JP2008276640A - Keyword prediction system and keyword prediction method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To predict a keyword related to an event that will become popular in the future, and effectively use it to marketing for accurately catching customer needs, listing advertisement for displaying a text advertisement on a retrieval result of a search engine for a fee, or the like. <P>SOLUTION: The method comprises acquiring television program information; extracting common nouns and proper nouns as keywords from the acquired television program information; inputting the extracted keywords to a predetermined keyword advice tool in order to extract a retrieval frequency of each of the input keywords; extracting a program name from the television program information to form a program name list; extracting an audience rate of the past time zone that becomes a weight of the television program from the television program information; dividing the retrieval frequency of each extracted keyword by the weight of the television program to acquire the rare degree of each keyword; and rearranging the keywords in the ascending order of the rare degree determined for each keyword. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a keyword prediction system and a keyword prediction method for predicting a keyword related to an event that will be popular in the future.

  Keywords are important clues for elucidating problems and content, and are used as clues for information retrieval. For example, in information retrieval, desired information can be browsed by inputting a keyword into an input field of a search engine (Web site) on the Internet.

  By the way, for example, a keyword input by a search engine is an index for knowing what interest a user visiting from the search engine has. If such an index is known, it is considered that it can be used effectively for things such as marketing that accurately grasps customer needs and listing advertisements that display paid text ads in search engine search results. .

As an application of such a keyword input to a search engine, in Patent Document 1, an advertiser keyword that can have an advertiser listing associated with an advertisement content page and a keyword with a bid face value set. The database includes a query processing subsection that is operatively coupled to the advertiser keyword database, submits search queries to the advertiser keyword database, and is associated with a partition in the advertising keyword database that processes the query. It proposes a computer system that is programmed to determine a set of listings and return search terms to a query client.
Special table 2006-516767

  In the computer system disclosed in Patent Document 1 described above, a plurality of sets of advertisement listings are adopted, and each set has a bidding mechanism, and therefore, for pay-for-performance advertisements associated with markets and sales channels. System can be realized.

  However, such systems submit search queries to the advertiser keyword database to determine the set of advertiser listings associated with the partition in the ad keyword database that will process the query. Predicting the keywords related to the event that will be performed, for example, marketing that can accurately grasp customer needs, and effective use for such things as listing ads that display paid text ads in search engine search results There was a problem that could not be planned.

  This invention is made | formed in view of such a condition, and it aims at providing the keyword prediction system and keyword prediction method which can solve the said problem.

The keyword prediction system of the present invention is a keyword prediction system that performs keyword prediction based on information acquired via a communication line, and includes an information management server and a plurality of web servers that provide a homepage, and the information management A server for acquiring television program information via a homepage from any of the web servers; a first extracting unit for extracting common nouns and proper nouns as keywords from the acquired television program information; A second extraction means for sequentially inputting the extracted keywords into a predetermined keyword advice tool obtained via a web page from any of the web servers, and extracting the number of searches for each inputted keyword; Creating means for extracting a program name from program information and creating a program name list; Third extraction means for extracting the audience rating in the past time period, which is the weight of the TV program from the program information, and dividing the number of searches for each of the extracted keywords by the weight of the TV program, Rare degree acquisition means for obtaining a rare degree and rearrangement means for rearranging the rare degree obtained for each keyword in order from the smallest.
The keyword prediction method of the present invention is a keyword prediction method that performs keyword prediction based on information acquired via a communication line, and includes an information management server and a plurality of web servers that provide a homepage, and the information management The server acquires TV program information via a homepage from any of the web servers, extracts a common noun and proper noun that are keywords from the acquired TV program information, A step of inputting the extracted keywords in order into a predetermined keyword advice tool obtained via a web page from a web server, extracting the number of searches for each input keyword, and extracting a program name from the television program information And creating a list of program names and weighting the TV programs from the TV program information. Extracting the audience rating for the past time period, dividing the number of searches for each extracted keyword by the weight of the television program, and determining the rare degree for each keyword, and each keyword And rearranging in descending order of the rareness obtained every time.
In the keyword prediction system and keyword prediction method of the present invention, TV program information is acquired via a homepage from any web server, and common nouns and proper nouns that are keywords are extracted from the acquired TV program information, The extracted keywords are sequentially input to a predetermined keyword advice tool obtained via a web page from any web server, the number of searches for each input keyword is extracted, and the program name is extracted from the TV program information. A list of program names is created, the audience rating for the past time period, which is the weight of the TV program, is extracted from the TV program information, and the number of searches for each extracted keyword is divided by the weight of the TV program. Rareness for each keyword is required, and the rare is calculated for each keyword. Sort order is carried out from one of the small.
Here, as the degree of rarity is smaller, the keyword is related to an event that will be popular in the future. This is because even if the number of keyword searches is small, if the audience rating used as a weight for a TV program is high, the probability that the keyword with a low degree of rarity will become popular in the future is high.

