JP2009211280A - Method, apparatus, and program for analyzing online page, and computer readable recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To analyze to express numerically intentions of a creator and a commentator of an online page and feature of characteristics of the page without depending on the semantic contents of the page. <P>SOLUTION: A method of analyzing an online page acquires an article part and a comment part of the page to calculate the number of article characters, the number of article pictograms, the number of comment averaged pictograms, and associates these numbers with page IDs to calculate a score on the basis of the number of characters calculated for each page ID, calculates, by using the score, the feature value of the article indicating whether the creator and the commentator give weight to accurately notify information or to notify with rich feelings to store the feature value in a feature value storage means, and retrieves the feature value of page IDs specified by an input query, or the page IDs having the feature value specified by the query from the storage means to output the retrieved feature value or the page IDs. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an online page analysis method, apparatus, program, and computer-readable recording medium, and more particularly, to an online page analysis method, apparatus, program, and computer-readable recording medium using the number of characters / pictograms.

  For details, please refer to the number of characters / pictograms included in the articles on the online page, and the number of characters / pictograms included in the comments when comments are attached to the articles, to analyze the features of online pages. The present invention relates to a page analysis method and apparatus, a program, and a computer-readable recording medium.

  In recent years, services such as blogs and SNS (Social Networking Service) that can easily post articles written by users as online pages have become widespread. Most of the authors of these articles are general users who are not related to the mass media. However, experience information and reviews unique to general users are often described, and a lot of useful information is buried. I can say that. In particular, companies that provide merchandise to general consumers are increasingly demanding to mine these online pages in order to grasp the reputation and reviews of specific merchandise.

  There are several ways to find useful information from these online pages. For example, as represented by Google, there is a method of determining the usefulness of the online page by regarding the number of linked pages of the online page as support votes from others (for example, see Non-Patent Document 1).

In addition, there is a method for evaluating the online page by using information such as the number of comments given to the online page and the number of comments given (for example, see Non-Patent Document 2).
L. Page. S. Brin, R. Motwani, and T. Winograd, "The PageRank Citation Ranking: Bringing Order to the Web". Technical report, Stanford Digital Library Technologies Project, 1998 Akihiro Miyata, Toshinobu Matsuoka, Shinichi Okano, Seto Yamada, Tomomi Ishiuchi, Noriyasu Arakawa, Yasuhisa Kato, “Blog Article Retrieval Method Using Resonance Characteristics Analysis”, Transactions of Information Processing Society of Japan, Vol. 48, No12, pp. 4041- 4050 (2007).

  As described above, online pages such as blogs and SNS contain valuable information such as reputation and word of mouth for third parties. On the other hand, these online pages contain a lot of diary and inner ring information that only the creator of the article or his friend can understand. For this reason, it is necessary to improve the accuracy of online page search / mining techniques, and there is an increasing demand for techniques for appropriately classifying these various types of information.

  For example, as shown in FIG. 15, there are an article A and an article B that include the same keyword (“pension” in the example in the figure), and in the article A, useful information for a third party such as a deep consideration about current affairs is written. Assume that article B contains information that is not useful to a third party, such as an individual's daily events. In this case, article A and article B cannot be distinguished by simply determining whether or not a keyword (“pension” in the example in the figure) is included. Therefore, as described in Non-Patent Document 1 described above, a method for determining the usefulness of the online page by regarding the number of linked links as support votes from others is quite effective in Web search. However, there are reports that blog articles have fewer links than traditional Web pages, and that blog articles with linked links account for just over 1% of the total (for example, Ko Fujimura, Takefumi Inoue, Masayuki Sugisaki, “The EigenRumor Algorithm for Ranking Blogs”, In Proceedings of the WWW 2005 2nd Annual Workshop on the Weblogging Ecosystem: Aggregation, Analysis and dynamics, May 2005.). Since SNS also has the same usage pattern as blogs, it seems that the tendency for the number of linked links to be small is the same.

  In this way, because the number of linked pages is small, comments can be given to articles in blogs and SNSs, so there are many readers who replace the response to articles with the act of simple comments instead of the act of links. It seems to be the cause. Focusing on this point, a method has been proposed for evaluating an online page by using information such as the number of comments given to articles and the number of people given comments, such as Non-Patent Document 2 above. . In the method of Non-Patent Document 2, for example, an article that continues to receive comments from many users over a long period of time is determined to contain content that has material value. Alternatively, it is determined that an article in which a small number of people have given comments many times contains content that is subject to deep discussion.

