JP5981386B2 - Representative page selection device and representative page selection program - Google Patents

Description

  The present invention relates to a technique for selecting a representative page representing a specified keyword and region.

  There is a need for a technique for selecting and presenting a representative page related to a keyword in the area from a document group from the specified keyword and area. For example, if “Ramen” is specified as the keyword and “Yokohama” is specified as the region, the menu page of the store that is famous for Ramen in Yokohama, the route to the store, and the official page with the map are presented as the representative page. It is.

  As a representative page selection / presentation method, there is a method using a general search engine on the Internet. This is a method in which a region converted into a character string such as a place name is added as a query and inputted as a query, and a higher page is selected and presented from a plurality of pages obtained as a search result.

  Further, according to Patent Document 1, a geographical search for searching a document from a document group based on the geographical distance from a specified keyword and position and the importance of a small area including the name of the place that appears in the document. A document search method has been proposed, and there is also a method of selecting and presenting the top of the document obtained from the search result.

Japanese Patent No. 5087377

  However, when using a general search engine, the place name entered as a query is treated as just one keyword as well as the specified keyword, so it is geographically close to the place name that appears in the document. There was a problem that was not considered.

  In this regard, in the above-described geographical document search method, a document is searched based on a geographical distance from a place name appearing in the document. By using this method, the geographical proximity to a specified region is used. Can be searched. However, in other cases, the documents are simply ranked based on the importance related to the keyword and the position, and therefore, it is not always possible to detect a document suitable for the representative. In other words, even if this method is used, there is a problem that the representativeness of the page is not considered.

  In addition, as a problem common to both of them, there is a problem that it is necessary to collect a large number of web pages and create an index.

  The present invention has been made in view of the above circumstances, and does not require collection and indexing of a large number of documents, considering the geographical proximity to a designated area, and the typicality of a page. The purpose is to select a considered representative page.

  The representative page selection device according to claim 1 is a part in which a document related to a designated keyword and a region is stored in a storage unit based on a geographical measure of a designated region and a place name included in the document. The search means for searching from the document group, the reference destination referred to by the document as a representative page candidate, the approximation of the character string between the keyword and each representative page candidate, and the place name included in the reference source document And a selection means for selecting one or more representative pages based on any one or more of the score, rank, or number of the document given based on the importance of the corresponding region. To do.

  The representative page selection device according to claim 2 is the representative page selection device according to claim 1, characterized in that the selection means calculates an approximation degree of the character string using an edit distance.

  The representative page selection device according to claim 3 is the representative page selection device according to claim 1 or 2, wherein the representative page selection device is based on a matching degree between a URL of a related page related to the keyword and a URL of each representative page candidate. The gist is to reduce the number of page candidates.

  A representative page selection apparatus according to a fourth aspect is the representative page selection apparatus according to the third aspect, wherein the related page includes a related page related to a related keyword related to the keyword.

  The gist of the representative page selection program according to claim 5 is to cause a computer to function as the representative page selection device according to any one of claims 1 to 4.

  As described above, according to the present invention, in order to search the document group for documents related to the specified keyword and the area based on the geographical scale of the specified area and the place name included in the document, the specified area and The representative page can be selected in consideration of the geographical proximity.

  Further, according to the present invention, since a document related to the specified keyword and region is searched from a part of the document group, it is possible to eliminate the necessity of processing for collecting and indexing a large number of document groups. it can.

  Further, according to the present invention, the reference destination referred to by the retrieved document is set as a representative page candidate, and the approximation degree of the character string between the designated keyword and each representative page candidate is included in the reference source document. Since one or more representative pages are selected based on any one or more of the score, rank, or number of the document assigned based on the importance of the area corresponding to the place name, the representativeness of the page is considered. Selected representative page can be selected.

  According to the present invention, it is possible to select a representative page that takes into consideration the geographical proximity to a designated area and does not require collection and indexing of a large amount of documents, and considers the representativeness of the page.