  According to the keyword prediction system and the keyword prediction method of the present invention, the number of searches for each extracted keyword is divided by the weight of the TV program so as to obtain the rare degree for each keyword. In addition to being able to predict keywords related to likely events, it is effective for things such as marketing that accurately grasps customer needs, listing ads that display paid text ads in search engine search results, etc. Can be used.

  In the present embodiment, the information management server acquires TV program information via a home page from any web server, extracts common nouns and proper nouns as keywords from the acquired TV program information, The keywords extracted in order to a predetermined keyword advice tool obtained through a web page from the web server of the web server are sequentially input, the number of searches for each input keyword is extracted, the program name is extracted from the TV program information, and the program is extracted. Create a list of names, extract the audience rating of the past time period, which is the weight of the TV program, from the TV program information, divide the number of searches for each extracted keyword by the weight of the TV program, and create a rare list for each keyword. By calculating the degree and rearranging in order from the lowest rarity obtained for each keyword, In addition to predicting keywords related to events that are likely to be prevalent, effectively use them for things such as marketing that accurately grasps customer needs and listing ads that display paid text ads in search engine search results. I was able to plan.

  Details of the embodiments of the present invention will be described below. FIG. 1 is a diagram for explaining an embodiment of the keyword prediction system of the present invention.

  The keyword prediction system shown in the figure includes a keyword prediction device 100, web servers 210 to 230, and user terminals 310 to 330, which are connected via the Internet 400, which is a communication line. In addition, as a communication line, a dedicated line, a wide area network (WAN: Wide Area Network), etc. can also be used.

  The keyword prediction device 100 is operated by a service provider that predicts a keyword related to an event that will be popular in the future and provides the prediction result to an advertiser who has registered as a member. A management server 120 and a database 130 are provided. Details of predicting a keyword related to an event that will be popular in the future will be described later.

  The information management terminal 110 is operated when changing the system of the information management server 120 or instructing execution of keyword prediction by the control unit 128 described later of the information management server 120. As the information management terminal 110, a desktop personal computer, a notebook personal computer, or the like can be used.

  The information management server 120 acquires information from, for example, various home pages provided by the web servers 210 to 230 in accordance with the operation of the information management terminal 110 and registers the information in the database 130. The information management server 120 has a function of predicting a keyword related to an event that will be popular in the future, and details thereof will be described later.

  The database 130 stores information from the various home pages described above registered by the information management server 120.

  The web servers 210 to 230 provide various home pages to the information management server 120 and the user terminals 310 to 330. Specifically, information such as HTML documents and images stored in the databases 210a to 230a is provided in response to a request through a predetermined search engine, for example.

  The user terminals 310 to 330 browse information such as HTML documents and images from the web servers 210 to 230 via, for example, a predetermined search engine. As the user terminals 310 to 330, a mobile phone, a PDA, a desktop personal computer, a notebook personal computer, or the like can be used.

  FIG. 2 is a block diagram for explaining the internal configuration of the information management server 120. As shown in the figure, the information management server 120 includes an interface unit 121, a receiving unit 122, a transmitting unit 123, a data writing unit 124, a data reading unit 125, a ROM 126, a RAM 127, and a control unit 128.

  The interface unit 121 is in charge of exchanging information with the web servers 210 to 230 via the Internet 400. The receiving unit 122 receives information from the web servers 210 to 230 via the interface unit 121. Here, the information received from the web servers 210 to 230 is obtained from, for example, (http://tv.so-net.ne.jp/guest/pc/portal.action: provided by Sonnet Entertainment Co., Ltd.). TV program information from a homepage as shown in FIG.