  However, the technique of Non-Patent Document 2 often fails to find articles useful to third parties. For example, it is assumed that there are an article C and an article D as shown in FIG. 16, and M comments are transmitted from N users in both articles. In the method of Non-Patent Document 2, Article C and Article D are determined to be equally useful, but in fact, Article C contains information that is useful to third parties, and the comments are debated. The article D has contents that can only be understood by the acquaintance of the article creator, and the comments often involve informal communication that only the parties can enjoy. Whether these articles are useful to third parties varies greatly, but existing methods cannot distinguish this difference.

  The present invention has been made in view of the above points, and does not depend on the semantic content of the online page, but analyzes the characteristics of the intention of the online page creator and comment giver, the characteristics of the online page, etc. It is an object of the present invention to provide an on-line page analysis method and apparatus, a program, and a computer-readable recording medium that can be used.

  For details, use the number of characters and pictograms included in the articles and comments on the online page, whether the person who created the article and the person who gave the comment attaches importance to accurately communicating the information, or is rich in emotion It is an object of the present invention to provide an on-line page analysis method and apparatus, a program, and a computer-readable recording medium capable of quantifying the intention such as whether or not the communication is important.

  FIG. 1 is a diagram for explaining the principle of the present invention.

The present invention (Claim 1) is an online page analysis method in an apparatus for analyzing online pages including blogs and SNSs on the Internet,
Data input means acquires the article part and comment part of the online page, calculates the number of article characters, the number of article pictograms, the average number of comments, the average number of comment characters, and stores the number of characters in association with the page ID that uniquely identifies the online page A data input process (step 1) to be stored in the means;
The score calculation means reads the number of article characters, the number of article pictograms, the average number of comments, the average number of comments for each page ID from the number-of-characters storage means, and the value increases as the number of characters or the number of pictograms increases. A score calculation process (step 2) of calculating a pictogram number score, a comment character count score, and a comment pictogram count score and storing them in the score storage means;
The document feature quantity calculation means reads out the score from the score storage means, and uses the score to give importance to the author and the comment giver to accurately convey the information, or to convey emotionally A feature amount calculation process (step 3) for calculating a feature amount indicating whether or not importance is placed and storing the feature amount in a feature amount storage unit;
An output step (step 4) in which the data output means searches the feature quantity storage means for a page ID having the feature quantity of the page ID designated by the inputted query or the feature quantity designated by the query (step 4); ,I do.

Further, the present invention (Claim 2), in the document feature amount calculation process (Step 3),
The value obtained by dividing the article character score by the article pictograph score is defined as the article CP ratio. When the article CP ratio is large, the article creator has a feature of accurately transmitting information, and when the article CP ratio is small. The article CP ratio is stored in the feature amount storage means as having the feature that the article creator conveys information in an emotional manner.

Further, the present invention (Claim 3), in the document feature amount calculation process (Step 3),
A value obtained by dividing the comment character score by the comment pictogram score is defined as a comment CP ratio. When the comment CP ratio is large, the comment giver accurately conveys information. When the comment CP ratio is small, The comment CP ratio is stored in the feature amount storage means as having the feature that the comment creator conveys information in an emotional manner. Further, the present invention (Claim 4) is a document feature amount calculation process (Step 3). ,
The online pages are grouped according to whether the article character score and the comment character score are larger or smaller than a predetermined value, and the page ID of the online page belonging to the group is associated with each group. Store in the feature quantity storage means.

  FIG. 2 is a principle configuration diagram of the present invention.