It is a figure which shows the functional block structure of a representative page selection apparatus. It is a figure which shows the operation | movement flow of a representative page selection apparatus. It is a figure which shows the example of keyword DB. It is a figure which shows the example of keyword related page DB. It is a figure which shows the example of a document acquisition result. It is a figure which shows the example of reference destination page DB. It is a figure which shows the example of the narrowing-down result of a representative page candidate. It is a figure which shows the example of a 1st representative page score. It is a figure which shows the example of a 2nd representative page score.

  Hereinafter, an embodiment for carrying out the present invention will be described with reference to the drawings.

  FIG. 1 is a diagram illustrating a functional block configuration of the representative page selection device 100. The representative page selection apparatus 100 is an apparatus that receives keyword information and area information designated by the user as input, selects a representative page related to the keyword in the area, and displays it on the screen.

  The keyword is a word that is used as a clue to search for things and information that the user is interested in and wants to know. A region is a range on a map divided by administrative districts and coordinates. The representative page is a page that represents a keyword such as detailed information related to the keyword.

  As shown in FIG. 1, the representative page selection device 100 includes a related keyword acquisition unit 1, a keyword related page acquisition unit 2, a document acquisition unit 3, a representative page candidate acquisition unit 4, and a representative page candidate narrowing unit. 5, a representative page selection unit 6, a keyword DB 7, a keyword related page DB 8, a search index DB 9, and a reference page DB 10. The function will be described below.

  The keyword DB 7 has a function of storing a correspondence table in which a plurality of keywords and related keywords related to the keywords are associated with each other. For example, it is conceivable to extract a noun phrase from a document in which an explanation about the keyword is written and use it as a related keyword.

  The keyword related page DB 8 has a function of storing a correspondence table in which a plurality of keywords including related keywords are associated with related pages (document URLs) related to the keywords. For example, if a page with a detailed explanation is introduced in a document with explanations about keywords, make it a related page, or use the company name as a keyword and the company's official page as a related page. It is possible to do.

  The search index DB 9 has a function of storing a search index created by using the geographical document search method disclosed in Patent Document 1 from a specified keyword and region.

  The reference page DB 10 has a function of storing a correspondence table that associates document identifiers for identifying a plurality of documents with reference destination pages (reference destination URLs) referred to from the respective documents. Yes. For example, when the document is described in HTML, it is conceivable to analyze the document and extract the reference destination page from the information of the anchor tag.

  The related keyword acquisition unit 1 has a function of referring to the keyword DB 7 and acquiring a related keyword related to the input keyword.

  The keyword related page acquisition unit 2 has a function of referring to the keyword related page DB 8 and acquiring a related page related to the input keyword and the related keyword acquired by the related keyword acquisition unit 1 as a keyword related page. Yes.

  The document acquisition unit 3 has a function of referring to the search index DB 9 and searching for and acquiring a document related to the input keyword and region using the geographical document search method described above.

  The representative page candidate acquisition unit 4 has a function of referring to the reference destination page DB 10 and acquiring the reference destination page referred to by the document acquired by the document acquisition unit 3 as a representative page candidate.

  The representative page candidate narrowing-down unit 5 is based on the matching degree between the URL of the keyword related page acquired by the keyword related page acquisition unit 2 and the URL of each representative page candidate acquired by the representative page candidate acquisition unit 4. A function to reduce (narrow) the number of page candidates is provided.

  The representative page selection unit 6 determines the score of each representative page candidate that has been narrowed down based on the approximation of the character string between the input keyword and each representative page candidate, and the document search result performed by the document acquisition unit 3. Are calculated, and one or more representative page candidates having high scores are selected as representative pages.

  Next, the operation of the representative page selection device 100 will be described with reference to FIG. In this operation example, “Ferris wheel” is designated as a keyword, and “Yokohama” is designated as an area.

  Initially, the related keyword acquisition part 1 acquires the related keyword corresponding to the input keyword from keyword DB7 (step S1). When the keyword DB 7 is shown in FIG. 3, “gondola”, “Chicago”, “XX amusement park”, and “ΔΔ land” are acquired for the keyword “Ferris wheel”.