  The transmission unit 123 transmits a keyword prediction result related to an event that will be popular in the future to an advertiser who has registered as a member. In addition, although the terminal which an advertiser etc. uses here is not shown in figure here, either of the web servers 210-230 may become an advertiser side terminal depending on the case. In this case, a keyword prediction result related to an event that will be popular in the future is transmitted to any of the web servers 210 to 230.

  The data writing unit 124 writes the above-described various information received by the receiving unit 122 into the database 130. The data reading unit 125 reads information necessary for predicting a keyword related to an event that will be popular in the future from the database 130. That is, in this case, image information such as an image becomes unnecessary, and text information excluding image information is read.

  The ROM 126 stores a control program necessary for control by the control unit 128. Here, the control program necessary for the control by the control unit 128 includes, for example, a program indicating a procedure for predicting a keyword related to an event that will be popular in the future and obtaining the prediction result. The RAM 127 stores information read by the data reading unit 125.

  The control unit 128 controls the operation of each unit according to a control program stored in the ROM 126. That is, the procedure executed by the control unit 128 is shown in the control program. Further, the following main control by the control unit 128 is executed in response to an instruction from the information management terminal 110 described above. That is, the control unit 128 extracts common nouns and proper nouns from the TV program information among the information read by the data reading unit 125. In this case, extraction of common nouns and proper nouns can be performed by a general keyword extraction method. However, when extracting common nouns and proper nouns, it is possible not to extract general names such as programs.

  Further, when the control unit 128 extracts common nouns or proper nouns, for example, as shown in FIG. 4 obtained from (http://inventory.jp.overture.com/d/searchinventory/suggestion/: provided by Overture Inc.). The number of searches corresponding to the keyword in the keyword advice tool shown on the web page is extracted. In this case, the order of input to the keyword advice tool for common nouns and proper nouns extracted from the TV program information is, for example, the order of extraction from the TV program information. Here, the order of extraction from the TV program information is, for example, the order from the smallest channel in the order of early broadcasting.

  In addition, the control unit 128 extracts a program name from the above TV program information and creates a program name list. In addition, the control unit 128 extracts a past audience rating (the audience rating here is an audience rating in a time period, not an audience rating of each program) from the TV program information. This audience rating is used as a weight for a television program. In this case, program search frequency (frequency of TV programs that are easily searched) can be taken into consideration.

  In addition, the control unit 128 divides the number of searches for each keyword in the above keyword advice tool by the weight of the above TV program, and obtains the rarity for each keyword. In addition, the control unit 128 performs rearrangement in order from the lowest rarity obtained for each keyword. Thus, the keywords arranged in order from the one with the lowest degree of rareness are the prediction results of the keywords related to the event that will be popular in the future, and are displayed on the screen of the information management terminal 110 described above. Here, as the degree of rarity is smaller, the keyword is related to an event that will be popular in the future. This is because even if the number of keyword searches described above is small, if the audience rating used as a weight for a television program is high, the probability that a keyword with a low degree of rarity will become popular in the future is high.

  Next, a keyword prediction method in the keyword prediction system described above will be described. First, as shown in FIG. 5, when there is an instruction to execute keyword prediction from the information management terminal 110 to the information management server 120 (step S <b> 1), first, the web server 210 through the Internet 400 is received by the receiving unit 122. The TV program information and the like shown in FIG. 3 are received from any of 230 (step S2).

  Next, when the received TV program information or the like is written into the database 130 by the data writing unit 124 (step S3), it is necessary for the data reading unit 125 to predict a keyword related to an event that will be popular in the future. Information is read from the database 130 (step S4).

  At this time, the control unit 128 extracts common nouns and proper nouns from the TV program information among the information read by the data reading unit 125 (step S5). In this case, when the general name of the program or the like is set not to be extracted, the extraction of the general name is excluded.

  Next, when common nouns and proper nouns are extracted, the extracted common nouns and proper nouns are input to the keyword advice tool shown in FIG. 4 described above, and the number of searches for each keyword is extracted (step). S6). At this time, a program name is extracted from the above TV program information and a program name list is created (step S7). At this time, the past audience rating (the audience rating here is not the audience rating of each program but the audience rating in the time zone) is extracted from the television program information and used as a weight for the television program. In this case, as described above, the program search frequency (the frequency of TV programs that are easily searched) can be taken into consideration.

