JP5610634B2 - Related information distribution program, server and method for distributing related information based on keywords of public contents - Google Patents

Description

  The present invention relates to a technique for controlling to distribute related information based on a keyword of public content.

  In recent years, various contents have been released to an unspecified number of third parties via the Internet. The public contents are various media contents such as TV broadcast contents, Web pages, music contents, and electronic books, and are contents that can be accessed by an unspecified number of third parties.

  For example, in the case of television broadcast content, not only video / audio information as a broadcast program but also various information for viewers are included. In particular, subtitle character information displayed in synchronization with the video / audio information is usually a short sentence summarizing the contents of the broadcast program, and a keyword included in the sentence appropriately represents the video / audio information. .

  On the other hand, comment texts from an unspecified number of third parties are sent to sites such as blogs (Web log) and mini blogs (for example, twitter (registered trademark)) via the Internet. It is actively transmitted. Such comment text is characterized by being discussed on one common topic. As such a common topic, there is public content distributed to an unspecified number of third parties as described above.

  For example, the user can discuss the broadcast content among a large number of unspecified users via a miniblog or the like while viewing the public content. In recent years, viewing styles such as watching TV broadcast contents (for example, watching a drama) and sending comments (for example, various messages not limited to a drama story) via a miniblog are becoming widespread. is there. This provides a feeling that one public content is shared with many users for viewing.

  As a conventional technique related to such a service, a characteristic keyword is automatically extracted from subtitle character information in a television broadcast content, a page in which a sentence including the keyword is described is searched from a website, and the page is searched. There is a technique of presenting to a user (see, for example, Patent Document 1). For example, the Web site may be an online shopping site that sells products related to the keyword. The user can recognize the Web page as advertisement information.

  It can also be assumed that a large number of viewers for the same TV broadcast content extract a keyword of interest from comment text written on a blog site and distribute advertisement information related to the keyword. Thereby, it is possible to effectively distribute the advertising information based on the keyword that the viewer is paying attention to.

JP 2010-087658 A JP 2008-048297 A

Reputation analysis software “Uwasa”, Institute of Language Engineering, [online], [searched on November 4, 2010], Internet <URL: http://www.gengokk.co.jp/koubunnews.htm>

  However, in many cases, the keyword of the comment text written by the viewer includes a keyword representing the contents of the public content (for example, television broadcast content). From an advertiser's perspective, even advertisement information searched based on keywords in a user's comment text is often only advertisement information that can be assumed in advance from the contents of the television broadcast content.

  On the other hand, there are cases where topics that cannot be assumed at all from the contents of the public content are raised and discussed on the blog site. For example, for TV broadcast content of a drama, there are cases where the main character has an interest in a bag that has nothing to do with the story. On the blog site, for example, "I want the bag, but where do you sell it?" In this case, the advertiser wants to distribute advertisement information about the bag to the viewer.

  Therefore, an object of the present invention is to provide a related information distribution program, a server, and a method capable of searching and distributing related information (advertisement information) based on a keyword that cannot be assumed from the contents of public content.

According to the present invention, for a public content for an unspecified number of third parties, a related information distribution program for causing a computer to search related information for the public content and to distribute the related information,
Content keyword extraction means for extracting a plurality of content keywords included in the published content;
Comment collecting means for collecting comment sentences from a large number of viewers who have viewed public content from a comment server;
Comment keyword extraction means for extracting a plurality of comment keywords included in a comment sentence;
Effective keyword extraction means for extracting keywords other than content keywords as effective keywords from comment keywords,
Related information search means for searching related information related to effective keywords from a database;
The computer is caused to function as distribution control means for distributing related information to a third party together with the public content.

According to another embodiment of the related information distribution program of the present invention,
The content keyword extracting means extracts a keyword from a sentence included in the public content by morphological analysis, and at the same time, extracts a characteristic word by TF-IDF (Term Frequency-Inverse Document Frequency). Extracted as keywords,
And / or
The comment keyword extracting means preferably extracts a keyword from a large number of comment sentences by morphological analysis, and also causes the computer to function so as to extract a characteristic word as a comment keyword by TF-IDF.

According to another embodiment of the related information distribution program of the present invention,
The public content is a television broadcast content, a web page, a music content, or an electronic book, and the content keyword is all keywords based on meta information and text included in the content,
The comment text is posted on a mini web log server, and it is also preferable to make the computer function so that the comment keywords are all keywords based on the text in a large number of comment texts.

