JP2017062525A - Advertisement device, advertisement extraction method, advertisement extraction program, terminal device, and terminal control program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enhance advertising effects.SOLUTION: An information display device of the application includes a calculation unit and an extraction unit. The calculation unit calculates index values that indicate the degrees of association between advertisement contents of distribution candidates which are listed together with contents and contents which are displayed around arrangement positions where the advertisement contents of the distribution candidates are arranged. The extraction unit extracts advertisement contents of distribution targets from the advertisement contents of distribution candidates, on the basis of the index values calculated by the calculation unit.SELECTED DRAWING: Figure 3

Description

  The present invention relates to an advertisement device, an advertisement extraction method, an advertisement extraction program, a terminal device, and a terminal control program.

  In recent years, with the rapid spread of the Internet, advertisement distribution via the Internet has been actively performed. For example, advertisement distribution is performed in which advertisement content such as a company or a product is displayed in an advertisement frame set on a web page, and when the advertisement content is clicked, the transition is made to the advertiser's web page.

  In such advertisement distribution, techniques for distributing advertisements that are of interest to users have been proposed in order to increase advertisement effectiveness. For example, a keyword is identified from content that presents an article posted by a user and an article posted by another user associated with the user in time series, and an advertisement related to the identified keyword is distributed as an advertisement that is displayed together with the content. The technique to do is proposed (patent document 1).

JP 2014-750006 A

  However, the above-described conventional technology does not always increase the advertising effect. Specifically, the above-described conventional technology is merely a technique for distributing advertisements of interest to users based on posted articles for each user using a web posting service such as Twitter (registered trademark), and the existence of posted articles. In the advertisement distribution that is not performed, that is, in the advertisement distribution that is widely performed in general, the advertisement effect is not necessarily improved.

  The present application has been made in view of the above, and an object thereof is to provide an advertising device, an advertising extraction method, an advertising extraction program, a terminal device, and a terminal control program that can enhance the advertising effect.

The advertising device according to the present application calculates an index value that indicates a degree of association between a distribution candidate advertisement content displayed together with the content and a content displayed around an arrangement position where the distribution candidate advertisement content is arranged. And an extraction unit that extracts the distribution target advertisement content from the distribution candidate advertisement content based on the index value calculated by the calculation unit.

  According to one aspect of the embodiment, there is an effect that the advertising effect can be enhanced.

FIG. 1 is a diagram illustrating an example of advertisement extraction processing according to the first embodiment. FIG. 2 is a diagram illustrating a configuration example of the advertisement extraction system according to the first embodiment. FIG. 3 is a diagram illustrating a configuration example of the advertising device according to the first embodiment. FIG. 4 is a diagram illustrating an example of the advertisement information storage unit according to the first embodiment. FIG. 5 is a diagram illustrating an example of an index value calculation process according to the first embodiment. FIG. 6 is an explanatory diagram for specifically explaining the advertisement extraction processing according to the first embodiment. FIG. 7 is a flowchart illustrating an advertisement extraction processing procedure by the advertising device according to the first embodiment. FIG. 8 is a diagram illustrating an example of category classification according to the modification example of the first embodiment. FIG. 9 is a diagram illustrating a configuration example of the advertising device according to the second embodiment. FIG. 10 is a diagram illustrating an example of advertisement extraction processing by the advertisement extraction system according to the second embodiment. FIG. 11 is a diagram illustrating a configuration example of an advertisement extraction system according to the third embodiment. FIG. 12 is a diagram illustrating a configuration example of a user apparatus according to the third embodiment. FIG. 13 is a diagram illustrating an example of advertisement extraction processing according to another embodiment. FIG. 14 is a hardware configuration diagram illustrating an example of a computer that realizes the function of the advertising device.

  Hereinafter, an embodiment (hereinafter, referred to as “embodiment”) for carrying out an advertisement device, an advertisement extraction method, an advertisement extraction program, a terminal device, and a terminal control program according to the present application will be described with reference to the drawings. Note that the advertisement apparatus, the advertisement extraction method, and the advertisement extraction program according to the present application are not limited by this embodiment. Moreover, in the following embodiment, the same code | symbol is attached | subjected to the same site | part and the overlapping description is abbreviate | omitted.

(First embodiment)
[1. (Advertising extraction process)
Hereinafter, an example of the advertisement extraction process according to the first embodiment will be described with reference to FIG. FIG. 1 is a diagram illustrating an example of advertisement extraction processing according to the first embodiment. FIG. 1 shows an example in which advertisement extraction processing is performed by the advertising device 100.

  The advertising device 100 accepts submission of advertising content from the advertiser. Then, when the advertising device 100 receives an advertising request that is an acquisition request for the advertising content included in the web page W1 from the user terminal 10, the advertising device 100 receives the advertising content received from the advertiser, that is, the advertising content to be distributed from the advertising content of the distribution candidate. Extract content. Specifically, the advertising device 100 calculates an index value indicating the degree of association between the distribution candidate advertising content and the content displayed around the arrangement position where the advertising content is arranged on the web page W1. Then, the advertising device 100 extracts the distribution target advertisement content from the distribution candidate advertisement content based on the calculated index value.

  First, the web page W1 according to the first embodiment will be described. As shown in FIG. 1 (a), the web page W1 according to the embodiment has a plurality of contents arranged in the vertical direction. In addition, a plurality of contents are displayed as a list on the display screen of the user terminal 10.

  As described above, the plurality of contents displayed in a list are roughly classified into article contents and advertisement contents. The article content is, for example, a news article, entertainment information, a blog, etc., and indicates content other than advertising content. Advertising content is placed between article content. The advertisement content of such a posting method is sometimes called “in-feed advertisement” or the like.

  Further, in FIG. 1A, an example in which seven contents including advertisement contents are arranged in a vertical row is illustrated as a web page W1, but in this embodiment, the web page W1 includes 21 contents. Are arranged in a vertical line. In such a case, in response to a single request from the user terminal 10, the 21 contents including the advertisement content are not delivered to the user terminal 10 as the web page W1 at a time, but a predetermined number (for example, 7) ) Are distributed one by one, and finally, a web page W1 including 21 contents is formed.

  For example, after the web page W1a in which seven contents are arranged in a vertical row is initially displayed, the web page W1a is moved upward by a user operation, and the seventh content from the top (of FIG. 1A) In the example, when the article content F) is included in the display screen, the user terminal 10 transmits a request again, thereby acquiring the web page W1b in which seven contents are arranged in a vertical line. And user terminal 10 displays what connected web page W1b under web page W1a as web page W1.

  So far, the web page W1 according to the first embodiment has been described. Below, the advertisement extraction process by the advertising device 100 will be specifically described.

  For example, it is assumed that a request for a web page W1a, which is a part to be initially displayed, is transmitted from the user terminal 10 among the web pages W1 including 21 contents. In response to this, as shown in FIG. 1B, the content server 30 determines the article contents C2 to C6 as distribution target article contents and also arranges the article contents C2 to C6 in order from the top. And

  In addition, the content server 30 transmits to the advertising device 100 the list 1 that associates the text including the title, description, and the like of each determined article content with the arrangement order. For example, in the list 1, for the article content C1, the content server 30 associates the text information of the descriptive text “Drama X 10th episode audience rating is 11.5%” with the arrangement order “1”.

