 2 in the link source document DB 22 (FIG. 4), Files 23 (FIG. 5) are compared. As a result, as shown in FIG. 10A, as the character strings whose character codes are completely matched, as shown in FIG. 10A, the matching character strings 81a and No. 6 of “No. A matching character string 81b that completely matches the keyword “Thai” 51 is extracted.

  Returning to the description of FIG. In step S22, the control unit 11 of the link providing apparatus 1 performs a morphological analysis on the document (the document content 33 in FIG. 3 or the document content 42 in FIG. 4) stored in the link destination document DB 21 (or the link source document DB 22). .

  For example, the document content 42 linked to the link source document information No. 2 in the link source document DB 22 (FIG. 4) indicates that “Nobunaga's change may cause Nobunaga to follow its style…” Then, as shown in FIG. 10B, a “hononji” morpheme 82a, a “no” morpheme 82b, a “weird” morpheme 82c, and a “style” morpheme 82d are extracted.

  Returning to FIG. In step S23, the control unit 11 of the link providing device 1 selects one matching character string (matching character strings 81a and 81b in the example of FIG. 10A) extracted in the process of step S21 (for example, The matching character string 81a is selected). In step S24, the control unit 11 of the link providing apparatus 1 matches the matching character string 81a selected in the process of step S23 and the result of the morpheme analysis by the process of step S22 (in the example of FIG. 10B, morphemes 82a to 82d). ).

  In step S25, the control unit 11 of the link assignment apparatus 1 determines whether the matching part is a complete combination of morphemes as a result of the process of step S24, and determines that the matching part is a complete combination of morphemes. In step S26, the matching character string is extracted as a keyword.

  For example, the matched character string 81a of “Honnoji” is compared with the “Honnoji” morpheme 82a, “No” morpheme 82b, “Hen” morpheme 82c, and “Style” morpheme 82d. The matched character string 81a is a complete combination of the morphemes 82a, 82b, and 82c, and therefore is extracted as the keyword 83 as shown in FIG.

  On the other hand, when the control unit 11 of the link providing apparatus 1 determines in step S25 that the matching part is not a complete combination of morphemes, the process of step S26 is skipped and the process proceeds to step S27.

  In step S27, the control unit 11 of the link assignment device 1 determines whether or not the processing has been completed for all the matching character strings extracted in the processing of step S21, and if there is a matching character string that has not been processed yet. When it determines, it returns to step S23 and performs the process mentioned above repeatedly. Then, in step S27, when the control unit 11 of the link assignment device 1 determines that the process has been completed for all the matching character strings extracted in the process of step S21, the keyword extraction process is terminated, and the step of FIG. Return to S3.

  In step S3, the control unit 11 of the link assigning apparatus 1 selects one keyword extracted in the keyword extraction process in step S2 (FIG. 9) (for example, selects the keyword 83 of “Honnoji”), and selects the selected word. To decide. In step S4, the control unit 11 of the link providing apparatus 1 determines whether or not the link destination document is uniquely determined for the selected word determined in the process of step S3. Proceed to

  For example, since the selected word “Honnoji no Hen” is the same as the unique heading 32 in the linked document DB 21, it is determined that the linked document is uniquely determined for the selected word.

  In step S5, the control unit 11 of the link providing device 1 stores link information with the link destination document DB 21 determined to be uniquely determined for the selected word in the link information DB 24 between the selected word and the link destination document. . Accordingly, for example, as shown in FIG. 6, the link information of No. 5 in the link information DB 24 between the selected word and the linked document includes the document name 61 of “Sengoku Busho Nobunaga” and the selected word of “Honnoji no Hen”. 62 and link destination document identification information 63 of “Honnoji no Hen” are stored.

  In step S4, when the control unit 11 of the link providing apparatus 1 determines that the link destination document is not uniquely determined for the selected word determined in the process of step S3, the process of step S5 is skipped, and the process proceeds to step S6. move on.

