JP4859893B2 - Advertisement distribution apparatus, advertisement distribution method, and advertisement distribution control program - Google Patents

Description

  The present invention relates to the fields of an advertisement distribution apparatus, an advertisement distribution method, and an advertisement distribution control program for distributing a web advertisement displayed together with web contents.

  Content match advertisements and content-linked advertisements that place advertisements in accordance with the contents of displayed web contents are known (for example, see Patent Document 1). When an advertiser designates an advertisement to be placed and a keyword related to the advertisement, the advertisement is placed on content (advertisement medium) related to the keyword. It is a mechanism that analyzes text content, keywords, etc. of web content and distributes advertisements highly relevant to the content. Advertisers can place advertisements on websites and web contents that are highly relevant to their products and services, so that a high advertising effect can be expected. In other words, according to such an advertising method, an advertisement that matches the viewer's preference is placed, so that it is possible to improve the effect of attracting customers by the web advertisement and the cost effectiveness.

  By the way, in such content match advertisements, it is preferable that the advertisements to be posted change according to current fluctuations because advertisements can be placed according to the trend of the world. For example, in addition to attracting advertisements for mobile phones to blogs about mobile phones, advertisements for newly released mobile phones and advertisements related to characters used in recent campaigns are also displayed. I want to do it.

As a related technique, there is a technique for posting a news article or the like related to an advertisement as a supplement to the advertisement so as to relate the current affair to the advertisement (for example, see Patent Document 2). Target content (advertisement) is used to create or identify supplemental content (eg, news articles), and the target content and supplemental content are combined to obtain an advertisement.
JP 2008-97351 A JP 2006-519425 A

  However, conventional content match advertisements do not always reflect advertisements that are in line with world trends and current fluctuations. For example, when a content match advertisement is posted on a blog that describes a mobile phone, if the date of the article has passed and the blog content has become obsolete, the blog is now an old model. There may still be advertisements about mobile phones. Also, when a content match advertisement is posted on a blog that describes DVDs or movies, if the date of the article has passed and the blog content has become obsolete, the blog is now an old work. There may be advertisements related to DVDs.

  In this way, the conventional content match advertisement distribution process does not directly reflect the latest trends and current events, and many users are highly interested in new products and services. Distribution of advertisements that attract interest is strongly desired from the viewpoint of advertising effectiveness.

  The invention described in Patent Document 1 can display related content such as a news article in combination with an advertisement to be posted, but only assigns a document related to the advertisement itself. It is not possible to allocate advertisements linked to related current events. In other words, even if the advertiser designates a keyword that matches the current affairs or trend, advertisement allocation using that keyword is not performed.

  The present invention has been proposed in view of the above-described conventional problems. The purpose of the present invention is to distribute advertisements that match content and distribute advertisements that are linked to current affairs. An advertisement distribution apparatus, an advertisement distribution method, and an advertisement distribution control program are provided.

  In order to solve the above-described problems, according to the present invention, as described in claim 1, a news content distribution device that distributes news content is connected via a network and corresponds to web content that is an advertising medium. An advertisement distribution apparatus that distributes advertisements, wherein a first feature word extraction unit that extracts a feature word from text information included in the web content and a feature extracted by the first feature word extraction unit Category determination means for determining a news category to which the web content belongs based on a word; news search means for searching for news content belonging to the news category determined by the category determination means from the news content distribution device; , Included in the news content retrieved by the news retrieval means From the text information, a second feature word extracting unit that extracts a feature word, and a content match advertisement to be distributed to the web content are searched based on the feature word extracted by the second feature word extracting unit. The gist of the present invention is an advertisement distribution device characterized by having an advertisement search means.

Further, as described in claim 2, in the advertisement delivery device according to claim 1, a word representing a news category to which the web content belongs and a related word related to a word representing the news category are provided. It has a category classification database registered in association with it,
The category determining unit may determine a news category to which the web content belongs based on the feature word extracted by the first feature word extracting unit and the category classification database.