  Here, when seven contents are arranged in a vertical line, it is assumed that it is determined in advance to insert the advertisement contents at the arrangement position L1, which is an intermediate position. When receiving the list 1 from the content server 30, the advertising device 100 calculates an index value indicating the degree of association between the distribution candidate advertisement content and the article content displayed around the arrangement position L1. The index value indicating the degree of association is a numerical value of the degree of relevance between the advertisement content and the article content. The higher the index value indicating the degree of association, the higher the relationship between the two. In the following embodiments, an index value indicating the degree of association may be simply expressed as “index value”.

  Further, in the present embodiment, “article content displayed around the placement position of the advertisement content” refers to up to two pieces of article content from the placement position of the advertisement content. In the example of FIG. 1B, the article content displayed around the arrangement position L1 indicates article contents C2 to C5. The “article content displayed around the placement position of the advertisement content” is not limited to such an example. For example, the article content indicates up to three articles content from the placement position of the advertisement content. Also good. Further, it may indicate up to four pieces of article content and up to two pieces of article content from the arrangement position of the advertisement content.

  As shown in FIG. 1B, the advertising device 100 has advertisement contents AD1 to AD3 as advertisement contents to be distributed. Therefore, when receiving the list 1, the advertising device 100 calculates the similarity between each distribution candidate advertisement content and each of the article contents C2 to C5, and calculates the sum of the calculated similarities as an index value.

  In FIG. 1B, the advertising device 100 calculates four similarities between the advertising content AD1 and each of the article contents C2 to C5, and calculates the sum of the calculated four similarities as the index value “S1”. An example is shown. Similarly, the advertising device 100 calculates four similarities between the advertising content AD2 and each of the article contents C2 to C5, and the sum of the calculated four similarities is calculated as the index value “S2”. Indicates. Furthermore, an example is shown in which the advertising device 100 calculates the index value “S3” for the advertising content AD3.

  Here, the advertising content AD3 is advertising content related to sports. Of the four article contents C2 to C5, two of the article contents C3 and C4 are article contents related to sports. Therefore, the advertising device 100 calculates the index value “S3” between the advertising content AD3 and the article content C2 to C5 to be the highest. Then, the advertising device 100 extracts the advertising content AD3 for which the highest index value “S3” has been calculated as the distribution target advertising content from the distribution candidate advertising content.

  As a result, the web pages W1a arranged from the top in the order of the article contents C1, C2, C3 and the advertisement contents AD3, C4, C5, C6 are distributed to the user terminal 10.

  Whether or not the user is interested in the in-feed advertisement shown in FIG. 1 and clicks may be influenced by article content around the advertisement content. For example, when sports-related advertising content is inserted while sports-related article content is gathered, before the sports-related advertising content enters the user's field of view, the sports-related article content enters the user's interest in sports. Will increase. This tends to make it easier for the user to click on advertising content related to such sports.

  On the other hand, for example, when fashion-related advertising content is inserted while article content related to sports is gathered, it is unlikely that a user who has increased interest in sports clicks on advertising content related to fashion.

  That is, the advertising device 100 according to the first embodiment considers the user characteristics as described above, calculates an index value indicating the degree of association between the advertising content and the article content, and thereby the surrounding article content and Since the advertising content with higher affinity can be extracted and the extracted advertising content can be displayed, the advertising effect can be enhanced.

[2. System configuration
The configuration of the advertisement extraction system 1 according to the first embodiment will be described with reference to FIG. FIG. 2 is a diagram illustrating a configuration example of the advertisement extraction system 1 according to the first embodiment. As shown in FIG. 2, the advertisement extraction system 1 includes a user terminal 10, an optimization server 20, a content server 30, and an advertising device 100. The user terminal 10, the optimization server 20, the content server 30, and the advertising device 100 are connected to be communicable via wire or wireless via a network. The advertisement extraction system 1 shown in FIG. 2 may include a plurality of user terminals 10, a plurality of optimization servers 20, a plurality of content servers 30, and a plurality of advertisement devices 100. .

  The user terminal 10 is an information processing device used by a user. For example, the user terminal 10 is a mobile phone such as a smartphone, a tablet terminal, a PDA (Personal Digital Assistant), a desktop PC (Personal Computer), a notebook PC, or the like. For example, the user terminal 10 transmits an acquisition request (request) for the web page W1 that is list information including the plurality of contents illustrated in FIG. Further, the user terminal 10 displays the web page W1 received from the optimization server 20 on the display screen.

  The optimization server 20 is a front-end server that directly transmits and receives information to and from the user terminal 10. For example, when receiving the request for the web page W1 from the user terminal 10, the optimization server 20 transmits a content request that is an article content acquisition request to the web server 30, and also receives an advertisement content acquisition request from the advertising device 100. Send an ad request that is. Further, the optimization server 20 generates a web page W1 using the article content received from the content server 30 and the advertisement content received from the advertising device 100, and distributes the generated web page W1 to the user terminal 10. In contrast to such an optimization server 20, the content server 30 and the advertising device 100 can be said to be back-end servers.

  The content server 30 is a server device that distributes article content to the optimization server 20. For example, when the content server 30 receives a content request from the optimization server 20, the content server 30 determines the arrangement order of the article content to be distributed and the article content to be distributed. Then, the content server 30 associates the determined article content to be distributed with the arrangement order and distributes it to the optimization server 20.

  The advertisement device is a server device that distributes advertisements. Upon receiving the content request from the optimization server 20, the advertising device 100 extracts the advertising content to be distributed from the advertising information storage unit 120 by performing the advertising extraction process illustrated in FIG. 1. Then, the advertising device 100 distributes the extracted advertising content to the optimization server 20.

[3. (Data flow)
Next, the advertisement extraction processing by the advertisement extraction system 1 according to the first embodiment will be described using FIG. 2 while clarifying the data flow. In addition, about the point which description overlaps with FIG. 1, description is abbreviate | omitted.

  First, it is assumed that the user terminal 10 transmits a request for the web page W1a, which is an initially displayed part of the web page W1, to the optimization server 20 in accordance with the user operation (step S11). ). When receiving the request for the web page W1a, the optimization server 20 transmits a content request for acquiring the article content included in the web page W1a to the content server 30 and acquires the advertisement content included in the web page W1a. An advertisement request is transmitted to the advertisement apparatus 100 (step S12).

  When the content server 30 receives the content request from the optimization server 20, for example, the content server 30 refers to a predetermined storage unit in its own device that stores various article content, and determines the distribution of the article content to be distributed and the article content to be distributed. The arrangement order is determined (step S13). In addition, the content server 30 generates a list 1 that lists the determined information. The list 1 shown in FIG. 2 indicates that the content server 30 has decided to arrange the article contents C1 to C6 to be distributed in a vertical row, such as first, second,... In addition, an example is shown in which the content server 30 associates text information such as a title and a description with each article content to be distributed.