  For example, when “Takamatsu Castle” is determined as the selected word, there are two “Takamatsu Castle (Kagawa Prefecture)” and “Takamatsu Castle (Okayama Prefecture)” in the unique heading 32 of the linked document DB 21. It is determined that the link destination document is not uniquely determined for the selected word.

  In step S6, the control unit 11 of the link assignment device 1 determines whether all the keywords extracted in the keyword extraction process in step S2 have been selected. Returning to S3, the above-described processing is repeatedly executed. If the controller 11 of the link providing device 1 determines in step S6 that all keywords have been selected, the process proceeds to step S7.

  In step S7, the control unit 11 of the link providing apparatus 1 selects one unsupported keyword (same synonym) from the document in the link source document DB 22 (document content 42 in FIG. 4) (for example, “Takamatsu” Select the keyword “castle” and decide the selected word. In step S8, the control part 11 of the link provision apparatus 1 performs an association process with respect to the selected word (same notation synonym) determined by the process of step S7.

  Here, with reference to the flowchart of FIG. 11, the details of the same notation synonym association processing in step S8 of FIG. 8 will be described. In the description of FIG. 11, the specific processing contents will also be described with reference to FIG.

In step S31, the control unit 11 of the link providing device 1 acquires the appearance frequency of the keyword included in the link destination document for the selected word, the peripheral word, and the peripheral word in the link source document stored in the link source document DB 22. , TF (term frequency), IDF (inverse document
frequency) method. In the TF / IDF method, known techniques (including improved techniques) can be used.

  For example, as shown in FIG. 12, in the link source document 91, if the selected word is “Takamatsu Castle” and the peripheral words are “Marugame Kaoru”, “Tokugawa Tsunakata”, “Hashiba Hideyoshi” The linked documents 92 are “Marugame Kaoru”, “Tokugawa Tsunakata”, and “Hashiba Hideyoshi”, respectively. The keywords 93 appearing in the peripheral word linked documents included in the peripheral word linked documents 92 are “Marugame Castle, Iki Kuni, Kagawa Prefecture”, “Iki Kuni, Takamatsu Satoshi, and Matsudaira Yoseshige, respectively. ”,“ Takemoto Mohri, Bi China, Nobunaga Oda ”. The control unit 11 of the link providing device 1 obtains the appearance frequency of the keyword 93 that appears in the link destination document 92 of the peripheral word, and generates the feature vector 94 of the selected word by weighting by the TF / IDF method.

In the process of step S31, if the peripheral word is a synonym of the same notation and the link destination document is not uniquely determined, the peripheral word is excluded from the target of the similarity calculation process. For example, in the example of FIG. 12, it is assumed that the description of the dictionary regarding “Marugame Kaoru” varies depending on the era and is divided into the following two types.
1. Marugame Kaoru (Ikoma's time)
2. Marugame Kaoru (era of Mr. Yamazaki and Mr. Kyogoku)

  If the linked document of “Marugame Kaoru” is not uniquely determined when performing the process of associating “Takamatsu Castle”, “Marugame Kaoru” is excluded from the similarity calculation process. That is, by excluding peripheral words whose link destination documents are not uniquely determined from processing targets, it is possible to prevent noise information from being mixed, and as a result, the accuracy of the association processing is improved.

  In step S32, the control unit 11 of the link assigning apparatus 1 acquires the appearance frequency of the keyword included in the keyword list file 23 (FIG. 5) in each link destination document candidate, and weights it by the TF-IDF method.

  For example, as shown in FIG. 12, keywords appearing in the linked document candidate 95a of “Takamatsu Castle (Kagawa Prefecture)” are “Marugame Kaoru, Sanuki Country, Hikida Castle, Kagawa Prefecture”, and “Takamatsu Castle ( The keywords appearing in the linked document candidate 95b of “Okayama Prefecture” are “Okayama Prefecture, Bin China, Hideyoshi Hashiba, and Mr. Ishikawa”. The control unit 11 of the link assigning apparatus 1 acquires the appearance frequencies of the keywords appearing in the link destination document candidates 95a and 95b, respectively, and weights them by the TF / IDF method, thereby obtaining the feature vectors 96a and 96b of the link destination document candidates. Generate.