According to another embodiment of the related information distribution program of the present invention,
The database is a collection of advertising information as related information linked to keywords.
The related information search means preferably uses a database to cause the computer to search for advertisement information as related information.

  According to another embodiment of the related information distribution program of the present invention, the related information distribution control means can acquire the related information together with the public content from the terminal operated by the user to the server that publishes the public content. It is also preferable to control the distribution.

According to another embodiment of the related information distribution program of the present invention,
It is also preferable to further cause the computer to function as reputation analysis means for extracting only effective keywords based on positive reputation based on the dependency expressions before and after the effective keyword output from the effective keyword extraction means.

According to the present invention, a content server that publishes public content to an unspecified number of third parties, a comment server that publishes comment texts from a large number of viewers who have viewed the public content, and a user operated. A related information distribution server that is communicable with a terminal, searches related information for the public content, and distributes the related information to the terminal,
Content keyword extraction means for extracting a plurality of content keywords included in the published content received from the content server;
Comment collecting means for collecting comment sentences from a large number of viewers who have viewed public content from a comment server;
Comment keyword extraction means for extracting a plurality of comment keywords included in a comment sentence;
Effective keyword extraction means for extracting keywords other than content keywords as effective keywords from comment keywords,
Related information search means for searching related information related to effective keywords from a database;
And distribution control means for distributing the related information to a third party together with the public content.

According to the present invention, a content server that publishes public content to an unspecified number of third parties, a comment server that publishes comment texts from a large number of viewers who have viewed the public content, and a user operated. A related information distribution method in a system having a terminal and a related information distribution server that searches related information for the published content and distributes the related information to the terminal,
A first step of extracting a plurality of content keywords included in the published content received from the content server;
A second step of collecting comment text from a large number of viewers who have viewed public content from a comment server;
A third step of extracting a plurality of comment keywords included in the comment text;
A fourth step of extracting a keyword other than the content keyword as an effective keyword from the comment keywords;
A fifth step of searching the database for related information related to the effective keyword;
And a sixth step of distributing the related information to a third party together with the public content.

  According to the related information distribution program, server, and method of the present invention, it is possible to search and distribute related information (advertisement information) based on keywords that cannot be assumed from the contents of public content.

It is a system configuration diagram in the present invention. It is a sequence diagram in the present invention. It is a function block diagram of the related information delivery server in this invention. It is explanatory drawing showing the flow of the text information based on FIG.

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

  FIG. 1 is a system configuration diagram according to the present invention.

  According to FIG. 1, the content publishing server 2 is connected to the Internet. The content publishing server 2 distributes the public content to an unspecified number of third parties. The public content is various media content, and may be, for example, a television broadcast content, a web page, music content, or an electronic book. In the case of television broadcast content broadcast via the air, it is distributed by terrestrial digital broadcast, one-segment broadcast, BS (Broadcast Satellite) broadcast, or the like. These broadcast contents basically include subtitle character information as well as video / audio information. In the case of a web page, music content, or electronic book, it is distributed via the Internet.

  The router 6 installed in the house may be connected to the Internet via an access network and may have a set-top box function for receiving broadcast waves.

  The terminal 4 is a device capable of reproducing public content, and may be a television, a personal computer, a mobile terminal, a smartphone, or the like. The terminal 4 receives and reproduces the public content distributed from the content public server 2 via the router 6. The user can view the public content reproduced on the terminal 4.

  According to FIG. 1, a miniblog server 3 is further connected to the Internet. The miniblog server 3 may be a twitter (registered trademark) site, for example. An unspecified number of third-party comment contributors can use the general-purpose terminal 5 to post comment text to the miniblog server 3 and view other person's comment text while viewing the public content.

  According to the present invention, the related information distribution server 1 is further connected to the Internet. The related information distribution server 1 searches the related information for the public content with respect to the public content for an unspecified number of third parties, and controls to distribute the related information.

  The related information distribution server 1 receives public content from the content public server 2 and collects comment text related to the public content from the miniblog server 3. Then, the related information distribution server 1 searches for related information (advertisement information) based on the comment text, and transmits the advertisement information to the content publishing server 2. The content publishing server 2 distributes the public content to a third party and also distributes the advertisement information. Thereby, the advertisement information can also be visually recognized by the user who is viewing the public content using the terminal 4.