  For example, in FIG. 2, the content server 30 associates the conceptual text information “T1” with the article content C1, but actually, as described in FIG. An explanatory sentence such as “Talk audience rating is 11.5%” is associated. The text information of the article contents C2 to C6 is also as illustrated in FIG. Further, as shown in the list 1, it is assumed that it is determined in advance that the advertising content is to be inserted into the arrangement position L1 (here, fourth from the top) which is an intermediate position.

  The content server 30 transmits the list 1 generated in this way to the advertising device 100 (step S14). Further, as shown in List 1, the content server 30 associates the data of the article content to be distributed with the arrangement order, and transmits them to the optimization server 20 (step S15).

  The advertising device 100 refers to the list 1 received from the content server 30, calculates an index value indicating the degree of association between the advertisement content as a distribution candidate and the article content group displayed around the arrangement position L1, and calculates the calculated index Based on the value, the advertisement content to be distributed is extracted (step S16). The advertisement extraction process is the same as described with reference to FIG.

  Then, the advertising device 100 transmits the extracted advertising content AD3 to the optimization server 20 (step S17). The optimization server 20 generates the web page W1a using the article content C1 to C6 received from the content server 30 and the advertisement content AD3 received from the advertising device 100 (step S18). Then, the optimization server 20 delivers the generated web page W1a to the user terminal 10 (step S19).

  Here, it is assumed that the web page W1a displayed on the display screen of the user terminal 10 is moved upward by a user operation, and the seventh article content C6 is included in the display screen. In such a case, the user terminal 10 transmits a request for acquiring the web page W1b to be displayed under the web page W1a to the optimization server 20 (step 21). It should be noted that it is possible to arbitrarily determine at what position the content is included in the display screen to transmit the request.

  When receiving the request for the web page W1b, the optimization server 20 transmits a content request for acquiring the article content included in the web page W1b to the content server 30, and acquires the advertisement content included in the web page W1b. An advertisement request is transmitted to the advertising device 100 (step S22).

  When the content server 30 receives the content request from the optimization server 20, for example, the content server 30 refers to a predetermined storage unit in its own device that stores various article content, and determines the distribution of the article content to be distributed and the article content to be distributed. The arrangement order is determined (step S23). In addition, the content server 30 generates a list 2 that lists the determined information. The list 2 shown in FIG. 2 indicates that the content server 30 has determined that the article contents C7 to C12 to be distributed are arranged in the vertical row in the eighth, ninth,... In addition, an example is shown in which the content server 30 associates text information such as a title and a description with each article content to be distributed. Further, as shown in the list 2, it is assumed that it is determined in advance that the advertising content is to be inserted into the arrangement position L2 (here, eleventh from the top) which is an intermediate position.

  The content server 30 transmits the list 2 generated in this way to the advertising device 100 (step S24). Further, as shown in list 2, the content server 30 associates the article content to be distributed with the arrangement order, and transmits them to the optimization server 20 (step S25).

  The advertising device 100 refers to the list 2 received from the content server 30, calculates an index value indicating the degree of association with the distribution candidate advertisement content and the article content group displayed around the arrangement position L 2, and calculates the calculated index value Based on the above, advertisement content to be distributed is extracted (step S26). Here, it is assumed that the advertising device 100 determines the advertising content AD1 as the advertising content to be distributed.

  The advertising device 100 transmits the extracted advertising content AD1 to the optimization server 20 (step S27). The optimization server 20 generates the web page W1b using the article content C7 to C12 received from the content server 30 and the advertisement content AD1 received from the advertising device 100 (step S28). Then, the advertising device 100 distributes the generated web page W1b to the user terminal 10 (step S29). The user terminal 10 displays, on the display screen, a web page W1 connected to the web page W1b below the web page W1a already displayed.

[4. Configuration of advertising device]
Next, the configuration of the advertising device 100 according to the first embodiment will be described with reference to FIG. FIG. 3 is a diagram illustrating a configuration example of the advertising device 100 according to the first embodiment. As illustrated in FIG. 3, the advertising device 100 includes a communication unit 110, an advertisement information storage unit 120, and a control unit 130.

  The communication unit 110 is realized by, for example, a NIC (Network Interface Card). The communication unit 21 is connected to the network by wire or wireless, and transmits / receives information to / from the optimization server 20 and the content server 30.

  The advertisement information storage unit 120 is realized by, for example, a semiconductor memory element such as a RAM (Random Access Memory) or a flash memory, or a storage device such as a hard disk or an optical disk. The advertisement information storage unit 120 stores various information related to the advertisement content received from the advertiser and distribution results that are result information when the advertisement content is distributed or clicked.

  Here, FIG. 4 is a diagram illustrating an example of the advertisement information storage unit 120 according to the first embodiment. In the example of FIG. 4, the advertisement information storage unit 120 includes items such as “advertiser ID”, “advertisement content”, “bid price”, “CTR”, and “eCPM”.

  “Advertiser ID” indicates identification information for identifying a terminal device (advertiser terminal) used by an advertiser or an advertiser for submission of advertising content. “Advertising content” indicates the advertising content submitted from the advertiser terminal 10. FIG. 4 shows an example in which conceptual information such as “AD1” is stored in “advertising content”, but actually, an image, a moving image, text data, a URL, or a file path name indicating the storage location thereof. Is stored.

  “Bid price” indicates an advertising fee designated when an advertiser submits advertising content. For example, the “bid price” corresponds to a unit price paid from the advertiser to the advertisement distributor (for example, the administrator of the advertising device 100) when the advertisement content is clicked once by the user.

  “CTR” indicates a value obtained by dividing the number of times the advertising content is clicked by the number of times the advertising content is displayed. The CTR of advertising content that has never been distributed to the terminal device 100 includes a predetermined fixed value, an average value of CTR for all advertising content, and the same advertising category (for example, car, travel). The average value of CTR and the like in all the advertising contents belonging to is stored. Further, “CTR” may store a predicted CTR predicted from a CTR prediction model or the like. Such a predicted CTR is predicted based on, for example, the type of advertising content, the type of web page on which the advertising content is displayed, and the like.

  “ECPM” (effective cost per mille) indicates an expected return amount of advertisement content around 1000 times, and can be calculated by, for example, multiplying CTR by a bid price.

  That is, FIG. 4 shows an example in which the advertiser identified by the advertiser ID “B1” submits the advertising content “AD1” and designates the bid price “150 yen”. FIG. 4 shows an example in which the CTR of the advertising content “AD1” is “0.02” and the eCPM is “3 yen”.

  Returning to FIG. 3, the control unit 130 executes various programs stored in the storage device inside the advertising device 100 using the RAM as a work area, for example, by a CPU (Central Processing Unit), an MPU (Micro Processing Unit), or the like. It is realized by doing. The control unit 130 is realized by an integrated circuit such as an application specific integrated circuit (ASIC) or a field programmable gate array (FPGA).

  As shown in FIG. 3, the control unit 130 includes a submission reception unit 131, a request reception unit 132, a calculation unit 133, an extraction unit 134, and a distribution unit 135. Implement or execute a function or action. The internal configuration of the control unit 130 is not limited to the information illustrated in FIG. 3, and may be another configuration as long as the information processing described later is performed. Further, the connection relationship between the processing units included in the control unit 130 is not limited to the connection relationship illustrated in FIG. 3, and may be another connection relationship.