  In addition, as described in claim 3, in the advertisement distribution device according to claim 1 or 2, the news determined by the category determining means of the feature words extracted by the first feature word extracting means. First weight value calculating means for calculating a weight value indicating the degree of association with news content belonging to the category, wherein the news search means is based on the weight value calculated by the first weight value calculating means. Thus, the news content can be searched for from the news content belonging to the news category determined by the category determining means.

  In order to solve the above problems, according to the present invention, as described in claim 4, web content that is connected to a news content distribution device that distributes news content via a network and is an advertising medium An advertisement distribution device that distributes advertisements in response to the first feature word extraction means for extracting feature words from the text information included in the web content, and the news content distribution device from the first Based on the feature words extracted by the feature word extraction means, a news search means for searching for news content, and a second feature for extracting feature words from text information included in the news content searched by the news search means Based on the feature words extracted by the word extraction means and the second feature word extraction means, the web controller Are summarized as advertisement distribution device characterized by having an advertisement search means for searching content matching ads to be delivered to Ntsu.

  Further, as described in claim 5, in the advertisement distribution device according to any one of claims 1 to 4, the query log storage device that stores the query word used for the user's search operation and the network are used. The second feature word extraction unit is connected, acquires a query word included in the news content searched by the news search unit from the query log storage device, and extracts the query word from the news content as a feature word To be able to.

  Further, as described in claim 6, in the advertisement distribution device according to any one of claims 1 to 5, the feature search extracted by the second feature word extraction unit is performed by the news search unit. A second weight value calculating unit that calculates a weight value indicating a degree of association with the searched news content, wherein the advertisement search unit includes the feature word extracted by the second feature word extracting unit; The content match advertisement to be distributed to the web content can be searched based on the weight value of the feature word calculated by the second weight value calculation means.

  Further, as described in claim 7, in the advertisement distribution device according to claim 6, the second weight value calculation means includes a feature word at the news body position in the feature word at the news title position of the news content. It is possible to calculate a larger weight value than that.

  Further, as described in claim 8, in the advertisement distribution device according to claim 6 or 7, the advertisement distribution device is connected to a query log storage device that stores a query word used for a user search operation via a network, and When calculating the weight value of the feature word in the news content searched by the news search unit, the second weight value calculation unit stores a query word included in the news content searched by the news search unit in the query log storage It is possible to calculate a weight value larger than a word other than the query word obtained from the device and obtained from the query word.

  In addition, what applied the arbitrary combination of the component of this invention, expression, or a component to a method, an apparatus, a system, a computer program, a recording medium, etc. is also effective as an aspect of this invention.

  According to the present invention, it is possible to provide an advertisement distribution apparatus, an advertisement distribution method, and an advertisement distribution control program that distribute advertisements that match contents and distribute advertisements that are linked to current affairs regarding distribution of content match advertisements. it can.

  The best mode for carrying out the present invention will be described below with reference to the drawings in each embodiment.

<System configuration>
First, before describing specific contents of the present invention, a system configuration for carrying out the present invention will be described.

(network)
FIG. 1 is a network configuration diagram in which an advertisement distribution apparatus 1, a Web server 2, a news server 3, and a terminal 4 according to the present invention are connected via a network 5.

  The advertisement distribution apparatus 1 is an advertisement distribution apparatus (advertisement distribution server) according to the present invention, and distributes advertisements related to content provided by the Web server 2. The Web server 2 is a content server and provides content in response to a content acquisition request from the terminal 4. The news server 3 is a content server related to news in particular and has news content (current information). The terminal 4 is a client terminal for browsing web content including advertisements. For example, it is composed of an information processing device such as a PC (Personal Computer), a PDA (Personal Digital Assistants), and a mobile phone, and includes a web browser (web content browsing software). The network 5 is a network including wired and wireless, for example, an Internet network.

  FIG. 1 shows a configuration example for explaining an embodiment of the present invention, and the advertisement distribution apparatus 1 can be constructed on a server shared with the Web server 2 and the news server 3. Alternatively, an internal function unit and DB (Data Base), which will be described later, may be constructed on different servers and arranged (connected) on the network 5.

(hardware)
Here, the hardware configuration of the advertisement distribution apparatus 1 will be briefly described. The advertisement distribution apparatus 1 includes, as main components, a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), an auxiliary storage device, a storage medium reader, an input device, a display device, and a communication device. Etc.

