JP2011013969A - Apparatus and method for supporting link creation, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a technique for sharply reducing creator's loads in link creation.SOLUTION: A link creation supporting apparatus includes: a reference word extraction part 110 for extracting a reference word to be a reference for extracting a link setting range candidate; a link setting range candidate extraction part 120 for extracting link setting range candidates in each link purpose candidate to be a candidate of a link purpose; a link destination candidate retrieval part 130 for retrieving link destination candidates to be candidates of a link destination in each group of a link purpose candidate and a link setting range candidate; a first determination part 140 for calculating a first evaluation value in each group of a link purpose candidate, a link setting range candidate and a link destination candidate, totalizing the first evaluation values in each purpose candidate, and determining a link purpose from the link purpose candidates based on the totalized first evaluation value; and a second determination part 150 for calculating a second evaluation value in each group of a link setting range candidate and a link destination candidate in the link purpose determined by the first determination part, and determining a link setting range from link setting range candidates based on the second evaluation value.

Description

  The present invention relates to a link creation support apparatus, a link creation support method, and a program.

  In general, creators of web contents such as blogs (hereinafter simply referred to as “contents”) add other contents within the contents being created in order to enrich the contents being created when the contents are created (description). Create a link to Link creation usually specifies the range (anchor text) for creating a link on the content being created, searches for the link destination within the designated range, and describes the URL of the found link destination in the designated range. To do. Every time a creator creates one link, the creator performs the above-described complicated work. Therefore, a mechanism that can reduce the burden on the creator when creating a link is desired.

  Currently, as a mechanism that can reduce the burden on creators when creating links, a database that stores keywords and URLs of explanation pages that explain the keywords is built in advance. A method of automatically assigning a link to is known (for example, see Patent Document 1).

  Further, when creating a document such as content, a set of an object (eg, restaurant A) and an attribute (eg, business hours) appearing in the document being created is recognized from a grammatical pattern, and a value for the attribute (eg, 10). 0: 00-24: 00) is automatically acquired based on information such as a list structure and presented to the creator (for example, see Non-Patent Document 1).

JP 2005-202895 A

“Text creation support environment using attributes / values dynamically acquired from the Web,” Kazumasa Nakamura, Naoki Yoshinaga, Kentaro Torizawa, DBWeb2007, Vol. 5B-1

  However, since the method of Patent Document 1 can only deal with keywords in the database, there is a problem that the burden on the creator at the time of creating a link cannot be reduced sufficiently. In addition, since the method of Non-Patent Document 1 can handle only a pair of an object and an attribute, there is a problem that the burden on the creator at the time of creating a link cannot be reduced sufficiently. That is, the above two methods have a problem that the creator feels a burden.

  The present invention has been made in view of such circumstances, and an object of the present invention is to provide a technique for greatly reducing the burden on the creator when creating a link.

  In order to solve the above problem, the link creation support apparatus according to one aspect of the present invention extracts link setting range candidates that are candidates for the link setting range in which a link destination should be set as anchor text from the content being created. A reference word extraction unit that extracts a reference word that serves as a reference, and for each link purpose candidate that is a link purpose candidate that is the purpose of link creation from the content being created, based on the reference word Using the link setting range candidate extraction unit for extracting the link setting range candidates and the search method according to the link purpose candidate and the link setting range candidate, the link destination candidate and link setting range candidate for each set of link destination candidates Link destination candidate search unit that searches for candidate link destination candidates, and appropriateness as a link destination for each set of link purpose candidates, link setting range candidates, and link destination candidates A first determination unit that calculates a first evaluation value to be displayed, aggregates the first evaluation value for each link purpose candidate, and determines a link purpose from the link purpose candidate based on the first evaluation value after the aggregation; For the link purpose determined by one determination unit, a second evaluation value indicating appropriateness as a link destination is calculated for each set of link setting range candidates and link destination candidates, and link setting is performed based on the second evaluation value And a second determination unit that determines a link setting range from the range candidates.

  In the link creation support apparatus, the first determination unit determines the link purpose from the link purpose candidates for each group of link setting range candidates that are candidates for one link setting range for which one link destination is to be set, The 2 determining unit may determine the link setting range from the link setting range candidates for each group of link setting range candidates that are candidates for one link setting range for which one link destination is to be set.

  In the link creation support apparatus, the first determination unit determines the link purpose candidate with the highest first evaluation value after aggregation as the link purpose, and the second determination unit determines the set with the highest second evaluation value. You may make it determine a link setting range candidate as a link setting range.

  In the link creation support apparatus, the first determination unit calculates a first evaluation value using the first prediction model generated by multivariate analysis, and the second determination unit is generated by multivariate analysis. Alternatively, the second evaluation value may be calculated using the second prediction model.

  In the link creation support apparatus, the link setting range candidate extraction unit may extract link setting range candidates having different numbers of words according to the link purpose candidates.

  In the link creation support apparatus, the reference word extraction unit may extract a reference word based on a word score indicating a frequency at which one word is in a link setting range in content including the word.

  In the link creation support apparatus, the link setting range candidate extraction unit may extract link setting range candidates having different numbers of words according to the word score.

  In the link creation support apparatus, the word score may be assigned to each set of a character string and a word attribute constituting the word.

  In the link creation support apparatus, the word score is higher in the link setting range in the content created by the user than in the link setting range in the content created by others. It may be.

  The link creation support apparatus further includes an anchor text setting unit that sets a link destination in the link setting range in the content being created, and the second determination unit determines from the link setting range candidates based on the second evaluation value. One link setting range is determined, and one link destination candidate corresponding to the link setting range is supplied to the anchor text setting unit. The anchor text setting unit adds the first link setting range to the link setting range acquired from the second determination unit. 2 You may make it set the link destination candidate acquired from the determination part.

  In order to solve the above-described problem, a link creation support method according to another aspect of the present invention includes a link setting range candidate to be a link setting range candidate to set a link destination as anchor text from within the content being created. Reference word extraction means for extracting a reference word that serves as a reference when extracting, and for each link purpose candidate that becomes a link purpose candidate that is the purpose of link creation, based on the reference word from the content being created, Link destination for each set of link purpose candidates and link setting range candidates using link setting range candidate extracting means for extracting one or more link setting range candidates and a search method according to the link purpose candidates and link setting range candidates Link destination candidate search means for searching for link destination candidates that are candidates for the link, and a link destination for each set of link purpose candidates, link setting range candidates, and link destination candidates First determination means for calculating a first evaluation value indicating appropriateness, totaling the first evaluation value for each link purpose candidate, and determining a link purpose from the link purpose candidate based on the first evaluation value after the totalization And, for the link purpose determined by the first determining means, a second evaluation value indicating the appropriateness as the link destination is calculated for each set of the link setting range candidate and the link destination candidate, and the second evaluation value is calculated based on the second evaluation value. And second determining means for determining the link setting range from the link setting range candidates.