  The submission receiving unit 131 receives submission of advertising content from the advertiser terminal. Specifically, the submission accepting unit 131 accepts advertisement content submission along with designation of a bid price. Then, the submission receiving unit 131 stores the received advertisement content and bid price in the advertisement information storage unit 120 in association with the advertiser ID for the advertiser of the submission source.

  The request receiving unit 132 receives an advertisement request from the optimization server 20. For example, the request reception unit 132 receives an HTTP request as an advertisement request. In addition, the request reception unit 132 receives information related to content to be distributed from the content server 30. For example, as shown in steps S14 and S24 of FIG. 2, the request reception unit 132 obtains the order of distribution of the article content to be distributed from the content server 30 and the text such as the title and description of the article content to be distributed. A list associated with each article content to be distributed is received.

  When the content and the advertisement content are displayed in a list as in the web page W1 shown in FIG. 1, the calculation unit 133 is displayed around the distribution candidate advertisement content and the advertisement content arrangement position. An index value indicating the degree of association with the content to be calculated is calculated. For example, the calculation unit 133 calculates the degree of similarity with the distribution candidate advertisement content for each content displayed around the arrangement position. Then, the calculation unit 133 calculates an index value with the content group displayed around the arrangement position by summing up the calculated similarities. Hereinafter, the index value calculation processing according to the first embodiment will be described with reference to FIG. FIG. 5 is a diagram illustrating an example of an index value calculation process according to the first embodiment. In the present embodiment, the arrangement position (display position) of the advertisement content and the article content indicates the arrangement order from the top when the advertisement content and the article content are arranged in a vertical line.

  The web page W1x shown in FIG. 5 conceptually shows a part of the web page W1 that is the list information shown in FIG. Here, since the advertisement content arranged at the arrangement position Lx, which is the arrangement position of the advertisement content, that is, the advertisement content to be distributed is extracted by the extraction unit 134 described later, the calculation unit 133 determines that the distribution candidate advertisement content and An index value calculation process for calculating an index value indicating a degree of association with article content will be described. In the index value calculation process, the following two patterns can be considered.

  First, the pattern 1 will be described. As shown in FIG. 5, the arrangement position Lx of the advertisement content, that is, the arrangement order is i-th. With this i-th as a reference, the arrangement positions of article content located above the arrangement position Lx can be determined as i−1, i-2, i-3,. . Further, with the i-th as a reference, the arrangement positions of article content located below the arrangement position Lx can be determined as i + 1, i + 2, i + 3,.

  Here, the calculation unit 133 determines the article content from the arrangement position Lx to the upper and lower two pieces as the article content around the arrangement position Lx, and calculates the similarity to the advertisement content I as a distribution candidate. In other words, in the example of FIG. 5, the calculation unit 133 performs the distribution candidate advertisement content I, the (i-2) th article content (article content B), the (i−1) th article content (article content C), and the (i + 1) th article. The similarity to each of the article content (article content D) and the i + 2nd article content (article content E) is calculated. Then, the calculation unit 133 calculates the index values of the advertising content I and the article content group from the arrangement position Lx to the upper and lower two pieces by summing the calculated similarities.

  The distribution candidate advertisement content I may indicate each advertisement content stored in the advertisement information storage unit 120 shown in FIG. 4, or may be stored in the advertisement information storage unit 120 shown in FIG. Each advertising content selected based on eCPM from each advertising content may be indicated.

  The calculation unit 133 calculates the similarity based on the feature information indicating the features of the advertisement content and the article content. For example, the calculation unit 133 extracts words (feature words) by performing morphological analysis on text included in the advertising content I, and generates a word vector indicating the characteristics of the advertising content. In addition, the calculation unit 133 extracts words (feature words) by performing morphological analysis on the text included in each article content from the arrangement position Lx to the upper and lower two pieces, and generates a word vector indicating the feature of each article content To do. Then, the calculation unit 133 calculates the similarity between the generated word vectors (for example, the cosine similarity between the word vectors) for the advertising content I and each article content. For example, the calculation unit 133 calculates a predetermined numerical value indicated as a cosine similarity in a range of “0 to 1”.

FIG. 5 illustrates an example in which the calculation unit 133 calculates the cosine similarity cos _i-2 as the similarity between the advertisement content I and the i-2th article content B. Further, an example is shown in which the calculation unit 133 calculates the cosine similarity cos _i-1 as the similarity between the advertisement content I and the (i-1) th article content C. The same applies to the (i + 1) th article content D and the (i + 2) th article content E.

  In addition, the calculation unit 133 adds a predetermined weight value to each calculated similarity. For example, in pattern 1, the calculation unit 133 multiplies the higher the similarity with the article content closer to the arrangement position Lx among the article contents from the arrangement position Lx to the upper and lower two pieces. In the present embodiment, the weight value r which is the highest weight value is set to “0.9”.

In the example of FIG. 5, since the (i-1) th article content C and the (i + 1) th article content D are adjacent to the arrangement position Lx, these article contents are located closest to the arrangement position Lx. Therefore, calculation unit 133, the similarity _{cos i + 1} of the similarity _{cos i-1} and ad content I and the (i + 1) th article content D of the advertisement content I and i-1 th article content C, the weight value r Multiply by (0.9).

On the other hand, the i-2th article content B and the i + 2nd article content E are separated from the arrangement position Lx by one content. Therefore, calculation unit 133, the similarity _{cos i + 2} the similarity _{cos i-2} and ad content I and i + 2-th article content E of the advertising content I and i-2-th article content B, the weight value r Multiply by a smaller weight value r ² (0.81).

  For example, it is considered that the user's interest in the arranged advertisement content can be increased by arranging the advertisement content more similar to the latest article content at the arrangement position Lx at the arrangement position Lx. This is because whether or not the user is interested in the advertising content tends to be easily influenced by the content of the latest article content. For example, when the article contents C and D are articles related to sports, by placing sports-related advertisement contents at the arrangement position Lx, the article contents C and D increase the user's interest in sports. It becomes easy to click with interest in the advertising content.

  In consideration of such user characteristics, the calculation unit 133, as described above, multiplies the similarity with the article content close to the arrangement position Lx by a higher weight value, thereby resembling the advertisement similar to the article content close to the arrangement position Lx. Content can be easily extracted.

The calculation unit 133, by summing the similarity multiplied by the weight value, calculating advertising content I, the index value S _I indicating the relevance of the article content group to be displayed in the peripheral position Lx . Specifically, in the index value calculation process in the Pattern 1 described so far, the calculation unit 133 can calculate the index value S _I by the following equation (1).

Index value S _I = r ² cos _i−2 + rcos _i−1 + rcos _{i + 1} + r ² cos _{i + 2} (1)

That is, the calculating unit 133, for each advertising content of each distribution candidate, using equation (1), calculates an index value S _I indicating the relevance of the article content group from position Lx up to two minutes up and down .

Then, calculating unit 133, based on the index value S _I calculated using Equation (1), calculates an advertisement evaluation value. For example, the calculation unit 133 calculates the total value of eCPM _I , which is the evaluation value of the advertising content I, and the index value S _I calculated using Expression (1) as the advertisement evaluation value E _I. Specifically, the calculation unit 133 calculates the advertisement evaluation value E _{I according} to the following formula (2).