  In step S33, the control unit 11 of the link providing apparatus 1 calculates the similarity between the selected word and each link destination document candidate using the weighted result obtained by the processing in steps S31 and S32.

  In the example of FIG. 12, the similarity between the feature vector 94 of the selected word and the feature vectors 96a and 96b of the linked document candidates is calculated.

  In step S34, the control unit 11 of the link assigning apparatus 1 sets the link destination document candidate having the maximum similarity as the link destination document for the selected word from the calculation result obtained in step S33, and sets the selected word and the link destination document. It is stored in the link information DB 24 (FIG. 6).

  In the example of FIG. 12, since the linked document candidate 95a has the highest similarity, the selected word 62 of “Takamatsu Castle (Kagawa Prefecture)” and the linked document identification information 63 of “Takamatsu Castle (Kagawa Prefecture)” Are linked and stored in the link information DB 24 between the selected word and the linked document.

  As described above, in addition to the frequency of appearance of peripheral words of keywords, the frequency of appearance of keywords appearing in the document content to which the peripheral words are associated is also considered in the similarity calculation, thereby increasing the types of words used for similarity calculation. The influence on a specific word can be reduced, and the accuracy of the association process can be improved.

  Returning to the description of FIG. In step S9, the control unit 11 of the link assignment device 1 determines whether or not all the synonym synonyms (unsupported keywords) have been selected from the link source document DB 22, and the synonym synonyms that have not yet been selected. If it is determined that there is, the process returns to step S7, and the above-described processing is repeatedly executed. In step S9, when the control unit 11 of the link providing apparatus 1 determines that all the same-notation synonyms have been selected, the process proceeds to step S10.

  In step S10, the control unit 11 of the link assignment device 1 refers to the link information DB 24 between the selected word and the link destination document, assigns a link to the link source document, and provides a link source document DB 25 after link assignment (FIG. 7). To store.

  As shown in the example of FIG. 7, in the document content 72 of the document name 71 of “History of Kagawa Prefecture”, “a href =” http: // ・ ・ ・ /marugamehan.txt/ ”” is added to “Marugame Kaoru”. A link is added, “a href =” http: // ・ ・ ・ /takamatsujyo-kagawaken.txt/ ”” is added to “Takamatsu Castle”, and “a href =” http: //...Tokugawatsunakata.txt/ ”” is added, and “a href =” http: // ・ /hashibahideyoshi.txt/ ”” is added to “Hashiba Hideyoshi”.

  In the above link assignment processing, the reason why the words that are not the same notation (that is, unique words) are processed first in steps S3 to S6 and the same notation is processed later in steps S7 to S9 is shown in FIG. As described above in step S31, if the peripheral word is a synonym of the same notation and the link destination document is not uniquely determined, the peripheral word is removed from the target of similarity calculation processing and appears in the sentence. This is because if the processing is performed in order, the number of neighboring words is reduced.

  In the example of FIG. 12, if “Takamatsu Castle” is processed after “Marugame Kaoru” (processing in the order in which they appear in the text), “Tokukawa Tsunakata” and “Hideshiba Hashiba” have unique linked documents. Since there is no such thing, the only word surrounding "Takamatsu Castle" is "Marugame Kaoru". In other words, by processing words that are not synonyms in advance, the number of peripheral words for the synonyms can be increased, and the accuracy of the similarity calculation process is improved.

  With the above link assignment processing, an appropriate link destination document link is assigned to a keyword included in an electronic book or a web page. The present invention can also provide a user with an electronic book, a web page, or the like to which a link of a linked document is assigned.

[Examples of using e-books after linking]
FIG. 13 is a diagram illustrating a configuration example of an electronic book providing system 100 that provides an electronic book with a link attached.

  An electronic book providing system 100 shown in FIG. 13 is configured by connecting an electronic book browsing terminal 101 and an electronic book providing server 102 to each other via a network (not shown). Note that the numbers of the electronic book browsing terminal 101 and the electronic book providing server 102 are arbitrary, and it is of course possible to provide a plurality of them.