  The CPU is composed of a microprocessor and its peripheral circuits, and is a circuit that controls the entire advertisement distribution apparatus 1. The ROM is a memory that stores a predetermined control program (software component) executed by the CPU. The RAM executes various control operations by executing a predetermined control program (software component) stored in the ROM. This is a memory used as a work area (work area) when performing.

  The auxiliary storage device is a device that stores a general-purpose OS (Operating System), various DBs related to advertisements, a program according to the present invention, and the like, and an HDD (Hard Disk Drive) that is a nonvolatile storage device is used. In addition to the auxiliary storage device, the various types of information may be stored in a storage medium such as a CD-ROM (Compact Disk-ROM) or a DVD (Digital Versatile Disk) or other media. Various stored information can be read through a drive device such as a storage medium reader. Therefore, various information can be obtained by setting the recording medium in the storage medium reading device as necessary.

  The input device is a device for a user (administrator) to perform various input operations. The input device includes a mouse, a keyboard, and the like. The display device is a device that displays a display screen. For example, it is composed of LCD (Liquid Crystal Display), CRT (Cathode Ray Tube) and the like. Next, the communication device is a device that communicates with other devices via a network. Supports communication according to various network forms including wired and wireless networks.

(function)
FIG. 2 is a functional block diagram showing main functions of an embodiment of the advertisement distribution apparatus 1 and the Web server 2 according to the embodiment of the present invention. The advertisement distribution apparatus 1 includes, as main functions, a content reception unit 101, a content feature word extraction unit 102, a category classification unit 103, a news search unit 104, a news feature word extraction unit 105, a content match advertisement search unit 106, and an advertisement transmission. The unit 107 is included. In addition, a DB includes a category classification DB 111 and a content match advertisement DB / bid word DB 112.

  The content receiving unit 101 has a function of receiving web content from the web server 2. You may receive from the web server 2 in the text information format already extracted from the web content.

  The content feature word extraction unit 102 has a function of extracting feature words together with weight values from web content. The feature words are extracted by performing a feature word extraction process from words included in the web content. For example, for words in web content, feature words that calculate weights corresponding to their appearance frequency, appearance location, character attributes, etc. as values (weight value calculation unit) and characterize the web content with words having higher weight values Extract as

  The category classification unit 103 has a function of referring to the category classification DB 111 and classifying into categories to which web contents belong based on feature words (and weight values). As will be described in detail later, the category classification DB 111 stores a category associated with a news category on the news server 3 and a related word associated with the category (name) in association with each other in order to categorize content. DB. There are various methods for acquiring related words, but as an example, a category name is thrown to a search engine, and related words are acquired from a snippet of a search result. A snippet is a text displayed under the title of a search result by a search engine.

  The news search unit 104 has a function of searching for news content from the news server 3. Search for news is performed based on the category of the web content classified by the category classification unit 103. Specifically, the news of the same category as the content category is searched. The news search is performed based on the feature word (and weight value) extracted by the content feature word extraction unit 102. As described above, the news search unit 104 searches for news based on either a category or a feature word. Details will also be described later.

  The news feature word extraction unit 105 has a function of extracting feature words (and weight values) from the news searched by the news search unit 104. The feature word may be the same as the method described above, and is extracted by performing a feature word extraction process from words included in the news content. For example, for a word in news content, a weight corresponding to its appearance frequency, appearance location (particularly, news title part), character attribute, etc. is calculated as a value (weight value calculating part), and a word having a higher weight value is calculated. It is extracted as a feature word that characterizes the news content.

  The news feature word extraction unit 105 refers to the query log DB 301 when extracting feature words (and weight values) from the news searched by the news search unit 104. The query log DB 301 is a log in which news and a query word (search word) corresponding to the news are recorded. When a feature word (and weight value) is extracted from the news, the query log DB 301 can be reflected in determination of the weight value of the feature word. Details will be described later.

  The content match advertisement search unit 106 searches for a content match advertisement that matches the feature word extracted by the news feature word extraction unit 105 from the content match advertisement DB 112 using a bid word, and determines an advertisement to be distributed. It has a function. This content match advertisement will be described later in detail.