  FIG. 2 is a sequence diagram in the present invention.

(S100) The content publishing server 2 distributes the public content to an unspecified number of third parties. The router 6 transfers the public content received via the air or the network to the terminal 4 (for example, personal computer) in the home network. The terminal 4 reproduces and displays the received public content, and the user can view the public content.

(S101) The related information distribution server 1 separates only text (character information) from the public content. In particular, the sentence to be separated varies depending on the type of public content.
[Public contents] [Text information]
TV broadcast content Program guide, subtitle characters Web page Meta information, text Music content Meta information (artist, song title), lyrics Electronic book Meta information (author, title), text

  When the public content is broadcast content, the text information may be, for example, iEPG (internet Electronic Program Guide) text. Further, the program stream may be a text included in an event information table of PSI / SI (Program Specific Information / Service Information) information.

(S102) Next, the related information distribution server 1 extracts a plurality of content keywords from the text information included in the public content by morphological analysis. Here, “morphological analysis” refers to a technique for dividing a sentence into meaningful words and discriminating parts of speech and contents using a dictionary. The “morpheme” means the smallest unit having meaning among the elements of the sentence.

  Then, the related information distribution server 1 extracts a characteristic word as a content keyword by TF-IDF (Term Frequency-Inverse Document Frequency: word appearance frequency-reverse appearance frequency) (see, for example, Patent Document 2). TF-IDF is a technique that weights each word, expresses a document from a query in a vector space, and ranks the documents based on the similarity between the query and the query. The higher the ranked value, the more important keywords are recognized. That is, the content of the public content is appropriately represented by the appearance frequency of the keyword appearing in the text information.

(S103) Thereafter, another user who is viewing the same public content has posted a comment sentence related to the public content to the miniblog server 3 using the general-purpose terminal 5. The miniblog server 3 accumulates the received comment text and makes it available to an unspecified number of third parties.

(S104) The related information distribution server 1 collects comment text related to the public content from the miniblog server 3. In order to collect comment texts related to public content, in the case of twitter, only comment texts with hash tags (# and English character strings) related to the public content are collected. Comment sentences to which the same hash tag is added can be grouped.

(S105) Next, the related information distribution server 1 extracts a plurality of comment keywords included in the comment text. A keyword is extracted from a large number of comment sentences by morphological analysis, and a characteristic word is extracted as a comment keyword by TF-IDF.

(S106) The effective keyword extraction unit 106 extracts keywords other than the content keyword from the comment keywords as effective keywords. This effective keyword is a keyword that cannot be assumed from the contents of the public content, and is a keyword that is interested by the comment poster.

(S107) Optionally, it is also preferable to extract only effective keywords based on a positive reputation based on dependency expressions before and after the effective keyword. This is because even if it is an effective keyword, it is not preferable from the standpoint of an advertiser to distribute advertising information using a keyword based on a negative reputation.

Conventionally, there is a technique of reputation analysis software (for example, see Non-Patent Document 1). According to this technology, a set of "dependency relationship" (combination of keywords and their directly related words) and "good and intent" from freely written sentences such as blogs, word-of-mouth sites, and questionnaires. Reputation information can be retrieved. According to this technology, sentence information (text) to be analyzed is analyzed with high accuracy by a combination of syntax analysis and a thesaurus (synonym dictionary), and “dependency relations” and “good and intention” are extracted. For example, there are the following examples.
“Jackets are ugly, but bags and shoes are not good.”
[Independent words] [independent words] [independent words] [independent words] [inside words] [intentions]
Jacket, is a daddy, but a bad bag, is a good, hey, good shoes, is a good, hey, good

(S108) Next, related information (advertisement information) related to the effective keyword is searched from the advertisement information database. The advertisement information database stores advertisement information as related information associated with keywords. The advertisement information is associated with a keyword as follows, for example.
[Keyword] [Advertising Information URL]
Jacket http://www.kddi.com/jackets.html
Bag http://www.kddi.com/bag.html
Shoes http://www.kddi.com/shoes.html
From the flow of S107, advertisement information URLs for bags and shoes are searched.

(S109) Then, the related information distribution server 1 transmits the advertisement information (related information) to the content publishing server 2.