The present invention (Claim 5) is an online page analyzer for analyzing online pages including blogs and SNSs on the Internet,
The article portion and the comment portion of the online page are acquired, the number of article characters, the number of article pictograms, the number of comment average characters, and the number of comment average pictograms are calculated and stored in the character count storage unit 24 in association with the page ID that uniquely identifies the online page. Data input means 13;
For each page ID, the number of article characters, the number of article pictograms, the number of comment average characters, the number of comment average pictograms are read out from the number-of-characters storage means 24, and the value increases as the number of characters or the number of pictograms increases. A score calculating means 25 for calculating a comment character number score and a comment pictogram number score and storing them in the score storage means 22;
Whether the score is read from the score storage means 22, and whether the author and the comment giver give importance to accurately communicating the information or whether the emphasis is to convey emotionally is used. A feature quantity calculating means 19 for calculating the feature quantity to be shown and storing it in the feature quantity storage means 23;
And a data output unit 20 that searches the feature amount storage unit 23 for a page ID having the feature amount of the page ID specified by the input query or the feature amount specified by the query, and outputs the page ID.

Further, according to the present invention (Claim 6), in the document feature amount calculation means 19,
The value obtained by dividing the article character score by the article pictograph score is defined as the article CP ratio. When the article CP ratio is large, the article creator has a feature of accurately transmitting information, and when the article CP ratio is small. The article creator includes a means for storing the article CP ratio in the feature amount storage means 23 as having the characteristic of conveying information richly in emotion.

Further, according to the present invention (Claim 7), in the document feature amount calculation means 19,
A value obtained by dividing the comment character score by the comment pictogram score is defined as a comment CP ratio. When the comment CP ratio is large, the comment giver accurately conveys information. When the comment CP ratio is small, The comment creator includes a means for storing the comment CP ratio in the feature amount storage means 23 as having the characteristic of conveying information in an emotional manner.

Further, according to the present invention (Claim 8), the document feature amount calculating means 19
The online pages are grouped according to whether the article character score and the comment character score are larger or smaller than a predetermined value, and the page ID of the online page belonging to the group is associated with each group. Means for storing in the feature quantity storage means 23;

  The present invention (Claim 9) is an online page analysis program for causing a computer to function as each means constituting the online page analysis apparatus according to any one of Claims 5 to 8.

  The present invention (Claim 10) is a computer-readable recording medium storing the online page analysis program according to Claim 9.

  As described above, according to the present invention, online pages such as blogs and SNSs can be analyzed in more detail, and the accuracy and flexibility of page search and data mining can be improved.

  Specifically, feature quantities such as article authors and commenters who place importance on accurately conveying information or empathic feelings, the types of information contained in articles, It will be possible to find online pages corresponding to queries that specify feature values such as reader responses to the information, and to perform data mining focusing on these feature values.

  In addition, since processing that depends on the semantic content of the document is not performed, it can be easily applied to pages such as blogs and the like that are difficult to analyze semantic content because there are many stylized sentences.

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

  In the present invention, blogs and SNSs are typical online pages on which comments can be added, but the created document can be posted on the network, and a user other than the document creator can view the document. And a service that allows the user to add a comment to the document. FIG. 3 shows a typical online page structure to which the present invention is directed.

  In the following description, the “pictogram” is realized by using an image such as a GIF (Graphics Interchange Format) file as shown in FIG. 4A, or as shown in FIG. 4B. And one or more characters. In the case of FIG. 5B, the constituent characters are not counted as the number of characters, but are counted as one pictogram.

  FIG. 5 shows the configuration of an online page analysis apparatus according to an embodiment of the present invention.

  The online page analysis apparatus 10 includes a data input unit 13, an article character number score calculation unit 14, an article pictogram number score calculation unit 15, a comment character number score calculation unit 17, a comment pictogram number score calculation unit 18, a document feature amount calculation unit 19, and data. The output unit 20, the score storage unit 22, and the feature amount storage unit 23 are configured.

  Outside the online page analysis device 10, there are an online page crawler 11, a page parser 12, a pictographic database 16, and a data mining device 21.

  The online page crawler 11 is a general device that can collect and store online pages.

  The page parser 12 is a general device that appropriately extracts an article part and a comment part from an HTML file or the like that is the substance of an online page stored in the online page crawler 11. Here, when there are a plurality of comments, the following description will be made assuming that each is distinguished.

  The pictographic database 16 is a general device and stores attributes (number of pictographs, etc.) related to pictographs.

  The data mining device 21 is a general device and mines acquired data. In the following, it is assumed that a query from a user is received, a query condition is passed to the data output unit 20, and a corresponding feature amount is acquired and mined.