  In order to solve the above problem, a program according to an aspect of the present invention provides a link setting range in which a link destination should be set as anchor text from the content being created on a computer of a link creation support apparatus that supports link creation. A reference word extraction step for extracting a reference word as a reference when extracting a link setting range candidate that is a candidate for the link, and a link purpose that is the purpose of creating a link based on the reference word from the content being created Link target candidates and links using a link setting range candidate extraction step for extracting one or more link setting range candidates for each candidate link purpose candidate, and a search method according to the link purpose candidate and the link setting range candidate Link destination candidate search step for searching for link destination candidates as link destination candidates for each set of setting range candidates, The first evaluation value indicating the appropriateness as the link destination is calculated for each set of the link setting range candidate and the link destination candidate, the first evaluation value is aggregated for each link purpose candidate, and the first evaluation after the aggregation is performed. In the first determination step of determining the link purpose from the link purpose candidates based on the value, and the link purpose determined in the first determination step, the appropriateness as the link destination is determined for each set of the link setting range candidate and the link destination candidate. A second evaluation value is calculated, and a second determination step of determining a link setting range from a link setting range candidate based on the second evaluation value is executed.

  According to the present invention, the burden on the creator when creating a link can be greatly reduced.

It is a functional block diagram of link creation support 10 by a 1st embodiment of the present invention. It is explanatory drawing explaining calculation of a word score. It is explanatory drawing explaining extraction of a link setting range candidate. 5 is a flowchart illustrating an example of an operation of a link setting range candidate extraction unit 120. It is explanatory drawing explaining the detection of a link destination candidate, and determination of the link objective. It is an example of the data for prediction model learning. It is a flowchart which shows an example of operation | movement of the link creation assistance 10. It is a functional block diagram of the link creation assistance 12 by the 2nd Embodiment of this invention. It is a flowchart which shows an example of operation | movement of the link creation assistance 12.

(First embodiment)
The link creation support 10 according to the first embodiment of the present invention will be described below. The link creation support 10 is an apparatus that supports content creation, and presents a link (such as a link creation position) that meets the link purpose of the content creator to the creator. The link purpose is the purpose and intent of creating a link when creating a link in the content being created. When creating a link, there are a plurality of link purposes, and an appropriate link destination is different for each link purpose. Therefore, in order to determine (estimate) an appropriate link destination, it is necessary to determine (estimate) an appropriate link purpose.

Typical link objectives are:
-Past article reference The purpose is to refer to content (past article) created by the creator himself in the past and related to the content being created. In other words, a link whose purpose is “reference past article” is a link for referring to a past article.
・ Keyword explanation The purpose is to refer to contents (explanatory contents) such as dictionary sites that explain the meaning of specific keywords. In other words, a link whose purpose is “keyword commentary” is a link for referring to commentary content.
・ Official website reference The purpose is to refer to official websites of companies and services. In other words, a link whose purpose is “reference official site” is a link for referring to the official site.
-Reference to information source page This is for the purpose of referring to the content (information source page) that provides detailed information related to the description content of the content being created and the basis of the description content. In other words, the link for the purpose of “information source page reference” is a link for referring to the information source page.

  Hereinafter, the link creation support 10 will be described with reference to the drawings. As shown in FIG. 1, the link creation support 10 includes an input / output unit 100, a reference word extraction unit 110, a link setting range candidate extraction unit 120, a link destination candidate search unit 130, a first determination unit 140, and a second determination unit 150. , A prediction model generation unit 160, a word score calculation unit 180, a word DB 190, a past article DB 191, an externally acquired content DB 192, and a prediction model DB 193.

  The input / output unit 100 acquires content that is being created (including content that is modified after creation). The input / output unit 100 supplies the content being created to the reference word extraction unit 110. The input / output unit 100 supplies the created content to the past article DB 191. It also outputs various messages to the creator.

  The reference word extraction unit 110 acquires the content being created from the input / output unit 100. When the reference word extraction unit 110 acquires the content being created, the reference word extraction unit 110 extracts the link setting range candidate as a link setting range candidate for which a link destination should be set as anchor text from the content being created. A reference word (characteristic word) is extracted. Specifically, the reference word extraction unit 110 calculates a reference word based on a word score that indicates the frequency (easy to be included in an anchor text) of a single word that is within a link setting range in the content including the word. Extract.

  This will be described in detail below. First, the reference word extraction unit 110 extracts words from the content being created by morphological analysis. The words extracted by the reference word extraction unit 110 are nouns and verbs, and function words such as particles and auxiliary verbs are not extracted.

  Subsequently, the reference word extraction unit 110 refers to the word DB 190 and assigns a word score to words (nouns and verbs) extracted from the content being created. Note that the word score calculated by the word score calculation unit 180 is stored in the word DB 190. Note that the word score calculation unit 180 analyzes existing content and anchor text and calculates a word score for each word, which will be described later.

  Subsequently, the reference word extraction unit 110 further extracts (extracts as a reference word) a word whose word score is greater than or equal to the first threshold or a word that is less than or equal to the second threshold. Note that the reference word extraction unit 110, when the number of words whose word score is equal to or higher than the first threshold value or 0 words equal to or lower than the second threshold value is 0, there is a portion suitable for link creation in the content being created. You may notify the input / output part 100 of not doing. The input / output unit 100 that has acquired the notification outputs a message indicating that there is no appropriate location for creating a link in the content being created.

  Subsequently, the reference word extraction unit 110 registers the word score in the technical term dictionary or encyclopedia stored in the word DB 190 with respect to a word having a word score equal to or higher than the first threshold or a word equal to or lower than the second threshold. A label indicating any one of “anchor word”, “anchor attached word”, and “anchor adjacent word” is given based on the frequency of occurrence immediately before and after the anchor text. For example, the reference word extraction unit 110 assigns a label indicating “anchor center word” to a word whose word score is equal to or higher than a first threshold and does not correspond to a general word registered in the technical term dictionary, A word indicating an “anchor attached word” is assigned to a word corresponding to a general word whose word score is equal to or higher than the first threshold and is not registered in the technical term dictionary. Further, a label indicating “anchor adjacent word” is given to a word whose word score is equal to or less than the second threshold and the ratio of occurrence before or after the anchor text stored in the word DB 190 is equal to or more than the third threshold. Give.