Advertising evaluation value E _I = eCPM _I + index value S _I (2)

That is, the calculation unit 133 calculates the advertisement evaluation value E _I for each of the distribution candidate advertisement contents using Expression (2).

  Next, the pattern 2 will be described. Pattern 2 is different from Pattern 1 in the way of assigning weight values. Other points are the same as those in the pattern 1. Specifically, in the pattern 1, the example in which the calculation unit 133 multiplies the weight value higher as the similarity to the article content near the arrangement position Lx is shown. However, the calculation unit 133 multiplies the similarity with the article content positioned above the arrangement position Lx by a higher weight value when multiplying the similarity with the article content closer to the arrangement position Lx by a higher weight value.

For example, as illustrated in the pattern 2 illustrated in FIG. 5, the calculation unit 133 calculates the relationship between the i−1th article content C and the advertisement content I that are closest to the placement position Lx and located above the placement position Lx. Multiply the similarity cos _i-1 by the highest weight value r (0.9). Further, the calculation unit 133 sets the similarity cos _i−2 between the _i− 2th article content B next to the placement position Lx and in the upward direction of the placement position Lx to the advertisement content I by 2 Multiply by the second highest weight value r ² (0.81).

In addition, the calculation unit 133 has the third highest weight value r for the similarity cos _{i + 1} between the i + 1th article content D that is closest to the placement position Lx and below the placement position Lx and the advertisement content I. Multiply by ³ (0.73). Furthermore, the calculation unit 133 next assigns the fourth highest weight to the degree of similarity cos _{i + 2} between the _{i +} 2th article content E, which is next to the placement position Lx and below the placement position Lx, and the advertisement content I. Multiply by the value r ⁴ (0.66).

  So far, it has been described that by placing the advertisement content more similar to the placement position Lx and the latest article content at the placement position Lx, the user's interest in the placed advertisement content can be enhanced. However, when browsing the list information in which a plurality of contents are arranged in a vertical row as in the web page W1 according to the first embodiment, the user scrolls the web page W1 upward, so that each content is scrolled upward. It is thought that there are many browsing.

  In such a case, whether or not the user is interested in the advertising content is influenced by the content of the latest article content, and the web page W1 is scrolled upward, so that the user can quickly notice the user. There seems to be a tendency to be easily influenced by the content of article content. For example, when the article content C is a sports-related article and the article content D is a fashion-related article, the sports-related advertising content is arranged at the arrangement position Lx rather than the fashion-related advertising content. The sports-related article content C, which is noticeable before the fashion-related article content D, increases the interest of the user in sports and makes it easy to click on sports-related advertising content.

  In consideration of such user characteristics, the calculation unit 133 multiplies the placement position Lx by a weight value that is higher as the degree of similarity to the article content that is close to the placement position Lx and located in the upper direction of the placement position Lx. It is possible to easily extract advertisement content similar to article content located near and above the arrangement position Lx.

The calculation unit 133, by summing the similarity multiplied by the weight value, and advertising content I, the index value S _I from the position Lx indicates the relevance of the articles content group up to 2 pieces of the upper and lower calculate. Specifically, in the index value calculation process in the Pattern 2 described so far, the calculation unit 133 can calculate the index value S _I by the following equation (3).

Index value S _I = r ² cos _i−2 + rcos _i−1 + r ³ cos _{i + 1} + r ⁴ cos _{i + 2} (3)

That is, the calculating unit 133, for each advertising content of each distribution candidate, using equation (3), calculates an index value S _I indicating the relevance of the article content group from position Lx up to two minutes up and down .

Then, calculating unit 133, based on the index value S _I calculated using Equation (3) to calculate the advertisement evaluation value. For example, the calculation unit 133 calculates the total value of the eCPM _{I of} the advertising content I and the index value S _I calculated for the advertising content I as the advertising evaluation value E _I using Expression (3). That is, the calculation unit 133 calculates the advertisement evaluation value E _{I according} to the above equation (2).

  Returning to FIG. 3, the extraction unit 134 extracts the advertisement content to be distributed from the distribution candidate advertisement content based on the index value calculated by the calculation unit 133. For example, the extraction unit 134 causes the calculation unit 133 to extract, as a distribution target, advertisement content of a distribution candidate having the highest advertisement evaluation value calculated using the above formula (2).

  The distribution unit 135 distributes advertisement content to be distributed. Specifically, the distribution unit 135 distributes the advertising content to be distributed extracted by the extraction unit 134 to the optimization server 20.

[5. Specific example of advertisement extraction processing)
Now, the advertisement extraction process that has been conceptually described with reference to FIG. 5 will be described using a specific example. FIG. 6 is an explanatory diagram for specifically explaining the advertisement extraction processing according to the first embodiment.

  First, it is assumed that the user terminal 10 transmits to the optimization server 20 a request for the web page W1a, which is an initially displayed portion of the web page W1, in accordance with a user operation. As a result, as shown in FIG. 6, the content server 30 determines the article content to be distributed as the article contents C1 to C6, so that the list 1 in which the arrangement order is associated with the text information T1 to T6 is displayed. Suppose that it transmitted to the advertising device 100. Further, when receiving the list 1, the advertising device 100 distributes the top three advertising contents AD1 to AD3 having a higher eCPM from the advertising contents stored in the advertising information storage unit 120 shown in FIG. Suppose that

  In the web page W1a, as described above, it is assumed that it is determined in advance to insert the advertisement content at the arrangement position L1 (fourth from the top) which is an intermediate position.

  Here, the calculation unit 133 of the advertising device 100 calculates the similarity between each of the article contents C2 to C5, which are article contents displayed around the arrangement position L1, with each of the distribution candidate advertisement contents AD1 to AD3.

For example, when the pattern 1 shown in FIG. 5 is used, the advertisement content AD3 as a distribution candidate will be described as an example. For example, the calculation unit 133 sets “0” as the similarity cos _i−2 between the advertisement content AD3 and the article content C2. .3 ”,“ 0.8 ”as the similarity cos _i−1 between the advertisement content AD3 and the article content C3,“ 0.8 ”as the similarity cos _{i + 1} between the advertisement content AD3 and the article content C4, “0.1” is calculated as the similarity cos _{i + 2} with the article content C5.

As shown in FIG. 6, the advertisement content AD3 is a sports-related advertisement, and the article content C2 is a fashion-related article. Since the similarity between sports and fashion is low, the calculation unit 133 displays the feature word extracted from the text “Large sports shop SS Japan first landing!” Of the advertisement content AD3 and the text T2 “article style of the article content C2”. For example, a lower “0.3” is calculated as the cosine similarity cos _i−2 with the feature word extracted from “Self-portrait technique”.

On the other hand, since the article contents C3 and C4 are sports-related articles and have high similarity to the advertisement content AD3, the calculation unit 133 and the feature word extracted from the text of the advertisement content AD3 and the text T3 “pro” of the article content C3 For example, a higher “0.8” is calculated as the cosine similarity cos _i−1 with the feature word extracted from “baseball news highlight”. In addition, the calculation unit 133 similarly uses the cosine similarity cos _{i + 1} between the feature word extracted from the text of the advertisement content AD3 and the feature word extracted from the text T4 “What is the problem that emerged in Japan Women ’s Valley?” Then, a higher “0.8” is calculated.