  The data input unit 13 calculates the number of article characters, the number of article pictograms, the number of comment average characters, the number of comment average pictograms based on the article portion and the comment portion extracted by the page parser 12, and associates the number of characters with the ID of the online page. Store in the storage unit 24. For pictograms, the pictogram database 16 is referenced to extract attributes (number of pictograms, etc.) relating to pictograms included in articles and comments. A data format stored in the character number storage unit 24 is shown in FIG. Instead of the average number of comments and the average number of comment pictograms, the maximum number or median value of the number of characters and the number of pictograms of each comment may be used.

  The article character number score calculation unit 14 calculates the article character number score (hereinafter referred to as “article C-Score”) of the online page using the number of article characters stored in the character number storage unit 24 and stores it in the score storage unit 22. . This is a score that increases as the number of characters increases. For example, there is a method in which the number of characters is used as a score. Alternatively, an appropriate calculation such as weighting the number of characters or normalization may be performed. Here, normalization is performed using the number of article characters of all online pages analyzed, and the article C-Score is expressed by a numerical value between 0 and 1 (all online pages analyzed) The article C-Score of the page having the smallest number of article characters is 0, and the article C-Score of the page having the largest number of article characters is 1).

  The article pictogram score calculation unit 15 calculates the article pictograph count score (hereinafter referred to as “article P-Score”) of the online page using the number of article pictograms stored in the character count storage unit 24, and the score storage unit 22. To store. This is a score that increases as the number of pictograms increases. For example, there is a method in which the number of pictograms is used as a score. Alternatively, an appropriate calculation such as weighting the number of pictograms or normalization may be performed. Here, it is assumed that normalization similar to the above-described article character number score calculation unit 14 is performed.

  The comment character number score calculation unit 17 calculates the comment character number score (hereinafter referred to as “comment C-Score”) of the online page using the number of comment characters stored in the character number storage unit 24 and stores it in the score storage unit 22. . This is a score that increases as the number of characters increases. For example, there is a method in which the number of characters is used as a score. Alternatively, an appropriate calculation such as weighting the number of characters or normalization may be performed. Here, normalization similar to the article character score calculation unit 14 is performed.

  The comment pictogram score calculation unit 18 calculates the comment pictograph count score (hereinafter referred to as “comment P-Score”) of the online page using the comment pictogram number stored in the character count storage unit 24, and the score storage unit 22. To store. This is a score that increases as the number of pictograms increases. For example, there is a method in which the number of pictograms is used as a score. Alternatively, an appropriate calculation such as weighting the number of pictograms or normalization may be performed. Here, normalization similar to the article character score calculation unit 14 is performed.

  Each score calculated by the above article character number score calculating unit 14, article pictogram number score calculating unit 15, comment character number score calculating unit 17, and comment pictogram number score calculating unit 18 is, for example, in a data format as shown in FIG. It is held in the score storage unit 22.

  The document feature quantity calculation unit 19 reads the score shown in FIG. 7 from the score storage unit 22, calculates the feature quantity of the online page based on the score, and stores it in the feature quantity storage unit 23. Specific methods for calculating the feature amount will be described in detail in the first to third embodiments.

  The data output unit 20 outputs the feature amount of the page ID of the online page specified by the query acquired from the general data mining device 21, or the page ID group of the online page having the feature amount specified by the query Is output.

  FIG. 8 is a flowchart of an outline operation in one embodiment of the present invention.

  Step 101) The data input unit 13 acquires the article part and comment part of the online page from the page parser 12.

  Step 102) The data input unit 13 calculates the number of article characters and the number of article pictograms, and stores them in the character number storage unit 24 in the format shown in FIG. Here, for the number of article pictograms, the pictogram database 16 is accessed and the pictogram count is acquired.

  Step 103) The data input unit 13 calculates the average number of comments and the average number of comment pictograms. The average number of comments is a value obtained by dividing the number of comments by all comments on the online page. Further, the average number of comment pictograms is the number obtained by dividing the number of pictograms obtained by accessing the pictogram database 16 by the number of comments. The number of characters calculated in this way is stored in the number-of-characters storage unit 24 in the format shown in FIG.