(S110) The content publishing server 2 distributes the public content to the terminal 4 being viewed by the user and also distributes the advertisement information thereof. The user can view the public content and browse advertisement information that cannot be assumed from the content of the public content.

FIG. 3 is a functional configuration diagram of the related information distribution server in the present invention.
FIG. 4 is an explanatory diagram showing the flow of text information based on FIG.

  According to FIG. 3, the related information distribution server 1 includes a communication interface unit 10 connected to the Internet, an advertisement information database 11, a text information separation unit 101, a content keyword extraction unit 102, a comment collection unit 104, A comment keyword extraction unit 105, an effective keyword extraction unit 106, a reputation analysis unit 107, a related information search unit 108, and a distribution control unit 109 are included. These functional components are realized by executing a program that causes a computer mounted on the server to function.

  The sentence information separation unit 101 separates only sentences from the public content (see S101 described above). According to FIG. 4, “Today I will introduce…” is separated as text information. The separated text information is output to the content keyword extraction unit 102.

  The content keyword extraction unit 102 extracts a plurality of content keywords included in the published content (see S102 described above). The content keyword extraction unit 102 extracts keywords from the text included in the public content by morphological analysis. And the content keyword extraction part 102 extracts a characteristic word as a content keyword by TF-IDF. According to FIG. 4, “mobile phone”, “au (registered trademark)”, “new brand”, and “ida (registered trademark)” are extracted as content keywords. The extracted content keyword is output to the effective keyword extraction unit 106.

  The comment collection unit 104 collects comment text posted from a large number of viewers who have viewed the public content from the miniblog server 3 (see S104 described above). According to Fig. 4, as comment sentences, "Iida's mobile phone ... I like it." "I use ..." "Is this actress's skin ... used?" "Guest guests ... where brands?" The collected comment text is output to the comment keyword extraction unit 105.

  The comment keyword extraction unit 105 extracts a plurality of comment keywords included in the comment text (see S105 described above). The comment keyword extraction unit 105 extracts all keywords from a large number of comment sentences by morphological analysis, and extracts characteristic words as comment keywords by TF-IDF. According to FIG. 4, “iida”, “au”, “mobile phone”, “cosmetics”, and “jacket” are extracted as comment keywords. The extracted comment keyword is output to the effective keyword extraction unit 106.

  The effective keyword extraction unit 106 extracts keywords other than the content keyword from the comment keywords as effective keywords (see S106 described above). According to FIG. 4, “cosmetics” and “jacket” are extracted as effective keywords. That is, “mobile phone”, “au”, “new brand”, and “ida” are excluded because they are also included in the content keyword. The extracted effective keyword is output to the reputation analysis unit 107.

  For example, even if a product name such as “au” or “ida” can be extracted as a keyword in the comment text of the viewer for the public content of a special program related to a mobile phone, those keywords are already mentioned in the program. It is what. Advertisement information based on such keywords cannot be said to have a high advertising effect on viewers.

  On the other hand, keywords such as “cosmetics” and “jacket” are keywords related to cosmetics used by program performers and costume brands, and are not directly mentioned in the program. Advertising information based on such keywords is related to products that are of interest to viewers and can be said to have a high advertising effect.

  The reputation analysis unit 107 extracts only effective keywords based on a positive reputation based on the dependency expressions before and after the effective keyword output from the effective keyword extraction unit (see S107 described above). Here, a negative dependency expression such as “Dasai” is used for “jacket”. Therefore, “jacket” is excluded from the effective keywords. Then, only “cosmetics” having a positive reputation is output to the related information search unit 108 as an effective keyword.

  The related information search unit 108 searches related information (advertisement information) related to the effective keyword “cosmetics” from the advertisement information database 11 (see S108 described above). The advertisement information related to the searched “cosmetics” is output to the distribution control unit 109.

  The distribution control unit 109 publishes the related information together with the public content to a third party (see S109 described above). Here, the distribution control unit 109 controls the distribution so that the related information can be acquired from the terminal operated by the user, together with the public content, with respect to the content public server.

  As described above in detail, according to the related information distribution program, server, and method of the present invention, it is possible to search and distribute related information (advertisement information) based on keywords that cannot be assumed from the contents of public content. . Specifically, keywords that have not appeared in the caption text of the TV broadcast (keywords that are not directly mentioned) are extracted from comment keywords posted on blog sites on the Internet (for example, twitter) Control is performed so that advertisement information related to the keyword is presented to the user.