  Next, the number of keyword searches by the keyword advice tool is divided by the weight of the television program described above, and the rare degree for each keyword is obtained (step S8). At this time, rearrangement is performed in order from the lowest rarity obtained for each keyword (step S9). Thus, the keywords arranged in order from the one with the lowest degree of rareness are the prediction results of the keywords related to the event that will be popular in the future, and are displayed on the screen of the information management terminal 110 described above. Here, as described above, the smaller the rarity, the more relevant the event that will be popular in the future. This is because, as described above, even if the number of keyword searches is small, if the audience rating used as a weighting for a TV program is high, the probability that a keyword with a low degree of rarity will become popular in the future is high. .

  The keyword prediction result related to the event that will be popular in the future obtained in this way is transmitted from the transmission unit 123 to the advertiser who has registered as a member.

  As described above, in this embodiment, the information management server 120 acquires television program information via the home page from any of the web servers 210 to 230, and the common nouns and keywords that are keywords from the acquired television program information. Extract nouns, sequentially input the extracted keywords into a predetermined keyword advice tool obtained via a web page from any of the web servers 210-230, extract the number of searches for each input keyword, The program name is extracted from the TV program information to create a list of program names, the audience rating in the past time period that is the weight of the TV program is extracted from the TV program information, and the number of searches for each extracted keyword is Divide by weighting to find the rarity for each keyword, and the rarity obtained for each keyword Sorting in ascending order, so it is possible to predict keywords related to events that will be popular in the future. For example, marketing that accurately grasps customer needs and paid search engine search results. Can be effectively used for such things as listing advertisements that display text advertisements.

  In the present embodiment, the information source for keyword prediction has been described as television program information. However, the present invention is not limited to this example, and the information source may be news information, market information, or the like distributed via the Internet 400. Any other information may be used.

  The information source is not limited to the information distributed via the Internet 400, but may be information recorded and distributed on a recording medium such as a CD or a DVD. It may be a weekly magazine. In the case of these monthly and weekly magazines, if the characters described therein are captured by a scanner and converted into text data, it is possible to save the trouble of inputting the characters described in the keyboard.

It is a figure for demonstrating one Example of the keyword prediction system of this invention. It is a block diagram for demonstrating the internal structure of the information management server of FIG. It is a figure which shows an example of the homepage which provides television program information. It is a figure which shows an example of the web page of a keyword advice tool. It is a flowchart for demonstrating the keyword prediction method in the keyword prediction system of FIG.

Explanation of symbols

100 keyword prediction device 110 information management terminal 120 information management server 121 interface unit 122 reception unit 123 transmission unit 124 data writing unit 125 data reading unit 128 control unit 130 database 210a to 230a database 210 to 230 web server 310 to 330 user terminal 400 the Internet

Claims (2)

A keyword prediction system that performs keyword prediction based on information acquired via a communication line,
An information management server;
With multiple web servers that provide homepages,
The information management server
Obtaining means for obtaining television program information via a homepage from any of the web servers;
First extraction means for extracting common nouns and proper nouns as keywords from the acquired television program information;
Second extraction means for sequentially inputting the extracted keywords into a predetermined keyword advice tool obtained via a web page from any of the web servers, and extracting the number of searches for each input keyword;
Creating means for extracting a program name from the TV program information and creating a program name list;
A third extraction means for extracting a viewing rate in the past time period which is a weighting of the television program from the television program information;
Rare degree acquisition means for dividing the number of searches for each extracted keyword by the weight of the television program, and obtaining a rare degree for each keyword;
A keyword prediction system, comprising: a sorting unit that sorts the keywords in order from the lowest rarity obtained for each keyword. A keyword prediction method for performing keyword prediction based on information acquired via a communication line,
An information management server;
With multiple web servers that provide homepages,
The information management server
Obtaining television program information via a home page from any of the web servers;
Extracting common nouns and proper nouns as keywords from the acquired TV program information;
Inputting the extracted keywords in order into a predetermined keyword advice tool obtained via a web page from any of the web servers, and extracting the number of searches for each entered keyword;
Extracting a program name from the TV program information and creating a program name list;
Extracting the audience rating of the past time zone which is the weight of the TV program from the TV program information;
Dividing the number of searches for each of the extracted keywords by the weight of the television program, and obtaining a rare degree for each keyword;
And a step of rearranging the keywords in the order of decreasing rarity obtained for each keyword.

Images