  Step 104) The article character number score calculation unit 14 calculates the article character number score (article C-Score) from the character number storage unit 24. As described above, the calculation method may be a calculation method in which the score increases as the number of characters increases. The obtained article character count score (article C-Score) is stored in the score storage unit 22 in association with the page ID of the online page to be processed.

  Step 105) The article pictogram score calculation unit 15 calculates the score of the article pictogram number (article P-Score) from the character number storage unit 24. As described above, the calculation method may be a calculation method in which the score increases as the number of characters increases. The obtained article pictogram score (article P-Score) is stored in the score storage unit 22 in association with the page ID of the online page to be processed.

  Step 106) The comment character number score calculation unit 17 calculates the article character number score (comment C-Score) from the character number storage unit 24. As described above, the calculation method may be a calculation method in which the score increases as the number of characters increases. The obtained comment character count score (comment C-Score) is stored in the score storage unit 22 in association with the page ID of the online page to be processed.

  Step 107) The comment pictogram score calculation unit 18 calculates a comment pictogram score (comment P-Score) from the character count storage unit 24. As described above, the calculation method may be a calculation method in which the score increases as the number of characters increases. The obtained comment pictogram number score (comment P-Score) is stored in the score storage unit 22 in association with the page ID of the online page to be processed.

  Step 108) The document feature quantity calculation unit 19 reads the score for each page ID from the score storage unit 22 and obtains the feature quantity. First, as a first method, a method for obtaining a CP ratio from an article C-Score and an article P-Score, and as a second method, a method for obtaining a CP ratio from a comment C-Score and a comment P-Score, three As an eye method, there is a method of grouping based on the magnitude relationship between the scores of the article C-Score and the comment C-Score, and acquiring the page ID of the group and the online page. Each of these methods will be described in the examples. Each obtained feature amount is stored in the feature amount storage unit 23 for each page ID.

  Step 109) The data output unit 20 acquires a query input from the data mining device 21 by the user.

  Step 110) The data output unit 20 searches the feature amount storage unit 23 based on the query, and outputs information (page ID or feature amount) corresponding to the query.

  Hereinafter, the present invention will be described in detail using specific examples.

[First embodiment]
In the present embodiment, as an example of processing performed by the document feature amount calculation unit 19, feature amount calculation using a value obtained by dividing the article C-Score by the article P-Score (hereinafter referred to as “article CP ratio”). State. Note that the processing of the other components of the online page analysis apparatus 10 is the same as that of the above-described embodiment.

  The document feature amount calculation unit 19 is a means that emphasizes that characters convey information accurately, and is an analysis that utilizes the emphasis that pictograms convey emotions richly.

  For example, an article having a large number of characters and few pictograms has a large article CP ratio. In the case of such an article, it can be said that the creator of the article uses a lot of means of characters that can express information more accurately and emphasizes accurately conveying the contents of the document to the reader. Articles that deeply examine current affairs and articles that introduce programming techniques often have a large article CP ratio.

  Conversely, an article with a small number of characters and a large number of pictograms has a small article CP ratio. In the case of such articles, article creators often try to convey emotionally rather than accurately convey information. Also, there are many cases where pictograms are frequently used to make articles appearing as if they contain everyday contents that seem to be enjoyable.

The document feature amount calculation unit 19 reads the article C-Score and the article P-Score from the score storage unit 22 and calculates the article CP ratio for each page ID of the online page.
Article CP ratio = Article C-Score / Article P-Score
And stored in the feature quantity storage unit 23 in the format shown in FIG.

  By calculating the article CP ratio of each online page as described above, the apparatus 10 places importance on the article creator's importance of accurately communicating information or on feeling-rich communication. This makes it possible to respond to queries that specify conditions. For this reason, for example, when this apparatus 10 is introduced into an online page search service, pages can be classified and searched in more detail than before, so that the user's search needs can be flexibly met.

[Second Embodiment]
In the present embodiment, when a comment is given to an article in the online page to be analyzed, the comment C-Score is divided by the comment P-Score as an example of processing performed by the document feature amount calculation unit 19. The feature amount calculation using the value (hereinafter referred to as “comment CP ratio”) will be described. Note that the processing of the other components of the online page analysis apparatus 10 is the same as that of the above-described embodiment.