  The anchor center word is a word that is easily included in the anchor text and is likely to be a central matter of the content indicated by the anchor text. An anchor center word can constitute anchor text by itself. An example of the anchor center word is a company name, a place name, a person name, or the like. The anchor attached word is a general word that is easily included in the anchor text. An anchor attached word cannot constitute anchor text by itself. Examples of anchor attached words are “article” and “announcement”. An anchor adjacent word is a word that is unlikely to be included in anchor text and is likely to appear immediately before or after the anchor text. An example of an anchor adjacent word is a verb such as “by” of “by” or “read” of “read”, and is a word following the contents of the link destination in the anchor text.

  Finally, the reference word extraction unit 110 supplies the reference word (with label) to the link setting range candidate extraction unit 120 together with the content being created acquired from the input / output unit 100. The reference word extraction unit 110 may supply the word score of each reference word to the link setting range candidate extraction unit 120 in addition to the content being created and the reference word (with label).

  The calculation of the word score will be described. The past article DB 191 stores content (past articles) acquired from the input / output unit 100, and the externally acquired content DB 192 stores external contents collected (acquired) by web crawling. ing. The word score calculation unit 180 refers to the past article DB 191 and the externally acquired content DB 192, and includes a web page including a certain word, a word appearing in the anchor text, a ratio of a certain word included in the anchor text, Statistical information such as the appearance position outside the anchor text is analyzed, and a word score is calculated.

  Specifically, for example, as shown in FIG. 2, the word score calculation unit 180 is a content in which a word appears from a set of existing contents in the externally acquired content DB 192 and a set of past articles in the past article DB 191. When the number is X and the number of contents including the word in the anchor text is X, Y / X is calculated as the word score of the word.

  In the example of FIG. 2, the number of contents X, the number of contents Y, and the word score X / Y are calculated for each word. However, in order to improve accuracy, the attribute of the word (for example, place name, person name) is considered, The number of contents X, the number of contents Y, and the word score X / Y may be calculated. That is, for example, a word score may be assigned to each set of character strings and word attributes constituting a word (for example, different word scores for “Yamaguchi”-“place name” and “Yamaguchi”-“person name”). Give). Further, when the anchor text is in the content (past article) created by the user, a higher score may be given compared to the anchor text in the content created by another person.

  The link setting range candidate extraction unit 120 acquires a reference word (with a label) from the reference word extraction unit 110 together with the content being created. The link setting range candidate extraction unit 120 that has acquired the reference word together with the content being created, sets one or more link settings for each link purpose candidate that is a candidate for the link purpose based on the reference word from the content being created. Extract range candidates. The link setting range candidate extraction unit 120 may extract link setting range candidates having different numbers of words according to the link purpose candidates. Note that the link setting range candidate extraction unit 120, in addition to the content being created and the reference word, when acquiring the word score of each reference word from the reference word extraction unit 110, depending on the word score or the purpose of the link Depending on the candidates and the word score, link setting range candidates having different numbers of words may be extracted.

  When the link setting range candidate extraction unit 120 extracts the link setting range candidate for each link purpose candidate, the link setting range candidate extraction unit 120 supplies the link setting range candidate for each link purpose candidate to the link destination candidate search unit 130. In other words, the link setting range candidate extraction unit 120 supplies the link destination candidate-link setting range candidate pair to the link destination candidate search unit 130.

  Hereinafter, extraction of a link setting range candidate will be described in detail. The link setting range candidate extraction unit 120 is based on the reference word (anchor center word, anchor attached word, anchor adjacent word) in the content being created based on the reference word of each label extracted by the reference word extraction unit 110. , Gradually adding words (setting the additional limit number n), gradually expanding the range on the content being created, and for each link purpose candidate, a plurality of link setting range candidates (links) from the content being created The setting range candidate pattern) is extracted. In other words, as illustrated in FIG. 3, the link setting range candidate extraction unit 120 extracts, for example, a link setting range candidate pattern for each link purpose candidate for the reference word W1, and the link purpose candidate-link setting range candidate. A list LW1 that stores a set of The link setting range candidate extraction unit 120 performs the same process on all reference words of the entire content being created, and creates a list L = [LW1, LW2,... Generate. In other words, the link setting range candidate extraction unit 120 extracts link setting range candidate patterns for all link purposes when the reference word Wk is used as a reference.

  The additional limit number n is set to a different value depending on the link purpose. The number of words in the anchor text for the link purpose “keyword explanation” and the link purpose “refer to the official website” is short, and the word in the anchor text for the link purpose “reference source page” and the link purpose “reference past article” This is because there is a correlation between the purpose of the link and the number of words in the anchor text, such as a long number. In the example shown in FIG. 3, in the case of the link purpose “keyword explanation”, the one word before and after is set as an additional limit number (additional limit number n = 1), and in the case of the link purpose “reference information source page”, before and after. Three words are set as the additional limit number (additional limit number n = 3). Different values may be set according to the word score. However, regardless of the additional limit number n, the extended range is limited to the range of one sentence (punctuation break) to which the reference word belongs (the same sentence is extended as a limit).

  The link setting range candidate extraction unit 120 may extract a link setting range candidate pattern in consideration of whether the reference word is an anchor center word, an anchor attached word, or an anchor adjacent word. For example, when one sentence includes a reference word that is an anchor center word, the link setting range candidate extraction unit 120 uses the reference word that is the anchor center word in the sentence as a reference (the anchor center word as a base point (axis)). Extract link setting range candidate patterns. The link setting range candidate extraction unit 120, when one sentence does not include a reference word that is an anchor central word but includes a reference word that is an anchor attached word, uses the anchor attached word in the sentence as a reference (based on the anchor attached word). As a result, a link setting range candidate pattern is extracted. The link setting range candidate extraction unit 120, when one sentence includes a reference word that is an anchor center word and a reference word that is an anchor attached word and does not include an anchor adjacent word, includes an anchor adjacent word in the sentence. Is used as a reference (the word immediately before or immediately after the anchor adjacent word as a base point) to extract a link setting range candidate pattern. Note that the link setting range candidate extraction unit 120 does not extract a link setting range candidate pattern from the sentence when it does not include a reference word.

  Hereinafter, the operation by the link setting range candidate extraction unit 120 will be described with reference to the flowchart of FIG. In FIG. 4, the link setting range candidate extraction unit 120 initializes an index i of a sentence that constitutes the content being created (step S201). The link setting range candidate extraction unit 120 extracts the i-th sentence from the content being created (step S202). The link setting range candidate extraction unit 120 determines whether or not a reference word that is an anchor center word is included in the i-th sentence (step S203). If the link setting range candidate extraction unit 120 determines that the reference word that is the anchor center word is included in the i-th sentence (step S203: Yes), the link setting range candidate extraction unit 120 uses the reference word that is the anchor center word as a reference. As described above, a link setting range candidate pattern is extracted (step S210).