  The advertisement transmission unit 107 has a function of transmitting, to the Web server 2, the content match advertisement to be distributed that has been searched by the content match advertisement search unit 106.

  Next, the function of the Web server 2 will be briefly described. The Web server 2 has a general Web server function, and includes a page request reception unit 201, a content transmission unit 202, an advertisement reception unit 203, a page creation unit 204, and a page transmission unit 205 as main functions. is there.

  The page request reception unit 201 receives a request for a web content page from the terminal 4. The content transmission unit 202 transmits the target content from the content DB 206 via the page creation unit 204 to the advertisement distribution apparatus 1 in order to receive distribution of an advertisement to be displayed (embedded) on the web content page. To do. The advertisement receiving unit 203 receives an advertisement (content match advertisement) from the advertisement distribution device 1. The page creation unit 204 creates a web content page that is finally provided to the terminal 4 (and thus the user) by embedding an advertisement in the web content page. The page transmission unit 205 transmits the final web content page to the terminal 4.

  The overall configuration of the advertisement distribution device 1 according to the embodiment of the present invention has been described above. In this configuration, the (processing) operation of the advertisement distribution apparatus 1 will be described below.

<Operation>
FIG. 3 is a flowchart showing an example of an advertisement distribution process of the advertisement distribution apparatus 1 according to the present invention. The following will be described with reference to the timely drawings.

  In step S301, text information is read from the web content received from the web server 2.

  In step S302, the content feature word extraction unit 102 performs feature word extraction processing from the read text information. First, words included in the text information are extracted, and a weight value is calculated for each word. The weight value is a value indicating the degree of association between the web content and the word. The weight value is calculated based on an element that is a basis for calculating the degree of association, for example, as described above, the appearance frequency, appearance location, character attribute (typeface data) of the word in the web content. Words with a high frequency of appearance in the web content have the meaning of directly characterizing the web content, and words located in places indicating titles in the web content have the meaning of characterizing the web content in summary. Have. Also. Focusing on character attributes (font size, bold, underlined characters, color characters, etc.), the web content creator has attached the font to the word with some intention, so the meaning of characterizing the web content as well Because it is considered to have.

  FIG. 4 is a diagram illustrating a state in which feature words and their weight values are extracted from web content. In this example, the content is a so-called blog, but this content is divided into a title block and a text block. The word located in the title block often includes a title or a word equivalent thereto, and well represents the content and topic. Therefore, a large weight value is given to the word located in the title block. Also, a large weight value is assigned to a word whose character font is larger than other characters. Therefore, a large weight value is given to the words “housewife”, “child care”, and the like. In addition, since the word “exercise” has a high appearance frequency and is underlined in some places, a large weight value is similarly given.

  In addition, as a weight value, it can be considered how much the word is used in general. You can simply increase the weight value according to the number of hits as a result of searching for that word in the search engine, or the proper noun is more relevant to the product name or the like than the general noun (common noun). Similarly, the weight value may be increased.

  As an example of such weighting, TF-IDF, which is an algorithm (natural language processing) for extracting characteristic words in sentences, is known. The TF-IDF value is obtained from the TF value representing the appearance frequency of the word obtained from the content and the DF value (or IDF value that is the reverse appearance frequency) that is the appearance frequency obtained as a result of searching the corpus. Feature words are extracted by using only those having a certain threshold value or more as feature words. The corpus is like a database containing a large number of words, and each word and its appearance frequency (DF value) are stored. General words appearing in many documents have a role of decreasing the importance and increasing the importance of words appearing only in a specific document. The feature word extraction condition is, for example, “all words whose TF-IDF value is equal to or greater than a certain threshold value”, “the top N words among words whose TF-IDF value is equal to or greater than a certain threshold value”, and the like.

  In this way, what words are extracted as feature words can be determined elastically. The calculated weight value is normalized so that numerical values can be easily handled.

  Returning to FIG. 3, the process proceeds to step S303, where the web content is classified into categories based on the feature words extracted by the content feature word extraction unit 102, and then news is searched based on the categories (news search unit 104). The process branches depending on whether the news is searched (news search unit 104) based on the feature word without being classified into categories. Further, news may be searched based on both the category and the feature word (news search unit 104), and finally an appropriate result may be selected.