  As in the first embodiment, this is a means for emphasizing that characters convey information accurately, and is an analysis that uses emphasis on emoticons for conveying emotions richly. . It also takes into account that pictograms are rarely included when comments indicate hostility to the article creator.

In this embodiment, the document feature amount calculation unit 19 reads the comment C-Score and the comment P-Score from the score storage unit 22, and calculates the comment CP ratio.
Comment CP ratio = Comment C-Score / Comment P-Score
And stored in the feature amount storage unit 23 in the format shown in FIG.

  For example, a comment with a large number of characters and a small number of pictograms has a large comment CP ratio. In the case of such a comment, it can be said that the comment giver attaches great importance to accurately using the means of characters that can express information more accurately and conveying the text content to the article creator or reader accurately. Seriously, comments that include discussion and comments that show hostility often have a high comment CP ratio.

  Conversely, a comment with a small number of characters and a large number of pictographs has a small comment CP ratio. In the case of such a comment, the comment giver often tries to convey the information more emotionally than to convey the information accurately. In many cases, the comment giver is often used as a pictograph when he is in close contact with the article creator.

  As described above, by calculating the comment CP ratio of each online page, the device 10 emphasizes that the comment giver gives importance to accurately communicating information or conveys emotionally. It is possible to respond to queries that specify the condition. For this reason, for example, when this apparatus 10 is introduced into an online page search service, pages can be classified and searched in more detail than before, so that the user's search needs can be flexibly met.

[Third embodiment]
In the present embodiment, when a comment is given to an article in an analysis-target online page, as an example of processing performed by the document feature amount calculation unit 19, the C-Score of the article and the comment is used. A case will be described in which the result of determining to which of the groups shown in FIG. 11 an online page belongs is calculated as a feature value, recorded and used in association with each page.

  In the following, regarding the size of the article C-Score and the comment C-Score, for example, the range of 0 to 0.5 is “small”, and the range of 0.51 to 1 is “large”. The range is determined, and the score is determined. In addition to the above magnitude determination, for example, “large” is 0.5 or more in advance and “small” is less than 0.5, or “high” is the top 50% of data scores and “less than 50% is low” A method such as “small” is conceivable.

  The document feature amount calculation unit 19 reads the article C-Score and the comment C-Score for each page ID of the online page from the score storage unit 22, performs grouping according to a predetermined range, and performs grouping as shown in FIG. Then, it is stored in a memory (not shown) inside the document feature value calculation unit 19.

  In FIG. 11, “Group 1” is a group of an article C-Score [Large] and a comment C-Score [Large]. On the page belonging to “Group 1”, a lot of information is written in the article, and the reader transmits a comment including a lot of information. This phenomenon is likely to occur when discussions are happening with readers agreeing and disagreeing, etc., using comments on articles that contain detailed information and opinions on current affairs.

  “Group 2” is a group of an article C-Score [small] and a comment C-Score [large]. For the pages belonging to “Group 2”, although the article does not contain much information, the reader transmits a comment containing much information. This phenomenon is likely to occur when content that has a very high impact is written in the article, and the reader uses the comments to give their opinions and impressions.

  “Group 3” is a group of an article C-Score [small] and a comment C-Score [small]. The pages belonging to “Group 3” do not contain much information in the article, and the reader also sends comments that do not contain much information.

  This phenomenon is likely to occur when an article contains a trivial event of an individual's daily life, and an acquainted reader comments and plays informal communication.

  “Group 4” is a group of an article C-Score [large] and a comment C-Score [small]. On the page belonging to “Group 4”, a lot of information is written in the article, and the reader transmits a comment that does not contain much information. This phenomenon is likely to occur when an article is intended for information transmission with detailed information, and the reader is thanking the reader for the information.

  By determining which group each online page belongs to as described above, the device 10 can respond to a query that specifies the type of information contained in the article and the reader's reaction to the information. become. Here, the data mining apparatus 21 has an interface as shown in FIG. 13 and inputs the group information checked in the figure to the data output unit 20.

  For example, when a query “data to be viewed on discussion topic or discussion process itself” (group 1) is input from the data mining device 21, the data output unit 20 displays, for example, the feature amount shown in FIG. Pages belonging to “group 1” are preferentially presented from the table in the storage unit 23. Further, for a query “I want to browse information with strong impact” (group 2), pages belonging to “group 2” are presented preferentially. For a query “I want to view information about an individual's daily events and informal communication that occurs here” (group 3), a page belonging to “group 3” is presented. In response to the query “want to browse a page for information transmission” (group 4), pages belonging to group 4 are presented preferentially.