  That is, when the i-th sentence includes the reference word Wa that is the anchor center word, the link setting range candidate extraction unit 120 generates a list LWa that stores link purpose-link setting range candidates based on the reference word Wa. Generate. More specifically, first, the link setting range candidate extraction unit 120 generates an LWa that stores a certain link purpose-reference word Wa. Next, the link setting range candidate extraction unit 120 sets the word addition count s to 1, and among the link setting range candidates composed of s + 1 words, selects the link setting range candidate whose t-th word is the reference word Wa. Add to LWa (Process 1). The link setting range candidate extraction unit 120 repeatedly executes the process 1 for t of 1 or more (s + 1) or less (process 2). Further, the link setting range candidate extraction unit 120 repeatedly executes the process 2 for s that is 1 or more and n (additional limit number) or less (step S210).

  On the other hand, when the link setting range candidate extraction unit 120 determines that the reference word that is the anchor central word is not included in the i-th sentence (step S203: No), the reference that is the anchor-attached word in the i-th sentence. It is determined whether or not a word is included (step S204). When the link setting range candidate extraction unit 120 determines that the reference word that is an anchor attached word is included in the i-th sentence (step S204: Yes), the link setting range candidate extraction unit 120 uses the reference word that is an anchor attached word as a reference. As described above, a link setting range candidate pattern is extracted (step S220).

  That is, the link setting range candidate extraction unit 120, when the i-th sentence does not include the reference word that is the anchor center word but includes the reference word Wb that is the anchor attached word, the link purpose based on the reference word Wb − A list LWb that stores link setting range candidates is generated. More specifically, first, the link setting range candidate extraction unit 120 sets the word addition number s to 1, and among the link setting range candidates composed of s + 1 words, the link whose t-th word is the reference word Wb. A setting range candidate is added to LWb (processing 1). The link setting range candidate extraction unit 120 repeatedly executes the process 1 for t of 1 or more (s + 1) or less (process 2). Note that LWb is generated, for example, when the process 1 is executed for the first time. Furthermore, the link setting range candidate extraction unit 120 repeatedly executes the process 2 for s that is 1 or more and n (additional limit number) or less (step S220).

  On the other hand, when the link setting range candidate extraction unit 120 determines that the reference word that is an anchor-attached word is not included in the i-th sentence (step S204: No), the reference that is the anchor adjacent word in the i-th sentence. It is determined whether or not a word is included (step S205). When the link setting range candidate extraction unit 120 determines that the reference word that is the anchor adjacent word is included in the i-th sentence (step S205: Yes), the link setting range candidate extraction unit 120 described above based on the reference word that is the anchor adjacent word. As described above, a link setting range candidate pattern is extracted (step S230).

  That is, the link setting range candidate extraction unit 120, when the i-th sentence does not include the reference word that is the anchor central word and the reference word that is the anchor attached word but includes the reference word Wc that is the anchor adjacent word, the reference word A list LWc that stores link purpose-link setting range candidates based on the word immediately before or immediately after Wc is generated. More specifically, first, the link setting range candidate extraction unit 120 sets the word addition number s to 0, and among the link setting range candidates composed of s + 1 words, the s + 1st word is one of the reference words Wc. The link setting range candidate that is the previous or s + 1 subsequent word is added to the LWc (processing 1). The link setting range candidate extraction unit 120 repeatedly executes the process 1 for s not less than 0 and not more than n (additional limit number) (step S230). Note that the LWc is generated, for example, when the process 1 is executed for the first time.

  When the link setting range candidate extraction unit 120 extracts the link setting range candidate pattern based on the reference word that is the anchor center word (step S210), the link setting range candidate extraction unit 120 uses the reference word that is the anchor attached word as a reference. When the pattern is extracted (step S220), when the link setting range candidate pattern is extracted based on the reference word that is the anchor adjacent word (step S230), or the reference that is the anchor adjacent word in the i-th sentence When it is determined that the word is not included (step S205: No), the index i is counted up (step S206). Subsequently, the link setting range candidate extraction unit 120 determines whether there is a sentence with index i in the content being created (step S207). If the link setting range candidate extraction unit 120 determines that a sentence with index i exists (step S207: Yes), the link setting range candidate extraction unit 120 returns to step S202 and extracts the next sentence. On the other hand, if the link setting range candidate extraction unit 120 determines that the sentence with the index i does not exist (step S207: No), the link setting range candidate extraction unit 120 determines whether at least one link setting range candidate has been extracted (step S208).

  When the link setting range candidate extraction unit 120 determines that at least one link setting range candidate has been extracted (step S208: Yes), this flowchart ends. On the other hand, if the link setting range candidate extraction unit 120 determines that no link setting range candidates have been extracted (step S208: No), the link setting range candidate extraction unit 120 enters that the content being created does not have an appropriate location for link creation. The input / output unit 100 that has notified the output unit 100 and acquired the notification outputs a message (for example, “extraction number 0”) indicating that there is no appropriate location for link creation in the content being created (step S209). ). And this flowchart is complete | finished.

  The link candidate search unit 130 acquires a link setting range candidate for each link purpose candidate from the link setting range candidate extraction unit 120. The link destination candidate search unit 130 that has acquired the link setting range candidate for each link purpose candidate uses a search method according to the link purpose candidate and the link setting range candidate, as shown in FIG. For each set of setting range candidates, a link destination candidate that is a link destination candidate is searched. For example, in the example illustrated in FIG. 5, the link destination candidate search unit 130 sets link destination candidates “K1”, “K2”,..., “K5” for the set of the link purpose candidate “keyword commentary” and the link setting range candidate “A”. Searching.

  The search method according to the link purpose candidate and the link setting range candidate is to use different search conditions (search character strings) according to the link purpose candidate and the link setting range candidate when searching for the link destination candidate. That is, the link destination candidate search unit 130 generates different search conditions for each set of link purpose candidates and link setting range candidates, and uses link search candidates (contents that are link destination candidates) using a general-purpose search API of the external Internet. )

  For example, when the link destination candidate is “keyword commentary” and the link setting range candidate is “tire replacement”, the link destination candidate search unit 130 generates a search character string such as “What is tire replacement”, The contents including the set of search character strings (entire search character string) are searched.

  For example, when the link destination candidate is “reference to official site”, the link destination candidate search unit 130 searches the content including the search character string using the link setting range candidate as the search character string.

  For example, when the link destination candidate is “reference to the information source page”, the link candidate search unit 130 TF-IDF, which is an algorithm that extracts, for example, characteristic words or important words in sentences. (Term Frequency-Inverse Document Frequency) is used to calculate the importance of each word, and each word is included, for example, within the range of a link setting range candidate or a predetermined length character before and after the link setting range candidate Select a word (search string) to be used for the search based on a predetermined priority order corresponding to the location within the range of the column or the entire content being created. Search for content that includes at least one.