Further, since the article content C5 is a political-related article and has a low similarity to the advertisement content AD3, the calculation unit 133 calculates “0.1”, which is lower than the article content C2, for example, as the similarity cos _{i + 2.} .

  Then, the calculation unit 133 multiplies each calculated similarity by a weight value. For example, according to the pattern 1 shown in FIG. 5, the calculation unit 133 multiplies a higher weight value as the degree of similarity with the article content located near the arrangement position L1.

And the calculation part 133 applies each similarity degree multiplied by the weight value to said Formula (1), and is the advertisement content AD3 of a delivery candidate, and the article content C2-C5 displayed around arrangement | positioning position L1. An index value S _AD3 is calculated. Specifically, the calculation unit 133 uses the above formula (1), and the index value S _AD3 = (0.81 × 0.3) + (0.9 × 0.8) + (0.9 × 0) .8) + (0.81 × 0.1) = 1.664 is calculated.

Further, the calculation unit 133 adds the eCPM _AD3 “2.2” of the advertisement content AD3 and the index value S _AD3 “1.764”, as shown in the above formula (2), so that the advertisement of the advertisement content AD3 is displayed. “3.9774” is calculated as the evaluation value E _AD3 .

Up to this point, using FIG. 5 and FIG. 6, the calculation unit 133 calculates the index value S _AD3 and the advertisement evaluation value E _AD3 of the advertisement content AD3 and the article contents C2 to C5 displayed around the placement position L1. An example to do. Calculator 133, the advertisement content AD1 delivery candidates, similarly, for example, to calculate the "3.863" as an index value _{S AD1} as "0.863" and advertisement evaluation value _{E AD1.} Similarly, the calculation unit 133 also calculates “1.278” as the index value S _AD2 and “2.328” as the advertisement evaluation value E _AD2 for the advertisement content AD2 as the distribution candidate.

  Next, the extraction unit 134 extracts the advertisement content AD3 for which the highest advertisement evaluation value is calculated among the index values calculated by the calculation unit 133 as a distribution target. Then, since the distribution unit 135 distributes the advertising content AD3 to the optimization server 20, the optimization server 20 distributes the web page W1a in which the advertising content AD3 is arranged at the arrangement position L1.

  For example, in the case of general advertisement distribution, it is considered that the advertisement content with the highest eCPM is simply extracted as the distribution target. In this case, in the example of FIG. 6, the advertising content AD1 is a distribution target. However, as in the present embodiment, when the advertisement content to be distributed is further extracted based on the index values of the advertisement content to be distributed and the article content around the arrangement position, the distribution is only when considering only eCPM. The target may be different.

  Since the advertisement content to be distributed extracted in this way has a relatively high eCPM and a high affinity with the article content around the position, it has a higher advertising effect than the advertisement content extracted based on the eCPM. Can be generated.

[6. Processing procedure)
Next, the procedure of the advertisement extraction process executed by the advertising device 100 according to the first embodiment will be described with reference to FIG. FIG. 7 is a flowchart illustrating an advertisement extraction processing procedure performed by the advertisement apparatus 100 according to the first embodiment.

  First, the request receiving unit 132 of the advertising device 100 determines whether an advertising request has been received from the optimization server 20 (Step S101). When the request reception unit 132 receives an advertisement request (step S101; Yes), the request reception unit 132 determines whether a list related to article content is received from the content server 30 (step S102). If the request reception unit 132 has received a list (step S102; Yes), the request reception unit 132 transmits the received list to the calculation unit 133.

  Here, the list relating to the article content is a list in which the order of distribution of the article content to be distributed and the text information are associated with each advertisement content to be distributed as described above. In addition, the request receiving unit 132 may perform steps S101 and S102 at the same time, or may be configured to receive an advertisement request including a list from the content server 30, for example.

  Next, the calculation unit 133 refers to the received list and identifies article content displayed around the arrangement position Lx where the advertising content is arranged (step S103). For example, the calculation unit 133 identifies article content displayed up to two from the top and bottom of the placement position Lx as article content displayed around the placement position Lx.

  In addition, the calculation unit 133 determines the distribution candidate advertisement content (step S104). For example, the calculation unit 133 refers to the advertisement information storage unit 120 and determines the upper predetermined number of advertisement contents having a high eCPM as the advertisement content of the distribution candidate.

  Then, the calculation unit 133 selects one of the distribution candidate advertisement contents (step S105). Here, the selected advertisement content is referred to as advertisement content I. Then, the calculation unit 133 calculates the similarity with the advertisement content I for each article content displayed around the arrangement position Lx (step S106). Furthermore, the calculation unit 133 weights each calculated similarity (step S107). For the weighting, either the pattern 1 or 2 described with reference to FIG. 5 may be used.

The calculation unit 133, by summing the similarity is weighted to calculate the advertising content I, the index value S _I of the article content group to be displayed in the peripheral position Lx (step S108). For example, in step S107, when weighted using the pattern 1, the calculation unit 133 calculates an index value _{S I} using equation (1). Furthermore, when weighted with the pattern 2, the calculation unit 133 calculates an index value S _I using equation (3).

Further, the calculation unit 133 calculates the advertisement evaluation value E _I by summing up the eCPM of the advertisement content I and the index value S _I (step S109). After that, the calculation unit 133 determines whether or not the advertisement evaluation value E _I has been calculated for all of the distribution candidate advertisement contents (step S110). If the calculation unit 133 determines that the advertisement evaluation value E _I has been calculated for all of the distribution candidate advertisement contents (step S110; Yes), the calculation unit 133 transmits the calculation result to the extraction unit 134.

Next, when receiving the calculation result, the extraction unit 134 compares the advertisement evaluation values E _I of the respective distribution candidate advertisement contents, and extracts the advertisement content having the highest advertisement evaluation value E _I as a distribution target (step S111). ). The distribution unit 135 distributes the extracted advertisement content to the optimization server 20 (step 112).

[7. (Modification)
The first embodiment described above may be implemented in various different forms other than the above embodiment. Accordingly, other embodiments will be described below.

[7-1. (Category classification)
In the first embodiment, as an example in which the calculation unit 133 of the advertising device 100 calculates the similarity to the advertising content for each article content, the feature word extracted from the text direction of the advertising content and the text of the article content are extracted. An example is shown in which the cosine similarity with the feature word is calculated. However, the calculation unit 133 may calculate the similarity (category similarity) between the category to which the distribution candidate advertisement content belongs and the category to which the article content displayed around the arrangement position belongs. This point will be described with reference to FIG. Also refer to FIGS. 5 and 6 as appropriate.

  FIG. 8 is a diagram illustrating an example of category classification according to the modification example of the first embodiment. As shown in FIG. 8, the category classification is divided into a plurality of hierarchies in a tree format. The first hierarchy is a large classification and corresponds to the widest category classification. The second hierarchy is a middle classification and corresponds to a category classification obtained by further subdividing each category of the first hierarchy. The third layer is a small classification and corresponds to a category classification obtained by subdividing each category of the second hierarchy.