  Various changes, modifications, and omissions of the above-described various embodiments of the present invention can be easily made by those skilled in the art. The above description is merely an example, and is not intended to be restrictive. The invention is limited only as defined in the following claims and the equivalents thereto.

DESCRIPTION OF SYMBOLS 1 Related information delivery server 10 Communication interface part 101 Text information separation part 102 Content keyword extraction part 104 Comment collection part 105 Comment keyword extraction part 106 Effective keyword extraction part 107 Reputation analysis part 108 Related information search part 109 Distribution control part 11 Advertisement information database 2 Content publishing server 3 Miniblog server 4 Terminal 5 General-purpose terminal for commenter 6 Router

Claims (8)

A related information distribution program for causing a computer to function to search related information for the public content and distribute the related information for the public content for an unspecified number of third parties,
Content keyword extraction means for extracting a plurality of content keywords included in the published content;
Comment collection means for collecting comment sentences from a large number of viewers who have viewed the public content from a comment server;
Comment keyword extraction means for extracting a plurality of comment keywords included in the comment sentence;
Effective keyword extraction means for extracting a keyword other than the content keyword as an effective keyword from the comment keywords;
Related information search means for searching related information related to the effective keyword from a database;
A related information distribution program for causing a computer to function as distribution control means for distributing the related information to a third party together with the published content. The content keyword extraction means extracts a keyword from a sentence included in the published content by morphological analysis, and extracts a characteristic word by TF-IDF (Term Frequency-Inverse Document Frequency). , Extracted as the content keyword,
And / or
The comment keyword extraction means causes a computer to function so as to extract a keyword from a large number of the comment sentences by morphological analysis and to extract a characteristic word as the comment keyword by TF-IDF. Item 2. A related information distribution program according to Item 1. The public content is a television broadcast content, a web page, a music content, or an electronic book, and the content keyword is all keywords based on meta information and text included in the content,
The comment sentence is posted on a mini Web log server, and the comment keyword is a function of the computer to be all keywords based on text in a large number of comment sentences. The related information distribution program according to claim 1 or 2. The database stores advertisement information as related information associated with keywords,
The related information distribution program according to any one of claims 1 to 3, wherein the related information search means causes a computer to search for advertisement information as the related information using the database. .   The said related information delivery control means controls delivery so that the said related information can be acquired from the terminal operated by the user with respect to the said public content with respect to the server which publishes the said public content. 5. The related information distribution program according to any one of 1 to 4.   The computer is further caused to function as reputation analysis means for extracting only effective keywords based on a positive reputation based on dependency expressions before and after the effective keyword output from the effective keyword extraction means. The related information distribution program according to any one of 1 to 5. A content server that publishes public content to an unspecified number of third parties, a comment server that publishes comment texts from a large number of viewers who viewed the public content, and a terminal operated by the user. A related information distribution server that searches related information for the public content and distributes the related information to the terminal,
Content keyword extraction means for extracting a plurality of content keywords included in the published content received from the content server;
Comment collecting means for collecting comment texts from a large number of viewers who viewed the public content from the comment server;
Comment keyword extraction means for extracting a plurality of comment keywords included in the comment sentence;
Effective keyword extraction means for extracting a keyword other than the content keyword as an effective keyword from the comment keywords;
Related information search means for searching related information related to the effective keyword from a database;
A related information distribution server comprising: a distribution control unit that distributes the related information to a third party together with the published content. A content server that publishes public content to an unspecified number of third parties, a comment server that publishes comment texts from a large number of viewers who viewed the public content, a terminal operated by the user, and the public A related information distribution method in a system including a related information distribution server that searches related information for content and distributes the related information to the terminal,
A first step of extracting a plurality of content keywords included in the published content received from the content server;
A second step of collecting comment text from a large number of viewers who have viewed the public content from the comment server;
A third step of extracting a plurality of comment keywords included in the comment sentence;
A fourth step of extracting a keyword other than the content keyword as an effective keyword from the comment keywords;
A fifth step of searching related information related to the effective keyword from a database;
And a sixth step of distributing the related information to a third party to be disclosed together with the published content.

Images