  For example, when the link destination candidate is “reference past article”, for example, the link destination candidate search unit 130 designates the past article DB 191 and is similar to the case where the link purpose candidate is “reference information source page”. Search for content. If the link setting range candidate or the content being created includes a word indicating time or time (for example, “last night”, “yesterday”, “last time”), the target may be narrowed down using the above time.

  When the link destination candidate searching unit 130 searches for a link destination candidate for each set of link purpose candidates and link setting range candidates, the link destination candidate searching unit 130 associates the link destination candidates with the link setting range candidates and specifies link destination candidates ( For example, URL) is supplied to the first determination unit 140. In other words, the link destination candidate search unit 130 supplies a set of link purpose candidates, link setting range candidates, and link destination candidates to the first determination unit 140.

  If the link destination candidate search unit 130 cannot search for more than a predetermined number of link destination candidates, the link destination candidate search unit 130 changes the search condition to search for a predetermined number of link destination candidates. May be.

  The first determination unit 140 acquires information (for example, URL) for specifying the link destination candidate in association with the link purpose candidate and the link setting range candidate from the link destination candidate search unit 130. In other words, the first determination unit 140 acquires a set of link purpose candidates, link setting range candidates, and link destination candidates. The 1st determination part 140 which acquired the group of a link purpose candidate, a link setting range candidate, and a link destination candidate is the 1st which shows the suitability as a link destination for every group of a link purpose candidate, a link setting range candidate, and a link destination candidate. The evaluation value is calculated, the first evaluation value is totaled for each link purpose candidate, and the link purpose is determined (adopted) from the link purpose candidates based on the first evaluation value after aggregation. Specifically, the first determination unit 140 determines the link purpose from the link purpose candidates for each group of link setting range candidates that are candidates for one link setting range for which one link destination is to be set. In other words, the first determination unit 140 determines a link purpose for each reference word that is used as a reference when extracting link setting range candidates (determines one link purpose for one link destination). A specific example relating to the first evaluation value will be described later.

  The first determination unit 140 may determine, for example, the top N link purpose candidates (N is 2 or more) in descending order of the first evaluation value after aggregation as the link purpose. Moreover, the 1st determination part 140 may determine a link objective candidate with the highest 1st evaluation value after totaling as a link objective.

  When the link purpose is determined from the link purpose candidates, the first determination unit 140 second determines the link purpose, the link setting range candidate corresponding to the link purpose, and information (for example, URL) for specifying the link destination candidate. To the unit 150. In other words, the first determination unit 140 supplies the second determination unit 150 with a set of the link purpose, the link setting range candidate, and the link destination candidate.

  That is, the first determination unit 140 targets all reference words of the entire content being created, and uses each reference word for the list L = [LW1, LW2,. The link purpose is determined for each [W1, W2,...], And the link destination-link setting range candidate pair and the link destination candidate corresponding to the link purpose-link setting range candidate pair are given to the second determining unit 150. Supply. For example, the first determination unit 140 determines a link purpose from a list LW1 that stores a link purpose candidate-link setting range candidate set related to the reference word W1, and the link purpose-link setting range candidate set and the link A link destination candidate corresponding to the set of purpose-link setting range candidates is supplied to the second determination unit 150.

  Hereinafter, the first evaluation value will be further described. The first determination unit 140 calculates a first evaluation value using the first prediction model generated by multivariate analysis. The first prediction model is a prediction model that does not specialize (depend on) the link purpose (link purpose candidate). Details of the first prediction model will be described later.

  For example, when processing is performed on the list LW extracted using the reference word Wa as a reference, first, the first determination unit 140 adds a set of lists LW (to the first prediction model stored in the prediction model DB 193 ( (Link target candidate-link setting range candidate) and a link destination candidate corresponding to the set are input, and a first evaluation value is acquired. For example, in the example illustrated in FIG. 5, the first determination unit 140 includes the URL “http: //www.k1/...” Of the link purpose candidate “keyword commentary”, the link setting range candidate “A”, and the link destination candidate “K1”. ”To obtain the first evaluation value“ 0.65 ”. The 1st determination part 140 performs the said process similarly about all the other groups in list | wrist LWa.

  Next, the first determination unit 140 calculates an average value of the first evaluation values for each link purpose candidate. For example, in the example illustrated in FIG. 5, the first determination unit 140 calculates the average value “0.43” of the first evaluation values related to the link purpose candidate “keyword commentary”. Next, the first determination unit 140 determines, for example, a link purpose candidate that is an average value of the highest first evaluation values as a link purpose. For example, in the example illustrated in FIG. 5, the first determination unit 140 determines the link purpose candidate “official site reference” having the average value “0.62” of the highest first evaluation value as the link purpose. As described above, the first determination unit 140 determines the link purpose “official site reference” related to the reference word Wa (list LWa). The 1st determination part 140 performs the said process similarly about all the other lists.

  Note that the first determination unit 140 may determine the link purpose when the average value of the first evaluation values is equal to or greater than a predetermined threshold value in order to improve accuracy. When there is no link purpose candidate that is equal to or greater than a predetermined threshold, the input / output unit 100 may be notified that there is no appropriate location for link creation in the content being created. The input / output unit 100 that has acquired the notification outputs a message indicating that there is no appropriate location for creating a link in the content being created.

  In addition, the 1st determination part 140 uses a matching degree with the structure pattern (a word label and a part of speech (for example, a noun, a particle)) of the link setting range classified by the link objective constructed | assembled previously in addition to a link candidate. It may be possible to improve accuracy and reduce the amount of calculation when determining the purpose. The degree of matching is a score indicating the degree of similarity between the link purpose candidate and the link setting range candidate with respect to the configuration pattern.

  The first evaluation value for each set of link purpose (link purpose candidate), link setting range candidate and link destination candidate, or the first evaluation value calculated (averaged) for each link purpose (link purpose candidate), It may be used when the second setting unit 150 determines the link setting range (described later).

  The second determination unit 150 acquires a set of a link purpose, a link setting range candidate, and a link destination candidate (information (for example, URL) specifying the link destination candidate) from the first determination unit 140. The second determination unit 150 that has acquired the set of the link purpose, the link setting range candidate, and the link destination candidate indicates the appropriateness as the link destination for each set of the link setting range candidate and the link destination candidate for the link purpose. The link setting range is determined (adopted) from the link setting range candidates based on the second evaluation value. Specifically, the second determination unit 140 determines a link setting range from the link setting range candidates for each group of link setting range candidates that are candidates for one link setting range for which one link destination is to be set. In other words, the second determination unit 150 determines a link setting range for each reference word that is used as a reference when extracting link setting range candidates (determines one link setting range for one link destination). A specific example relating to the second evaluation value will be described later.