  Here, in the above, the calculation unit 133 calculates the index value by summing each cosine similarity weighted with the pattern 1 as shown in the equation (1), or as shown in the equation (3). An example is shown in which the index value is calculated by summing up each cosine similarity weighted by the pattern 2. Even when the category similarity is used, the calculation unit 133 calculates the index value by summing each category similarity weighted by the pattern 1 or sums each category similarity weighted by the pattern 2. Thus, the index value can be calculated.

Specifically, when each category similarity is weighted with pattern 1, the calculation unit 133 uses the following expression (4) to calculate the advertisement content I to be distributed and the article content group displayed around the arrangement position Lx. it is possible to calculate the index value S _I of.

Index value S _I = r ² CG _i−2 + rCG _i−1 + rCG _{i + 1} + r ² CG _{i + 2} (4)

Referring to FIG. 5, “CG _i-2 ” indicates the category similarity between the advertisement content I and the i-2th article content B. “CG _i−1 ” indicates the category similarity between the advertisement content I and the (i−1) th article content C. “CG _{i + 1} ” indicates the category similarity between the advertisement content I and the i + 1th article content D. “CG _{i + 2} ” indicates the category similarity between the advertisement content I and the i + 2nd article content E.

  Below, the specific example which calculates a similarity using the category tree shown in FIG. 8 is shown. For example, as illustrated in FIG. 6, the advertising device 100 receives the list 1 in which the order of distribution of the article content to be distributed and the text information are associated with each other, and determines the advertising content AD1 to AD3 as the distribution candidate advertising content. And

  Here, the calculation unit 133 of the advertising device 100 calculates the category similarity with each of the distribution candidate advertisement contents AD1 to AD3 for each of the article contents C2 to C5 that are article contents displayed around the arrangement position L1.

For example, the advertisement content AD3 as a distribution candidate will be described as an example. When calculating the category similarity CG _i-2 between the advertisement content AD3 and the article content C2, as shown in FIG. 6, the fourth (i-th) advertisement is calculated. The content AD3 includes a text such as “Large sports shop SS first landing in Japan!”. Therefore, for example, the calculation unit 133 specifies the category of the advertisement content AD3 as “sports”. Further, the second (i-2) article content C2 includes text such as “Self-portrait tech that shows style well”. Therefore, for example, the calculation unit 133 specifies the category of the article content C2 as “fashion”.

  Here, according to the category tree shown in FIG. 8, “sports” and “fashion” are categories that are broadly classified, and thus it can be said that both are not very similar. When calculating the similarity between the categories of such large classifications, for example, the calculation unit 133 subtracts “0.5” from the maximum value “1” of the similarity to calculate “sport” and “fashion”. The similarity, that is, the similarity “0.5” between the advertisement content AD3 and the article content C2 is calculated.

When calculating the category similarity CG _i-1 between the advertisement content AD3 and the article content C3, as shown in FIG. 6, the third (i-1) article content C3 is “Pro Baseball News Highlight”. Is included. For this reason, for example, the calculation unit 133 specifies the category of the article content C3 as “professional baseball”.

  According to FIG. 8, “professional baseball” is a category that falls under “sports”, and therefore it can be said that the two are relatively high in similarity. “Sports” and “professional baseball” are separated by two levels. In such a case, for example, the calculation unit 133 adjusts the maximum similarity value “1” using “0.1” in order to trace one hierarchy, and the maximum similarity value “1” is changed from “1” to “ Subtract “0.1 × 2 (for two layers)”. Then, the calculation unit 133 calculates the similarity between “sports” and “professional baseball”, that is, the similarity “0.8” between the advertisement content AD3 and the article content C3.

Further, when calculating the category similarity CG _{i + 1} between the advertisement content AD3 and the article content C4, as shown in FIG. 6, the fifth (i + 1) article content C4 is “What is the problem that emerged from the Japan Women's Valley?” Is included. For this reason, for example, the calculation unit 133 specifies the category of the article content C4 as “valley”.

  According to FIG. 8, “valley” is a category that falls under “sports”, and therefore it can be said that both have a relatively high degree of similarity. “Sports” and “valley” are separated by one layer. In such a case, for example, the calculation unit 133 subtracts “0.1 (for one layer)” from the maximum value “1” of the similarity. Then, the calculating unit 133 calculates the similarity between “sports” and “valley”, that is, the similarity “0.9” between the advertisement content AD3 and the article content C4.

Further, when the category similarity CG _{i + 2} between the advertisement content AD3 and the article content C5 is calculated, as shown in FIG. 6, the sixth (i + 2) article content C5 is “the joint exercise between the US and Korea is finished”. Contains text. For this reason, for example, the calculation unit 133 specifies the category of the article content C5 as “politics”.

  According to FIG. 8, it can be said that “sports” and “politics” are not similar to each other because they are categories roughly classified. For this reason, for example, the calculation unit 133 subtracts “0.5” from the maximum value “1” of the similarity to obtain the similarity between “sports” and “politics”, that is, the advertising content AD3 and the article content C2. The similarity “0.5” is calculated.

  Then, the calculation unit 133 multiplies each calculated category similarity by a weight value. For example, according to the pattern 1 shown in FIG. 5, the calculation unit 133 multiplies a higher weight value as the degree of similarity with the article content located near the arrangement position L1.

And the calculation part 133 applies each similarity degree multiplied by the weight value to said Formula (4), and is the advertisement content AD3 of a delivery candidate, and the article content C2-C5 displayed around the position L1. An index value S _AD3 is calculated. Specifically, the calculation unit 133 uses the above formula (4), and the index value S _AD3 = (0.81 × 0.5) + (0.9 × 0.8) + (0.9 × 0) .9) + (0.81 × 0.5) = 2.34 is calculated.

Further, the calculation unit 133 adds the eCPM _AD3 “2.2” of the advertisement content AD3 and the index value S _AD3 “2.34”, as shown in the above formula (2), so that the advertisement of the advertisement content AD3 is displayed. “4.54” is calculated as the evaluation value E _AD3 .

So far, the calculation unit 133 has shown an example in which the index value S _AD3 and the advertisement evaluation value E _AD3 of the advertisement content AD3 and the article contents C2 to C5 displayed around the arrangement position L1 are calculated. Similarly, the calculation unit 133 calculates the index value S _AD1 and the advertisement evaluation value E _AD1 from the category similarity with each of the article contents C2 to C5 for the advertisement content AD1 to be distributed. Similarly, the calculation unit 133 calculates the index value S _AD2 and the advertisement evaluation value E _AD2 from the category similarity with each of the article contents C2 to C5 for the advertisement content AD2 to be distributed.

In addition, when each category similarity is weighted with the pattern 2 illustrated in FIG. 5, the calculation unit 133 uses the following formula (5) to calculate the advertisement content I to be distributed and the article content displayed around the arrangement position Lx. it is possible to calculate the index value S _I with the group.