  In step S304, the category classification unit 103 refers to the category classification DB 111 and classifies the web content into categories based on the feature words extracted in step S302. FIG. 5 is a schematic diagram showing the data structure of the category classification DB 111. The category classification DB 111 is a DB used for classifying web content based on feature words extracted from the content. The category corresponding to the news category (news field) on the news server 3 and the category ( This is a DB that stores related words related to (name). There are various methods of acquiring related words. For example, a category (name) is thrown to a search engine, and related words (which may be frequent words or feature words) are acquired together with a weight value from a snippet of a search result. Also, TF-IDF may be used here. FIG. 6 is a diagram illustrating a state in which feature words extracted from content are classified into (temporary) categories. Some feature words have overlapping categories, but they are classified for each category even if they overlap.

  Next, the category classification unit 103 calculates a score from the weight value of the feature word and the weight value of the related word. The score is one index representing the degree of association for each category with respect to the content. FIG. 7 is a diagram illustrating how scores are calculated for each category. Here, it can be calculated by multiplying the weight value of the feature word and the related word. Although weight values have been considered so far in order to improve accuracy, there is also a method of simply calculating a score based on the number of appearances of feature words. For example, if the feature word appears three times in the category “childcare”, the feature word appears once in the category “region”, and the feature word appears twice in the category “health”, the score is 3, 1 respectively. 2 is possible. That is, the score may be an index representing the degree of association for each category with respect to the content.

  FIG. 8 is a diagram illustrating a state in which categories are rearranged in the order of scores and upper categories are extracted. Although the score is calculated by the category classification unit 103, only the predetermined upper category is extracted by rearranging in order of score value (descending order). Here, the top one category is extracted, and the lower categories are drawn and are not considered in the news search. Thus, in this example, the web content is finally classified into the category “childcare”.

  Returning to FIG. 3, the process proceeds to step S <b> 305, and the news search unit 104 searches for news from the news server 3. As described above, the news search unit 104 searches for news based on either the category or the feature word of the web content.

  First, news is searched based on the category of web content. FIG. 9 is a diagram illustrating an example of news content belonging to the news category “childcare” searched (acquired) from the news server 3. Since the web content is classified into the category “childcare” by the category classification unit 103, the news server 3 searches the news content belonging to the news category “childcare”, for example, the latest (latest) five. As described above, since the category of the category classification DB 111 and the news category from the news server 3 have a corresponding relationship, the news search unit 104 can easily search for the corresponding news content. Further, the relationship between the feature word extracted by the content feature word extraction unit 102 and the news content belonging to the category of the news classified by the category classification unit 103 so as to search for news whose document distance is close to the web content. Based on the calculated weight value, the news search unit 104 can also search for news content from the news content belonging to the classified news category based on the calculated weight value. For calculating the weight value, TF-IDF or the like can be used as described above.

  Next, a process for searching for news based on the feature words (and weight values) extracted by the content feature word extraction unit 102 will be described. Returning to FIG. 4, a state in which feature words (and weight values) are extracted from the web content is shown. News is searched based on this feature word (and weight value). For example, news content including a characteristic word is searched from news content updated most recently (for example, the same day). The match score may be calculated from the feature word weight value and the news content feature word weight value. For example, using a vector space model or the like, a match score representing the degree of matching (relevance) with news content that matches a feature word is calculated. The vector space model is an algorithm for performing information retrieval. The weight value is regarded as a vector and expressed in a vector space, and the correlation amount is calculated by cosine, inner product, distance, etc., and the degree of association is obtained. The news content is rearranged in the descending order of the calculated match score, and a predetermined upper one is extracted. In this way, the match score for each news content is calculated, and finally the news content with a predetermined higher match score is obtained.

  As described above, in step 305, the news search unit 104 searches for news content based on either the category or feature word of the web content. When searching for news content based on a category of web content, the news categories of a plurality of searched news content are all the same. On the other hand, when searching for news content based on one of the feature words, it depends on the feature word (and weight), and therefore, the news category is not necessarily the same, and there are various cases.