  These queries and other queries (such as “I want to view a page with a high TF / IDF (Term Frequency / Inverse Term Frequency) value for a certain keyword”) may specify multiple conditions at the same time. Appropriate weights may be applied to. For example, a weight of 10 is applied to query 1 "I want to view a page containing a topic that is in discussion", and a weight of 20 is applied to query 2 "I want to view information with strong impact". When query 3 “I want to view a page with a high keyword TF / IDF value” is multiplied by 30, the score of the page belonging to “Group 1” is multiplied by 10 and the page belonging to “Group 2” The score of 20 is added by 20 times, and the score of the page including “Pension” is added by multiplying the TF / IDF value of “Pension” by 30 and presented to the user in order from the page with the highest total score. It works like this.

  Here, the case where each page is determined to belong to only one group has been described as an example. However, for example, the center point of each group is defined, and the center of each page and each group is defined. "Nearness (reciprocal of distance, etc.)" may be recorded as an attribute of the page. In this case, for example, page A has attributes such as “1” close to group 1 and “2” close to group 2, and “I want to view a page including a topic that is calling for discussion”. If the query 1 is multiplied by a weight of 10 and the query 2 “I want to view information with strong impact” is multiplied by a weight of 20, the score will be 10 × 1 + 20 × 2 = 50.

  Further, the first embodiment and the second embodiment may be incorporated in the third embodiment. In this case, in addition to the above, the online page is more detailed and flexible in consideration of whether the author or commentator places importance on accurately communicating information or on empathic communication. Can be analyzed.

  It should be noted that the function of each component of the online page analysis apparatus 10 shown in FIG. 3 is constructed as a program and installed and executed on a computer used as the online page analysis apparatus 10, or distributed via a network. It is possible.

  Further, the constructed program can be stored in a portable storage medium such as a hard disk, a flexible disk, or a CD-ROM, and can be installed or distributed in a computer.

  The present invention is not limited to the above-described embodiments and examples, and various modifications and applications can be made within the scope of the claims.

  In the present invention, online pages such as blogs, SNSs, and documents created by document creators can be posted on the network, and users other than the document creator can view the documents. It is applicable to services that can be granted.

It is a figure for demonstrating the principle of this invention. It is a principle block diagram of this invention. It is an example of the online page made into the object of this invention. It is an example of the pictogram used on the online page of this invention. It is a block diagram of the online page analysis apparatus in one embodiment of this invention. It is an example of data of the character number memory | storage part in one embodiment of this invention. It is an example of the calculated score in one embodiment of the present invention. It is a flowchart of the outline | summary operation | movement in one embodiment of this invention. It is an example of the feature-value calculation result in 1st Example of this invention. It is an example of the feature-value calculation result in 2nd Example of this invention. It is a figure for demonstrating the group determination in the document feature-value calculation part of 3rd Example of this invention. It is an example of grouping in the 3rd example of the present invention. It is an example of the group information input from the data mining apparatus in 3rd Example of this invention. It is an example of the table of the feature-value storage part in 3rd Example of this invention. It is an example of the article text of an online page. This is an example of the text of an article and comments.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Online page analysis apparatus 11 Online page crawler 12 Page parser 13 Data input means, data input part 14 Article character number score calculation part 15 Article pictogram number score calculation part 16 Pictogram database 17 Comment character number score calculation part 18 Comment pictogram number score calculation part 19 Feature quantity calculation means, document feature quantity calculation section 20 Data output means, data output section 21 Data mining device 22 Score storage means, score storage section 23 Feature quantity storage means, feature quantity storage section 24 Character number storage means, Character number storage section 25 Score Calculation means

Claims (10)