The electronic book browsing terminal 101 is a CPU (Central
Processing Unit), ROM (Read Only
Memory), RAM (Random Access)
A computer equipped with a memory, a hard disk drive (HDD), an input unit, a display screen, and the like. The electronic book browsing terminal 101 is preinstalled with a viewer application for browsing an electronic book, and displays the electronic book downloaded by the user. When a certain phrase of the electronic book being browsed is selected by the user, the electronic book browsing terminal 101 requests the electronic book providing server 102 to transmit a document linked to the phrase, and the electronic book providing server 102 The document data provided from 102 is received and presented to the user.

  The electronic book providing server 102 is a computer system including a CPU, a ROM, a RAM, an HDD, and the like, and has the function of the link providing device 1 shown in FIG. The electronic book providing server 102 receives the phrase data selected by the electronic book browsing terminal 101, searches the database group in the storage unit 12 for the document data linked to the phrase, and displays the search results and the like as electronic data. It is transmitted (provided) to the book browsing terminal 101.

  The electronic book providing process will be described with reference to FIG.

  In step S101, when a predetermined electronic book is selected by the user, the control unit 11 of the electronic book viewing terminal 101 requests transmission of the electronic book. In step S <b> 102, the control unit 11 of the electronic book providing server 102 reads out the requested electronic book from the link source document DB 25 after link addition based on a request from the electronic book viewing terminal 101, and transmits it to the electronic book viewing terminal 101. To do.

  The control unit of the electronic book browsing terminal 101 displays the electronic book received from the electronic book providing server 102 on the display unit W1. In step S103, when a predetermined word / phrase is selected by the user in the electronic book displayed on the display unit W1 in the electronic book browsing terminal 101 (“Takamatsu Castle” in the example of FIG. 13). , Request transmission of the document linked to the phrase. The phrase selected by the user is given a link of “a href =” http: // ・ ・ ・ /takamatsujyo-kagawaken.txt/ ”” (Fig. 7). Send is requested.

  In step S104, the control unit 11 of the electronic book providing server 102, based on the link destination (a href = ”http: //... /Takamatsujyo-kagawaken.txt/”) received from the electronic book browsing terminal 101, A link destination document (NO2 link destination document in the example of FIG. 3) is read from the link destination document DB 21 of the storage unit 12 and transmitted to the electronic book browsing terminal 101. The control unit of the electronic book browsing terminal 101 displays the electronic book received from the electronic book providing server 102 on the display unit W2.

[Embodiments of the present invention]
As described above, in an electronic book, the meaning of a word or phrase appearing in a certain document can be determined by associating the document in consideration of the meaning of the keyword based on the document content in advance and storing the associated information in the database. When checking, even if there are a plurality of candidate documents, an appropriate document can be presented.

[Evaluation comparison between the present invention and the prior art]
An example of evaluation in the case where the correct answer of association is determined manually by comparing the correct answer rate of the association by the technique of the present invention with the correct answer rate of the association by the conventional technique will be described.

  First, evaluation data will be described. The related source document is “Aozora Bunko New Character New Kana 150 works”, the related destination document is “Japanese Wikipedia”, the number of keywords used is “456116 types”, the keyword selection criterion is “Wikipedia heading, heading that starts in Japanese” Among them, exclude headlines composed only of hiragana, morphological analyzer is “tea bowl”, morphemes that are excluded from keyword elements are “particles”, “auxiliary verbs”, “prefixes”, “connectives”, “fillers”, "Noun-pronoun", "noun-number", "noun-independent", "noun-non-independent", "noun-conjunctive", "noun-quoted string", "noun" -Special "," Noun-suffix "," Verb-non-independent "," Symbol "," Other ".

  FIG. 14 is a diagram illustrating an example of evaluation results of the method of the present invention (example) and the conventional method (comparative example). The evaluation data uses the data described above. In the conventional method, a feature vector is generated only from the related source keyword and peripheral words, and the keyword and the document are associated with each other.