  Here, when topics that belong to relatively the same category are taken up as web content, even if the news content is searched based on either the category or feature word of the web content, relatively similar news content is obtained. Is likely to be searched. On the other hand, when topics belonging to a plurality of relatively different categories are taken up, there is a high possibility that different news contents are searched. When the search of news content is based on the category of web content, since the advertisement is determined by focusing on the advertisement related to a specific category, a high advertising effect may be obtained. On the other hand, it comes with a sacrificial topic. In addition, when the search of news content is based on feature words, related advertisements are determined over a wide range of feature words appearing in web content, so it is considered that balanced advertisement delivery is achieved.

  The process proceeds to FIG. 3, step S306. When feature words (and weights) are extracted from the news content searched in step S305 with reference to the query log DB 301, a query log corresponding to the news content is acquired in step S307. If the query log DB 301 is not referred to (considered), the process proceeds to step S308 without acquiring the query log.

  Here, the query word here is a word (word) associated with the news content and is a search word input to the search window of the search engine. More specifically, it is a search term that is input to the search window immediately after a user browses a certain news content. In general, when a user is interested in a specific word in a news article, a search engine is often used to search for that word. Therefore, it can be said that the word entered in the search window immediately after browsing the news after tracking the behavior history of the user who browsed the news is the word that the user showed interest in the news, that is, the feature word itself. The query log DB 301 is a log in which a news content identifier for identifying (specifying) news content of the news server 3 and a query word (search word) corresponding to the news content are recorded. ) Can be reflected in feature word extraction and weight value determination. FIG. 10 is a diagram illustrating an example of a query log.

  In step S308 in FIG. 3, the news feature word extraction unit 105 extracts feature words and weight values from the news content searched in step S305. As described above, the extraction method calculates a weight corresponding to the appearance frequency, appearance location, character attribute, and the like of a word in news content as a value (weight value calculation unit), and selects a word having a higher weight value. What is necessary is just to extract as a characteristic word which characterizes the news content. Also, TF-IDF may be used here.

  Further, it is possible to weight words focusing on points unique to news content. In general, a title (heading) is attached to news content. Since the title indicates the news content in summary, the accuracy of the content match (match score) with the assigned advertisement can be improved by giving a larger weight value to the word included in the title.

  If a query log is acquired, feature words and weight values are extracted reflecting the query log. Specifically, a large weight value is assigned to a word recorded in the query log of a certain news content. This is because the word can be said to be a word that shows interest to a considerable degree when the user browses the news content.

  FIG. 11 is a diagram illustrating a state in which feature words and weight values are extracted from news content. As described above, what word is extracted as a feature word can be determined elastically, but the news feature word extraction unit 105 of the present invention pays particular attention to points unique to news content. Word weighting and weighting reflecting the query log are performed.

  In step S309 in FIG. 3, the content match advertisement search unit 106 searches for a content match advertisement based on the feature word and the weight value extracted from the news content. The content match advertisement search unit 106 searches and determines a content match advertisement that matches the feature word (and weight value) from the content match advertisement DB 112 using the bid word.

  FIG. 12 is a schematic diagram showing the data structure of the content match advertisement DB 112. The content match advertisement DB 112 is a DB in which an advertisement (advertisement content) purchased by an advertiser and a bid word are registered in association with each other. An advertiser designates an advertisement and a bid word (keyword) related to the advertisement and purchases the advertisement. The bid word is a keyword for matching the advertisement to be distributed with the content of the advertisement medium.

  FIG. 13 is a diagram showing how the content match advertisement search unit 106 uses the bid word DB 112 to calculate a match score representing the degree of matching between the content that is the advertisement medium and the bid word. In the bid word DB 112, a bid word and its related word (and weight value) are registered in advance in association with each other. Based on the weight value of the feature word extracted from the news content by the news feature word extraction unit 105 and the weight value of the related word, for example, using a vector space model or the like, the fitness of the content that is the advertising medium and the bid word ( A match score representing the degree of association is calculated. The vector space model is an algorithm for performing information retrieval. The weight value is regarded as a vector and expressed in a vector space, and the correlation amount is calculated by cosine, inner product, distance, etc., and the degree of association is obtained. The bid words are rearranged in the descending order of the calculated match score, and a predetermined higher bid word is extracted. Then, an advertisement corresponding to the bid word is determined (searched) in the content match advertisement DB 112.