An online page analysis method in an apparatus for analyzing online pages including blogs and SNS (Social Networking Service) on the Internet,
The data input means acquires the article portion and comment portion of the online page, calculates the number of article characters, the number of article pictograms, the average number of comments, the average number of comment characters, and associates the online page with a page ID that uniquely identifies the number of characters. A data input process to be stored in the storage means;
The score calculation means reads the number of article characters, the number of article pictograms, the number of comment average characters, the number of comment average pictograms for each page ID from the character count storage means, and the value increases as the number of characters or the number of pictograms increases. An article character count score, an article pictogram score, a comment character count score, a comment pictogram score, and a score calculation process for storing the score in a score storage means;
The document feature amount calculation means reads out the score from the score storage means, and uses the score to give importance to the author and the comment giver to accurately convey the information or convey it in an emotional manner. A feature amount calculation process for calculating a feature amount indicating whether importance is attached to the feature amount and storing the feature amount in a feature amount storage unit;
An output step in which the data output means retrieves and outputs the feature amount of the page ID specified by the input query or the page ID having the feature amount specified by the query from the feature amount storage means;
An online page analysis method characterized by: In the document feature amount calculation process,
A value obtained by dividing the article character count score by the article pictogram score is defined as an article CP ratio. When the article CP ratio is large, the article creator has a feature of accurately transmitting information, and the article CP ratio is small. The online page analysis method according to claim 1, wherein the article CP ratio is stored in the feature amount storage means, assuming that the article creator has a characteristic of conveying information in an emotional manner. In the document feature amount calculation process,
A value obtained by dividing the comment character score by the comment pictogram score is defined as a comment CP ratio. When the comment CP ratio is large, the comment giver accurately conveys information, and the comment CP ratio is small. The online page analysis method according to claim 1, wherein the comment CP ratio is stored in the feature amount storage means, assuming that the comment creator has a feature of conveying information in an emotional manner. In the document feature amount calculation process,
Online pages are grouped according to whether the article character number score and the comment character number score are larger or smaller than a predetermined value, and the page ID and group of the online page belonging to the group are grouped for each group. The on-line page analysis method according to claim 1, wherein the information is stored in the feature quantity storage unit in association with each other. An online page analysis device for analyzing online pages including blogs and SNS (Social Networking Service) on the Internet,
The article portion and comment portion of the online page are acquired, the number of article characters, the number of article pictograms, the average number of comment characters, and the average number of comment pictograms are calculated and stored in the character count storage means in association with the page ID that uniquely identifies the online page. Data input means;
For each page ID, the number of article characters, the number of article pictograms, the number of comment average characters, and the number of comment average pictograms are read from the character number storage means for each page ID, and the value increases as the number of characters or the number of pictograms increases. Calculating a pictogram number score, a comment character count score, a comment pictogram count score, and storing the score in the score storage means;
Whether the score is read from the score storage means, and using the score, whether the author and comment give importance to accurately communicating information or whether to emphasize emotionally A feature quantity calculating means for calculating the feature quantity to be shown and storing it in the feature quantity storage means;
A data output means for searching and outputting the feature amount of the page ID specified by the input query or the page ID having the feature amount specified by the query from the feature amount storage means;
An on-line page analysis apparatus characterized by comprising: The document feature amount calculating means includes:
A value obtained by dividing the article character count score by the article pictogram score is defined as an article CP ratio. When the article CP ratio is large, the article creator has a feature of accurately transmitting information, and the article CP ratio is small. 6. The online page analyzing apparatus according to claim 5, further comprising means for storing the article CP ratio in the feature amount storage means, assuming that the article creator has a characteristic of conveying information in an emotional manner. The document feature amount calculating means includes:
A value obtained by dividing the comment character score by the comment pictogram score is defined as a comment CP ratio. When the comment CP ratio is large, the comment giver accurately conveys information, and the comment CP ratio is small. 6. The on-line page analysis apparatus according to claim 5, further comprising means for storing the comment CP ratio in the feature amount storage means as having a feature that the comment creator conveys information in an emotional manner. The document feature amount calculating means includes:
Online pages are grouped according to whether the article character number score and the comment character number score are larger or smaller than a predetermined value, and the page ID and group of the online page belonging to the group are grouped for each group. 6. The online page analyzing apparatus according to claim 5, further comprising means for associating and storing in the feature quantity storage means.   An online page analysis program for causing a computer to function as each means constituting the online page analysis device according to claim 5.   A computer-readable recording medium storing the online page analysis program according to claim 9.

Images