  In the example of FIG. 14, the correct answer rate is calculated from 224 keywords in which one of the example and the comparative example is correct among the correct answer determination results by hand. In the example, 138 cases were correct and the correct rate was 61.6%, while in the comparative example, 86 cases were correct and 38.4%.

  FIG. 15 is a diagram illustrating an example of an evaluation result when the number of correct answers is calculated for each number of neighboring words. The evaluation data uses the data described above.

  In the example of FIG. 15, keywords with the number of surrounding words of 30 or less occupy a correct answer rate of 91%, and in those keywords, the correct answer rate of the example is higher than the correct answer rate of the comparative example. .

  As described above, it can be seen from the evaluation results of FIGS. 14 and 15 that the accuracy rate of the example increases when the number of neighboring words is small. That is, by applying the present embodiment, it is possible to give an appropriate link destination to a keyword appearing in an electronic book, a web page, or the like.

  As mentioned above, although suitable embodiment, such as the link provision apparatus concerning this invention, was described, referring an accompanying drawing, this invention is not limited to this example. It will be apparent to those skilled in the art that various changes or modifications can be conceived within the scope of the technical idea disclosed in the present application, and these naturally belong to the technical scope of the present invention. Understood.

DESCRIPTION OF SYMBOLS 1 ......... Link provision apparatus 11 ......... Control part 12 ......... Storage part 21 ......... Link destination document DB
22 ... Link source document DB
23 ......... Keyword list file 24 ......... Link information DB of selected word and linked document
25 ... …… Link source document DB after link is granted

Claims (5)

A keyword list creation means for creating a keyword list from the headings of linked documents;
Using the keyword list, keyword extraction means for extracting keywords from the linked document and the linked document;
A selected word determining means for selecting one of the synonyms in the keyword of the link source document and determining the selected word;
In the link source document, the appearance frequency of the keyword included in the link destination document with respect to the selected word, the peripheral word, and the peripheral word is acquired, and the frequency of appearance of the keyword included in the keyword list is determined for each link destination document candidate. A similarity calculation means for acquiring and calculating a similarity between the selected word and each linked document candidate;
A selected word link destination document determining means for determining a link destination document candidate having the maximum similarity as a link destination document for the selected word;
A link providing device comprising: The link assignment apparatus according to claim 1, wherein the similarity calculation unit excludes the peripheral words whose link destination documents are not uniquely determined from the target of similarity calculation processing. The selected word determining means selects a unique word prior to the same notation among the keywords of the link source document,
The selected word link destination document determining means determines the link destination document for the unique word prior to the link destination document for the synonymous heterogeneity according to the selection by the selected word determining means. The link provision apparatus of description. A link granting method executed by the link granting device,
A control unit of the link providing device, a keyword list creating step of creating a keyword list from a heading of a linked document;
The control unit uses the keyword list to extract a keyword from a link destination document and a link source document;
The control unit selects one of the same synonyms among the keywords of the link source document and determines a selected word;
The control unit obtains the selection word, the peripheral word, and the appearance frequency of the keyword included in the link destination document for the peripheral word in the link source document, and is included in the keyword list in each link destination document candidate A similarity calculation step of acquiring a keyword appearance frequency and calculating a similarity between the selected word and each linked document candidate;
A selection word link destination document determination step in which the control unit determines a link destination document candidate having the maximum similarity as a link destination document for the selection word;
The link provision method characterized by including this. A keyword list creation means for creating a keyword list from the headings of linked documents;
Using the keyword list, keyword extraction means for extracting keywords from the linked document and the linked document;
A selected word determining means for selecting one of the synonyms in the keyword of the link source document and determining the selected word;
In the link source document, the appearance frequency of the keyword included in the link destination document with respect to the selected word, the peripheral word, and the peripheral word is acquired, and the frequency of appearance of the keyword included in the keyword list is determined for each link destination document candidate. A similarity calculation means for acquiring and calculating a similarity between the selected word and each linked document candidate;
A selected word link destination document determining means for determining a link destination document candidate having the maximum similarity as a link destination document for the selected word;
A program for causing a computer to function as a link providing device having

Images