  In FIG. 3, the process proceeds to step S <b> 309, and since the advertisement to be distributed is determined by the content match advertisement search unit 106, the advertisement transmission unit 107 transmits the advertisement to the Web server 2. When the web server 2 receives the advertisement, the web server 2 embeds the advertisement (advertisement information) in the content (advertisement medium), and finally transmits it as a web content page to the user terminal 4. Then, a web content page including an advertisement is displayed on the terminal 4 by a web browser or the like. FIG. 14 is an example of the displayed web content page. Advertisements related to web contents (blogs), which are advertising media, are also distributed that reflect the latest news (trends). The advertisement includes a text advertisement and an image advertisement (banner) format. In addition, link information such as URL (Uniform Resource Locator) is embedded in the text advertisement and the image advertisement.

<Summary>
As described above, the advertisement distribution apparatus 1 according to the present invention searches for news content that matches web content, and allocates advertisements based on the news content. News content has information on trends and current affairs, so it can directly reflect the latest trends and current affairs when matching with ads, and deliver ads that attract more users' interests. As a result, improvement in advertising effectiveness can be expected.

  From the advertiser's side, even if the web content (advertising media) does not include content that matches current affairs or trends, if the advertiser specifies keywords (bid words) that match current affairs or trends. You can allocate ads for that keyword.

  In addition, when an advertisement is assigned to a web content, a feature word that can be easily assigned an advertisement such as a product name or a company name that cannot be extracted from the web content can be extracted by interposing news.

  Also, by searching for news content based on the degree of association between web content and news content, it is possible to extract feature words based on news that is close to the content of the web content. Can be used for ads linked to.

  As described above, according to the present invention, an advertisement distribution apparatus, an advertisement distribution method, and an advertisement distribution control program that distribute advertisements that match contents and distribute advertisements that are linked to current affairs are provided for distribution of content match advertisements. It becomes possible.

  Although the present invention has been described with reference to preferred embodiments and specific specific examples of the present invention, these specific examples have been described without departing from the broad spirit and scope of the present invention as defined in the claims. Obviously, various modifications and changes can be made to the. In other words, the present invention should not be construed as being limited by the details of the specific examples and the accompanying drawings.

1 is a network configuration diagram according to the present invention. FIG. It is a functional block diagram which shows the main functions of one Embodiment of the advertisement delivery apparatus and web server which concern on embodiment of this invention. It is a flowchart which shows the example of an advertisement delivery process of the advertisement delivery apparatus by this invention. It is a figure which shows a mode that the feature word and its weight value were extracted from the web content. It is a schematic diagram which shows the data structure of category classification DB. It is a figure which shows a mode that a feature word is classified into categories. It is a figure which shows a mode that a score is calculated for every category. It is a figure which shows a mode that a high-order category is extracted. It is an example of news content. It is a figure which shows a mode that a query log is extracted. It is a figure which shows a mode that the feature word and its weight value were extracted from news content. It is a schematic diagram which shows the data structure of content match advertisement DB. It is a figure which shows a mode that a match score is calculated. It is an example of the web content page displayed.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Advertisement delivery apparatus 2 Web server 3 News server 4 Terminal 5 Network 101 Content receiving part 102 Content feature word extraction part 103 Category classification | category part 104 News search part 105 News feature word extraction part 106 Content match advertisement search part 107 Advertisement transmission part 111 Category Classification DB
112 Content Match Ad DB / Bid Word DB
201 page request reception unit 202 content transmission unit 203 advertisement reception unit 204 page creation unit 205 page transmission unit 206 content DB
301 Query log DB

Claims (11)