  For example, the second determination unit 150 may determine the upper M link setting range candidates (M is 2 or more) in descending order of the second evaluation value as the link setting range. The second determination unit 150 may determine the link setting range candidate having the highest second evaluation value as the link setting range.

  When determining the link setting range from the link setting range candidates, the second determining unit 150 supplies the link setting range and the link purpose corresponding to the link setting range to the input / output unit 110. In other words, the second determination unit 150 supplies the link purpose to the input / output unit 110 in association with the link setting range. The input / output unit 110 that has acquired the link purpose in association with the link setting range from the second determination unit 150 presents the link purpose in association with the link setting range to the creator via a display unit (not shown) ( indicate.

  Further, when the second determination unit 150 determines the link setting range from the link setting range candidates, the link setting range, the link purpose corresponding to the link setting range, and the link destination (the link destination corresponding to the link setting range) Candidates) may be supplied to the input / output unit 110. In other words, the second determination unit 150 supplies the link purpose and the link destination to the input / output unit 110 in association with the link setting range. The input / output unit 110 that has acquired the link purpose and the link destination in association with the link setting range from the second determination unit 150 associates the link purpose and the link destination with the link setting range via a display unit (not shown). Is presented (displayed) to the creator.

  Hereinafter, the second evaluation value will be further described. The second determination unit 150 calculates a second evaluation value using the second prediction model generated by multivariate analysis. The second prediction model is a prediction model specialized (dependent) for the link purpose. Details of the second prediction model will be described later.

  For example, when processing the list LWa that includes only the set related to the link purpose determined by the first determination unit 140, first, the second determination unit 150 first stores the link purpose stored in the prediction model DB 193. A link setting range candidate of one set of the list LW (link purpose-link setting range candidate) and a link destination candidate corresponding to the link setting range are input to the second prediction model specialized for the second evaluation model. Get the value.

  However, when the second evaluation unit 150 uses the first evaluation value used in determining the link purpose as the second evaluation value, in the example illustrated in FIG. , The URL “http: //www.k21/...” Of the link setting range candidate “A” and the link destination candidate “K21” is input, and the first (second) evaluation value “0.65” is acquired. . The 2nd determination part 150 performs the said process similarly about all the other groups. In addition, the 2nd determination part 150 may acquire the applicable 1st evaluation value from the 1st determination part 140, when utilizing a 1st evaluation value as a 2nd evaluation value.

  Next, the second determination unit 150 determines, for example, a set of the link setting range candidate and the link destination candidate that is the highest second evaluation value as the link setting range and the link destination. The second determination unit 150 performs the above process in the same manner for all other lists.

  Note that the second determination unit 150 may determine the link setting range (and the link destination) when the average value of the second evaluation values is equal to or greater than a predetermined threshold value in order to improve accuracy. When there is no link setting range (and a link destination) that is equal to or greater than a predetermined threshold, the input / output unit 100 may be notified that there is no location suitable for link creation in the content being created. The input / output unit 100 that has acquired the notification outputs a message indicating that there is no appropriate location for creating a link in the content being created.

  Next, the first prediction model and the second prediction model will be described. The first prediction model and the second prediction model are generated by the prediction model generation unit 160. The prediction model generation unit 160 generates a first prediction model and a second prediction model using the prediction model learning data. The prediction model learning data is stored, for example, in the externally acquired content DB 192.

FIG. 6 is an example of prediction model learning data. In the prediction model learning data shown in FIG. 6, one link destination candidate for a link (and an anchor text) existing in existing content corresponds to one entry. Each column is as follows.
(Content of each column)
“Link destination candidate URL”: URL of the link destination candidate acquired by the search
"Link purpose": Link purpose related to each link destination candidate (given manually)
"Evaluation value": Appropriateness of each link destination candidate (given manually)
"Scale": Content feature amount of link destination candidate calculated from link source and link destination candidate content

  The freshness shown in FIG. 6 is a measure representing the novelty of the content. The freshness value is higher for a link destination candidate with a new last update date. For the last update date, for example, refer to the Last-Modified field value included in the HTTP header of the link destination candidate. The reliability is a measure representing the reliability of the content of the link destination candidate. The reliability value is higher for link destination candidates belonging to Web sites that are considered to be reliable as information sources. For example, link destination candidates belonging to a trusted domain (such as a government domain such as “go.jp”) are high. The similarity is a scale representing the similarity between the content text of the link source and the content text of the link destination candidate. The similarity value is higher for link destination candidates whose contents are similar to the link source content. The URL / title matching degree is a scale representing the degree of matching between the word included in the link source anchor text and the link destination candidate URL or title. The URL / title matching value is higher as the number of words included in the link source anchor text is included in the URL or title of the link destination candidate.

  The prediction model generation unit 160 constructs the first prediction model and the second prediction model by performing multivariate analysis using the evaluation value as an objective variable and the scale value as an explanatory variable. Various methods such as multiple regression analysis, SVM (Support Vector Machine), and NN (Neural Network) can be used as multivariate analysis. An example of multiple regression analysis will be described. When the values of the explanatory variables (scale values) are f, r, s, and u, and the respective coefficients (degrees of importance) are a, b, c, and d, the link destination candidate that is the objective variable. The evaluation value y shall be represented by the following formula (1).

y = a * f + b * r + c * s + d * u + i (1)
However, i is an intercept.

  The prediction model generation unit 160 calculates the weight (a, b, c, d) of each scale by performing multiple regression analysis on the learning data. The prediction model is a mathematical formula using the calculated weight. Specifically, the prediction model generation unit 160 constructs a first prediction model by calculating weights using the entire learning data. For example, the prediction model generation unit 160 performs the process of No. A 1st prediction model is constructed | assembled using the data of 1-8. Further, the prediction model generation unit 160 constructs a second prediction model by calculating a weight using the learning data specific to the link purpose. For example, the prediction model generation unit 160 performs the process of No. The second prediction model for the link purpose “keyword commentary” is constructed using the data of 1 to 4, and the prediction model generation unit 160 displays the No. 1 in FIG. The second prediction model for the link purpose “official site reference” is constructed using the data of 5-8.

  Subsequently, the operation of the link creation support 10 will be described with reference to FIG. Note that the flowchart shown in FIG. 7 starts when the reference word extraction unit 110 acquires content being created from the input / output unit 100. The reference word extraction unit 110 extracts a reference word that serves as a reference when extracting link setting range candidates from the content being created (step S100). The reference word extraction unit 110 that has extracted the reference words supplies the reference word (with label) to the link setting range candidate extraction unit 120 together with the content being created.

  The link setting range candidate extraction unit 120 that has acquired the reference word (with label) together with the content being created extracts the link setting range candidate from the content being created (step S200). This process is as shown in FIG. The link setting range candidate extraction unit 120 that has extracted the link setting range candidates supplies the link destination candidate-link setting range candidate pair to the link destination candidate search unit 130.