  Next, the keyword related page acquisition unit 2 acquires a keyword related page corresponding to the input keyword and the related keyword acquired in step S1 from the keyword related page DB 8 (step S2). When the keyword-related page DB 8 is shown in FIG. 4, six document URLs are acquired.

  Next, the document acquisition unit 3 uses the search index DB 9 to search and acquire a document related to the input keyword and region (step S3).

  Specifically, a document is acquired by creating a search index using the geographical document search method described above and performing a document search. Since this geographical document search method searches for a document based on a geographical measure between a designated area and a place name included in the document, the geographical proximity to the input area is considered.

  In addition, since this geographic document search method assigns scores and ranks to the searched documents based on the importance of the area corresponding to the place name included in the document, the document acquisition result as shown in FIG. 5 is obtained. It will be. As a result, five documents D1, D2, D3, D4, and D5 are acquired.

  Note that web pages stored in a storage device (storage means such as a hard disk) on the Internet may be directly searched, and only a part (for example, blog articles) of various web pages is searched. Therefore, it is possible to eliminate the necessity of collecting and indexing a large number of web page groups. Any document search / acquisition method may be used as long as it can acquire a document related to a keyword and a region.

  Next, the representative page candidate acquisition unit 4 acquires the reference page corresponding to the document acquired in step S3 from the reference page DB 10 as a representative page candidate (step S4). Referring to the reference page DB 10 in FIG. 6, seven reference pages are acquired for the five acquired documents, and these are used as representative page candidates.

  Next, the representative page candidate narrowing-down unit 5 narrows down the number of representative page candidates acquired in step S4 using the keyword-related page acquired in step S2 (step S5).

  Specifically, the search is narrowed down to representative page candidates such that any one of the URLs of the keyword related pages matches the URL of the representative page candidate. When the seven representative page candidates in FIG. 6 are checked for forward matching by referring to the six keyword-related pages in FIG. 4, the representative page candidates are narrowed down to three as shown in FIG. Note that the narrowing down method is not limited to this, and narrowing down may be performed by perfect matching or narrowing down to only those that match only the domain.

  Finally, the representative page selection unit 6 calculates the representative page candidate score as the representative page score based on the proximity of the character string between the input keyword and the representative page candidate and the document search result performed in step S3. Then, the high score is selected as the representative page (step S6).

  Here, the product of the first representative page score obtained from the proximity of the character string of the keyword and the representative page candidate and the second representative page score obtained from the document search result is calculated as the representative page score. The method for calculating the representative page score is not limited to this, and a sum may be calculated instead of a product.

  First, the first representative page score is obtained from the closeness of the character string between the keyword and the representative page candidate. As a method of calculating the closeness of the character string between the keyword and the representative page candidate, a character string obtained by converting the keyword reading into a Roman character (keyword character string), and a character string of the URL of the representative page candidate (URL character string) The edit distance (the cost of the procedure necessary to convert one character string into the other character string) is calculated. The keyword character string may be a character string in which the keyword is translated into English.

  The insertion cost at the time of conversion from the keyword character string to the URL character string is set to “0”. As a result, it is possible to avoid the influence of a part not related to the keyword included in the URL character string. The insertion cost may not be “0” as long as it is a small value.

  In order to absorb the difference between Japanese and foreign language, the replacement cost from “k” to “c” is set to “0”. As a result, the edit distance between “kanransha” in which the keyword is converted into Roman characters and “canransha” included in the URL character string of the representative page candidate can be reduced. If the replacement cost is also a small value, it may not be “0”.

  Then, the reciprocal of the value obtained by adding “1” to the edit distance is calculated as the first representative page score. The first representative page score may be calculated as long as the edit distance is small. When the edit distance from the keyword for the representative page candidate of FIG. 7 is obtained, it becomes 6, 0, 0, respectively, so the first representative page score is as shown in FIG.

  Next, a second representative page score is obtained from the document search result. The document search result is information related to the document searched and acquired in step S3. Specifically, it refers to the document score and document rank assigned based on the importance of the area corresponding to the place name included in the reference source document, the number of reference source documents, and the like.