An advertisement distribution apparatus that is connected to a news content distribution apparatus that distributes news content via a network and distributes advertisements according to web content that is an advertisement medium,
First feature word extraction means for extracting feature words from text information included in the web content;
Category determining means for determining a news category to which the web content belongs based on the feature words extracted by the first feature word extracting means;
News search means for searching for news content belonging to the news category determined by the category determination means from the news content distribution device;
Second feature word extraction means for extracting feature words from text information included in the news content searched by the news search means;
Advertisement search means for searching for a content match advertisement to be distributed to the web content based on the feature words extracted by the second feature word extraction means;
An advertisement distribution apparatus comprising: A category classification database in which a word representing a news category to which the web content belongs and a related word related to a word representing the news category are registered in association with each other;
The category determining unit determines a news category to which the web content belongs based on the feature word extracted by the first feature word extracting unit and the category classification database;
The advertisement distribution apparatus according to claim 1. First weight value calculating means for calculating a weight value indicating a degree of association between the feature word extracted by the first feature word extracting means and the news content belonging to the news category determined by the category determining means; Have
The news search means, based on the weight value calculated by the first weight value calculation means, to search for news content from the news content belonging to the news category determined by the category determination means;
The advertisement distribution device according to claim 1 or 2, wherein An advertisement distribution apparatus that is connected to a news content distribution apparatus that distributes news content via a network and distributes advertisements according to web content that is an advertisement medium,
First feature word extraction means for extracting feature words from text information included in the web content;
News search means for searching for news content based on the feature words extracted by the first feature word extraction means from the news content distribution device;
Second feature word extraction means for extracting feature words from text information included in the news content searched by the news search means;
Advertisement search means for searching for a content match advertisement to be distributed to the web content based on the feature words extracted by the second feature word extraction means;
An advertisement distribution apparatus comprising: Connected to a query log storage device for storing a query word used for a user's search operation via a network;
The second feature word extraction unit obtains a query word included in the news content searched by the news search unit from the query log storage device, and extracts the query word from the news content as a feature word;
The advertisement distribution device according to any one of claims 1 to 4, wherein Second weight value calculating means for calculating a weight value indicating a degree of association between the feature word extracted by the second feature word extracting means and the news content searched by the news searching means;
The advertisement search means is distributed to the web content based on the feature word extracted by the second feature word extraction means and the weight value of the feature word calculated by the second weight value calculation means. Search for content match ads to do,
The advertisement distribution device according to any one of claims 1 to 5, wherein: The second weight value calculating means calculates a weight value larger than the feature word at the news body position for the feature word at the news title position of the news content;
The advertisement distribution apparatus according to claim 6. Connected to a query log storage device for storing a query word used for a user's search operation via a network;
The second weight value calculating means calculates a query word weight value included in the news content searched by the news search means when calculating a weight value of a feature word in the news content searched by the news search means. Obtaining from the storage device, and calculating a greater weight value for the obtained query word than words other than the query word;
The advertisement distribution device according to claim 6 or 7. A word representing a news category to which a web content serving as an advertising medium belongs is associated with a news content distribution apparatus that distributes news content, and a related word related to the word representing the news category is associated with each other. An advertisement distribution method in an advertisement distribution apparatus that includes a registered category classification database and distributes advertisements according to the web content,
A first feature word extraction step of extracting feature words from text information included in the web content;
A category determination step for determining a news category to which the web content belongs based on the feature words extracted by the first feature word extraction step;
A news search step of searching for news content belonging to the news category determined by the category determination step from the news content distribution device;
A second feature word extraction step of extracting a feature word from text information included in the news content searched by the news search step;
An advertisement search step of searching for a content match advertisement to be distributed to the web content based on the feature words extracted by the second feature word extraction step;
An advertisement distribution method characterized by comprising: An advertisement distribution method in an advertisement distribution apparatus that is connected to a news content distribution apparatus that distributes news content via a network and distributes advertisements according to web content that is an advertisement medium,
A first feature word extraction step of extracting feature words from text information included in the web content;
A news search step of searching for news content based on the feature word extracted by the first feature word extraction step from the news content distribution device;
A second feature word extraction step of extracting a feature word from text information included in the news content searched by the news search step;
An advertisement search step of searching for a content match advertisement to be distributed to the web content based on the feature words extracted by the second feature word extraction step;
An advertisement distribution method characterized by comprising:   An advertisement distribution control program for causing a computer to execute the advertisement distribution method according to claim 9.

Images