  The link destination candidate search unit 130 that has acquired the link target candidate-link setting range candidate pair searches for the link destination candidate for each set of the link purpose candidate and the link setting range candidate (step S300). The link destination candidate search unit 130 that has searched for the link destination candidate for each set of the link purpose candidate and the link setting range candidate, the link destination candidate search unit 130 determines the set of the link purpose candidate, the link setting range candidate, and the link destination candidate. 1 is supplied to the determination unit 140.

  The 1st determination part 140 which acquired the group of a link purpose candidate, a link setting range candidate, and a link destination candidate determines a link purpose for every reference | standard word used as the reference at the time of link setting range candidate extraction (step S400). The first determination unit 140 that has determined the link purpose for each reference word supplies the second determination unit 150 with a set of the link purpose, the link setting range candidate, and the link destination candidate.

  The 2nd determination part 150 which acquired the group of a link purpose, a link setting range candidate, and a link destination candidate determines a link setting range for every reference | standard word used as the reference | standard at the time of link setting range candidate extraction (step S500). The second determination unit 150 that has determined the link setting range for each reference word supplies the link purpose and the link destination to the input / output unit 110 in association with the link setting range.

  The input / output unit 110 that has acquired the link purpose and link destination in association with the link setting range outputs (displays) the link purpose and link destination in association with the link setting range to the creator via a display unit (not shown). (Step S600).

  As described above, in the text-based content production process such as a blog, the link creation support apparatus 10 estimates a range for creating a link in the content, a purpose for creating the link within the range, and a range for creating the link. Is an automatic link creation device that automatically presents According to the link creation support apparatus 10, a link (link setting range, link purpose / link setting range, link purpose, link destination) that follows the link purpose of the content creator is presented to the creator. The burden on the creator of can be greatly reduced. Depending on the creation style of the creator, one link purpose, one link setting range, and one link destination may be presented to the creator for a certain reference word. A plurality of link setting ranges and a link destination corresponding to each link setting range may be presented to the creator.

(Second Embodiment)
The link creation support 12 according to the second embodiment of the present invention will be described below. The link creation support 12 is a device that supports content creation, and automatically creates a link in the content being created that meets the link purpose of the content creator.

  Hereinafter, the link creation support 12 will be described with reference to the drawings. As shown in FIG. 8, the link creation support 12 includes an input / output unit 100, a reference word extraction unit 110, a link setting range candidate extraction unit 120, a link destination candidate search unit 130, a first determination unit 140, and a second determination unit 151. A prediction model generation unit 160, an anchor text setting unit 170, a word score calculation unit 180, a word DB 190, a past article DB 191, an externally acquired content DB 192, and a prediction model DB 193. Other than the second determination unit 151 and the anchor text setting unit 170 are the same as the link creation support 10 shown in FIG.

  Similar to the second determination unit 150, the second determination unit 151 determines the link setting range from the link setting range candidates. However, the second determination unit 151 determines one link setting range from the link setting range candidates based on the second evaluation value. That is, the second determination unit 151 determines the link setting range candidate having the highest second evaluation value as the link setting range.

  When determining the link setting range from the link setting range candidates, the second determining unit 151 anchors the link setting range and the link destination corresponding to the link setting range (link destination candidate corresponding to the link setting range). This is supplied to the text setting unit 170. In other words, the second determination unit 151 supplies the link destination to the anchor text setting unit 170 in association with the link setting range.

  Further, when the second setting unit 151 determines the link setting range from the link setting range candidates, the second determining unit 151 determines the link setting range, the link destination corresponding to the link setting range, and the link purpose corresponding to the link setting range. You may supply to the anchor text setting part 170. FIG. In other words, the second determination unit 151 may supply the link destination and the link purpose to the anchor text setting unit 170 in association with the link setting range.

  For example, the anchor text setting unit 170 acquires the content being created from the input / output unit 100 or the reference word extraction unit 110. In addition, the anchor text setting unit 170 acquires a link destination from the second determination unit 151 in association with the link setting range. The anchor text setting unit 170 may acquire the link destination and the link purpose from the second determination unit 151 in association with the link setting range.

  The anchor text setting unit 170 that has acquired the link destination in association with the link setting range sets the link destination in the link setting range in the content being created. In addition, the anchor text setting unit 170 that has acquired the link destination and the link purpose in association with the link setting range sets the link destination in the link setting range and in the vicinity of the link setting range in the content being created. The link purpose is displayed (for example, it may be displayed when the cursor is placed).

  FIG. 9 is a flowchart illustrating an example of the operation of the link creation support 12. Steps S100, S200, S300, S400, and S500 in FIG. 9 are the same as the operations of the link creation support 10 shown in FIG. As shown in FIG. 9, in the case of the link creation support 12, the anchor text setting unit 170 sets a link destination in the link setting range in the content being created (step S610). Further, as described above, the anchor text setting unit 170 displays the link purpose in the vicinity of the link setting range.

  As described above, in the text-based content production process such as a blog, the link creation support apparatus 12 estimates the range for creating a link in the content and the purpose for creating the link within the range, and automatically This is an automatic link creation apparatus that creates a link (sets a range for creating a link and a link destination based on the range). According to the link creation support device 12, a link (link setting range, link destination) that matches the link purpose of the content creator is automatically created in the content being created. The burden can be further greatly reduced.

  As described above, according to the link creation support apparatus 10 or the link creation support apparatus 12, the creator needs complicated operations such as selection of anchor text, search and selection of link destination content, and URL transcription when creating a link. Therefore, the burden of creating links can be eliminated (or reduced) and content creation can be focused. Even if the content is vague at the start of content creation, a specific description can be supported by referring to the anchor text-link purpose-link destination content presented by the present invention. Since content related to the content is automatically presented, the creator can find an error in the content created by the creator by referring to the content. This makes it possible to prevent incorrect information from being disclosed on the Web. Since the present invention recommends contents suitable for each link purpose, the contents can be enhanced by creating links to these contents. By creating a large number of contents using the present invention, it is possible to increase the proportion of contents that are enriched among contents existing on the Web. This also contributes to an improvement in satisfaction with a normal Web search.

  The techniques described in Patent Document 1 and Non-Patent Document 1 do not mainly have the concept of a link purpose. Therefore, an optimal link position and link destination that are in accordance with the intention of the creator, or optimal link generation Is very difficult.