  Here, the document score is used as the document search result, and the sum of the document scores of the documents including the representative page candidates is set as the second representative page score. FIG. 9 shows the result of setting the sum of the document scores of the documents including the representative page candidates as the second representative page score for the representative page candidates in FIG.

  For example, in the case of the representative page candidate of number 2, there are two documents including the representative page candidate D2 and D3 from FIG. 6, and the document scores of these documents are 3.0 and 2.0 respectively from FIG. Therefore, the sum 5.0 is the second representative page score. The same applies to the rest.

  The second representative page score may be calculated using a document search result. For example, the number of documents including the representative page candidate is set as the second representative page score, the sum of the reciprocal of the document order of the document including the representative page is set as the second representative page score, or scores related to various document information are combined. The second representative page score may be used.

  When the representative page score is calculated based on FIGS. 8 and 9, the representative page score of number 1 is 0.644, the representative page score of number 2 is 5.0, and the representative page score of number 3 is 1.2. It becomes. Based on this, a representative page is selected.

  Here, the one with the highest representative page score is selected as the representative page. Since the representative page candidate with the highest representative page score is number 2, the representative page candidate is selected as the representative page.

  Note that the method for selecting the representative page is not particularly limited as long as the one having a high score is selected. For example, the top N cases may be selected or the top X% may be selected.

  Since the representative page selected in this way is obtained from a search result document using the geographical document search method, the geographical proximity of the place is considered. In addition, since the selection is based on the proximity of the character string between the keyword and the representative page candidate and the search result of the document, the representativeness of the page is taken into consideration. Further, since the document of the representative page itself does not need to be collected and indexed, the representative page can be selected from a small amount of documents.

  As described above, according to the present embodiment, a document related to a specified keyword and a region is searched from a document group based on a geographical scale between a specified region and a place name included in the document. A representative page that takes into account the geographical proximity to the region can be selected.

  In addition, according to the present embodiment, since a document related to a specified keyword and region is searched from a part of the document group, it is not necessary to collect and index a large number of document groups. be able to.

  Further, according to the present embodiment, the reference destination referred to by the retrieved document is set as the representative page candidate, the approximation degree of the character string between the designated keyword and each representative page candidate, and the reference source document Since one or more representative pages are selected based on any one or more of the score, rank, or number of the document given based on the importance of the area corresponding to the included place name, the representativeness of the page Can be selected representative page.

  Finally, the representative page selection apparatus 100 described in the present embodiment can be realized by a computer having a memory and a CPU. The processing of each functional unit can be realized by a program that can execute them.

DESCRIPTION OF SYMBOLS 100 ... Representative page selection apparatus 1 ... Related keyword acquisition part 2 ... Keyword related page acquisition part 3 ... Document acquisition part 4 ... Representative page candidate acquisition part 5 ... Representative page candidate narrowing part 6 ... Representative page selection part 7 ... Keyword DB
8 ... Keyword related page DB
9 ... Search index DB
10 ... Reference page DB
S1-S6 ... Step

Claims (5)

A search means for searching a document related to the specified keyword and the area from a part of the document group stored in the storage means based on a geographical measure of the specified area and the place name included in the document;
The reference destination referred to by the document is set as a representative page candidate, and is assigned based on the similarity of the character string between the keyword and each representative page candidate, and the importance of the area corresponding to the place name included in the reference source document Selection means for selecting one or more representative pages based on any one or more of the score, rank, or number of the document
A representative page selection device. The selection means includes
The representative page selection apparatus according to claim 1, wherein the approximation degree of the character string is calculated using an edit distance.   3. The representative page selection device according to claim 1, wherein the number of representative page candidates is reduced based on a degree of match between a URL of a related page related to the keyword and a URL of each representative page candidate. In the related page,
The representative page selection device according to claim 3, wherein a related page related to a related keyword related to the keyword is included.   A representative page selection program for causing a computer to function as the representative page selection device according to any one of claims 1 to 4.

Images