  Note that a program for executing each process of the link creation support apparatuses 10 and 12 according to an embodiment of the present invention is recorded on a computer-readable recording medium, and the program recorded on the recording medium is read into a computer system. The above-described various processes related to the processes of the link creation support apparatuses 10 and 12 according to the embodiment of the present invention may be performed. Here, the “computer system” may include an OS and hardware such as peripheral devices. Further, the “computer system” includes a homepage providing environment (or display environment) if a WWW system is used. The “computer-readable recording medium” means a flexible disk, a magneto-optical disk, a ROM, a writable nonvolatile memory such as a flash memory, a portable medium such as a CD-ROM, a hard disk built in a computer system, etc. This is a storage device.

  Further, the “computer-readable recording medium” means a volatile memory (for example, DRAM (Dynamic DRAM) in a computer system that becomes a server or a client when a program is transmitted through a network such as the Internet or a communication line such as a telephone line. Random Access Memory)), etc., which hold programs for a certain period of time. The program may be transmitted from a computer system storing the program in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the “transmission medium” for transmitting the program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line. The program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, and what is called a difference file (difference program) may be sufficient.

  The embodiment of the present invention has been described in detail with reference to the drawings. However, the specific configuration is not limited to this embodiment, and includes designs and the like that do not depart from the gist of the present invention.

10, 12 Link creation support device 100 Input / output unit 110 Reference word extraction unit 120 Link setting range candidate extraction unit 130 Link destination candidate search unit 140 First determination unit 150, 151 Second determination unit 160 Prediction model generation unit 170 Anchor test setting Unit 180 Word score calculation unit 190 Word DB 191 Past article DB 192 Externally acquired content DB 193 Prediction model DB

Claims (12)

A reference word extraction unit that extracts a reference word that serves as a reference when extracting a link setting range candidate that is a candidate of a link setting range for which a link destination is to be set as an anchor text from the content being created;
A link setting range candidate extraction unit that extracts one or more link setting range candidates for each link purpose candidate that is a link purpose candidate that is the purpose of link creation from the content being created, based on the reference word When,
Link destination candidate search for searching for a link destination candidate as a link destination candidate for each set of the link purpose candidate and the link setting range candidate using a search method according to the link purpose candidate and the link setting range candidate And
A first evaluation value indicating appropriateness as a link destination is calculated for each set of the link purpose candidate, the link setting range candidate, and the link destination candidate, and the first evaluation value is aggregated for each link purpose candidate. A first determination unit that determines the link purpose from the link purpose candidates based on the first evaluation value after the aggregation;
In the link purpose determined by the first determination unit, a second evaluation value indicating suitability as a link destination is calculated for each set of the link setting range candidate and the link destination candidate, and the second evaluation is performed. A link creation support apparatus comprising: a second determination unit that determines the link setting range from the link setting range candidates based on a value. The first determination unit includes:
For each group of link setting range candidates that are candidates for one link setting range to be set to one link destination, the link purpose is determined from the link purpose candidates,
The second determination unit includes
The link setting range is determined from the link setting range candidates for each group of the link setting range candidates that are candidates for the one link setting range to be set as one link destination. The link creation support device described. The first determination unit includes:
The link purpose candidate having the highest first evaluation value after the aggregation is determined as the link purpose,
The second determination unit includes
The link creation support apparatus according to claim 1 or 2, wherein the link setting range candidate of the set having the highest second evaluation value is determined as the link setting range. The first determination unit includes:
Using the first prediction model generated by multivariate analysis, calculating the first evaluation value,
The second determination unit includes
The link creation support apparatus according to any one of claims 1 to 3, wherein the second evaluation value is calculated using a second prediction model generated by multivariate analysis. The link setting range candidate extraction unit
5. The link creation support apparatus according to claim 1, wherein the link setting range candidates having a different number of words are extracted according to the link purpose candidates. The reference word extraction unit
6. The reference word according to claim 1, wherein the reference word is extracted based on a word score indicating a frequency at which one word is included in the link setting range in content including the word. The link creation support device described in 1. The link setting range candidate extraction unit
The link creation support apparatus according to claim 6, wherein the link setting range candidates having a different number of words are extracted according to the word score. The word score is
8. The link creation support apparatus according to claim 6, wherein the link creation support apparatus is attached to each set of a character string and a word attribute constituting the word. The word score is
7. The link setting range in the content created by the user is given a higher score than the case of the link setting range in the content created by another person. The link creation support apparatus according to claim 8. An anchor text setting unit that sets a link destination in the link setting range in the content being created,
The second determination unit includes
Determining one link setting range from the link setting range candidates based on the second evaluation value, and supplying one link destination candidate corresponding to the link setting range to the anchor text setting unit;
The anchor text setting unit
The link creation support apparatus according to claim 1, wherein the link destination candidate acquired from the second determination unit is set in the link setting range acquired from the second determination unit. A reference word extraction unit that extracts a reference word that serves as a reference when extracting a link setting range candidate that is a candidate of a link setting range for which a link destination is to be set as an anchor text from the content being created;
Link setting range candidate extraction means for extracting one or more link setting range candidates for each link purpose candidate that is a link purpose candidate that is the purpose of link creation from the content being created with reference to the reference word When,
Link destination candidate search for searching for a link destination candidate as a link destination candidate for each set of the link purpose candidate and the link setting range candidate using a search method according to the link purpose candidate and the link setting range candidate Means,
A first evaluation value indicating appropriateness as a link destination is calculated for each set of the link purpose candidate, the link setting range candidate, and the link destination candidate, and the first evaluation value is aggregated for each link purpose candidate. First determining means for determining the link purpose from the link purpose candidates based on the first evaluation value after the aggregation;
For the link purpose determined by the first determining means, a second evaluation value indicating suitability as a link destination is calculated for each set of the link setting range candidate and the link destination candidate, and the second evaluation A link creation support method comprising: a second determination unit that determines the link setting range from the link setting range candidates based on a value. In the computer of the link creation support device that supports link creation,
A reference word extraction step for extracting a reference word that serves as a reference when extracting a link setting range candidate to be a link setting range candidate for which a link destination should be set as an anchor text from within the content being created;
A link setting range candidate extraction step for extracting one or more link setting range candidates for each link purpose candidate that is a link purpose candidate that is a purpose of link creation from the content being created with reference to the reference word When,
Link destination candidate search for searching for a link destination candidate as a link destination candidate for each set of the link purpose candidate and the link setting range candidate using a search method according to the link purpose candidate and the link setting range candidate Steps,
A first evaluation value indicating appropriateness as a link destination is calculated for each set of the link purpose candidate, the link setting range candidate, and the link destination candidate, and the first evaluation value is aggregated for each link purpose candidate. A first determination step of determining the link purpose from the link purpose candidates based on the first evaluation value after the aggregation;
In the link purpose determined by the first determination step, a second evaluation value indicating suitability as a link destination is calculated for each set of the link setting range candidate and the link destination candidate, and the second evaluation And a second determination step of determining the link setting range from the link setting range candidates based on a value.

Images