Index value S _I = r ² CG _i−2 + rCG _i−1 + r ³ CG _{i + 1} + r ⁴ CG _{i + 2} (5)

  As described above, the advertising device 100 according to the modified example of the first embodiment calculates the category similarity with the distribution candidate advertisement content for each article content displayed around the arrangement position, and calculates the calculated similarity. To calculate the index value. Thus, the advertising device 100 can perform category specification and category similarity calculation between the advertising content and the article content by using the category tree as shown in FIG. There is no need to perform this, and the processing until the advertisement evaluation value is calculated can be accelerated compared to the case where the cosine similarity calculation shown in FIG. 5 is performed.

[8. effect〕
The advertising device 100 according to the first embodiment includes a calculation unit 133 and an extraction unit 134. The calculation unit 133 calculates an index value indicating the degree of association between the distribution candidate advertisement content displayed together with the content and the content displayed around the arrangement position where the distribution candidate advertisement content is arranged. Based on the index value calculated by the calculation unit 133, the extraction unit 134 extracts the advertisement content to be distributed from the distribution candidate advertisement content.

  Thereby, since the advertising device 100 according to the first embodiment can display the advertising content having higher affinity with the content displayed around the arrangement position on the user terminal 10, the advertising effect can be enhanced. .

  In addition, the calculation unit 133 calculates the similarity with the advertisement content as the distribution candidate for each content displayed around the arrangement position, and uses the calculated similarity to display the content group displayed around the arrangement position. An index value indicating the degree of relevance with the distribution candidate advertisement content is calculated.

  As described above, the advertising device 100 according to the first embodiment calculates the degree of similarity with the distribution candidate advertisement content for each content displayed around the arrangement position, thereby displaying the content displayed around the arrangement position. The index value indicating the degree of association between the group and the advertisement content of the distribution candidate can be calculated with high accuracy.

  In addition, the calculation unit 133 assigns a higher weight to the similarity with the content close to the arrangement position among the contents displayed around the arrangement position.

  Accordingly, the advertising device 100 according to the first embodiment calculates the index value in consideration of user characteristics such that whether the user is interested in the advertising content is easily influenced by the content closer to the arrangement position. Therefore, it is possible to display on the user terminal 10 advertisement content having a high affinity with content closer to the arrangement position. And thereby, the advertising device 100 can enhance the effect.

  In addition, the calculation unit assigns higher weight to the similarity between the content displayed in the vicinity of the arrangement position and the content positioned above the arrangement position.

  As a result, the advertising device 100 according to the first embodiment determines whether or not the user is interested in the advertising content because the user is easily influenced by the content located closer to the arrangement position and above the arrangement position. Since the index value can be calculated in consideration of the user characteristics, it is possible to display on the user terminal 10 advertising content that is closer to the arrangement position and that has a high affinity with the content located above the arrangement position. And thereby, the advertising device 100 can enhance the effect.

  In addition, the calculation unit 133 calculates the degree of similarity between information extracted from the text included in the distribution candidate advertisement content and information extracted from the text included in the content displayed around the arrangement position.

  As described above, since the advertising device 100 according to the first embodiment calculates the similarity using the information extracted from the text, it can calculate the similarity with high accuracy.

  Further, the calculation unit 133 calculates the similarity between the category to which the distribution candidate advertisement content belongs and the category to which the content displayed around the arrangement position belongs.

  As described above, the advertising device 100 according to the first embodiment calculates the degree of similarity by comparing the category to which the advertising content belongs and the category to which the content displayed around the arrangement position belongs. The processing speed can be increased as compared with the case where the similarity is calculated using the obtained information.

  Further, the calculation unit 133 calculates an index value indicating the degree of association by summing up the similarities, and the extraction unit 134 calculates the sum of the index value calculated by the calculation unit 133 and the evaluation value of the distribution candidate content. Based on the above, the advertisement content to be distributed is extracted from the advertisement content as the distribution candidate.

  Thereby, advertising device 100 concerning a 1st embodiment can raise advertising effectiveness.

(Second Embodiment)
[1. Configuration of advertising device]
In the first embodiment, the calculation unit 133 of the advertising device 100 calculates the similarity with the distribution candidate advertisement content for each article content around the arrangement position, and uses the calculated similarity to calculate the index value. An example of calculation has been described. However, in the list information in which the article content and the advertisement content are displayed as a list like the web page W1, the article content and the advertisement content that the user is interested in are predicted, and the user is expected to be interested. You may perform the process of making it easy to extract advertisement content with high affinity with a content.

  And below, the advertising device 200 which added the structure which performs such a prediction process with respect to the advertising device 100 mentioned above is demonstrated. FIG. 9 is a diagram illustrating a configuration example of the advertising device 200 according to the second embodiment. As illustrated in FIG. 9, the advertising device 200 includes a communication unit 110, an advertisement information storage unit 120, and a control unit 230.

  As illustrated in FIG. 9, the control unit 230 includes a submission receiving unit 131, a request receiving unit 132, a calculation unit 133, an extraction unit 134, a distribution unit 135, and a prediction unit 236. That is, the advertising device 200 corresponds to a device in which the prediction unit 236 is further added to the advertising device 100.

  The prediction unit 236 predicts article content and advertisement content that the user is interested in among the contents included in the web page W1. For example, the prediction unit 236 predicts article content and advertisement content in which the user is interested by analyzing a user operation log transmitted from the user terminal 10.

  For example, the prediction unit 236 calculates a moving speed at which the web page W1 is moved by a user operation on the web page W1. Then, the prediction unit 236 specifies content that is located near the center of the display screen of the user terminal 10 and predicts that the user is interested in the specified content when the moving speed is equal to or lower than the predetermined speed. To do.

  For example, when a user browses each content while scrolling the web page W1, if the content that attracts interest is noticeable, the user unconsciously decreases the scroll speed or stops the scroll operation. It is believed that there is. Based on such characteristics, the prediction unit 236 predicts that the user is interested in the content located near the center of the display screen of the user terminal 10 when the moving speed is equal to or lower than the predetermined speed. . Note that the prediction unit 236 does not necessarily have to predict that the user is interested in the content located near the center of the display screen, and may be any position on the display screen.

  Further, for example, when the user browses each content while scrolling the web page W1 and there is content that attracts interest, the user selects the content and makes a transition to the detail page. For this reason, the prediction unit 236 may predict the content selected by the user among the contents included in the web page W1 as the content that the user is interested in.

Then, the calculation unit 133 weights the similarity with the article content predicted to be interested in the user by the prediction unit 236, and assigns the weighted similarity to the advertisement content I to be distributed and the arrangement position. The index value S _i is calculated by adding the similarity to each article content displayed in the vicinity and weighting each similarity as shown in the pattern 1 or 2.

For example, when the cosine similarity between the article content Y and the advertising content I predicted to be interested in the user by the prediction unit 236 is cos _Y , the calculation unit 133 sets the cosine similarity to cos _Y most. Multiply by a high weight value r (0.9). Then, calculating unit 133, the case of using the pattern 1, the following equation (6), calculates an index value _{S I.}

Index value S _I = (r ² cos _i−2 + rcos _i−1 + rcos _{i + 1} + r ² cos _{i + 2} ) + rcos _Y (6)

The expression in parentheses corresponds to the above expression (1). In other words, the calculation unit 133 calculates rcos _Y obtained by multiplying the cosine similarity cos _Y between the article content Y and the advertising content I predicted to be interested in the user by the prediction unit 236 by the highest weight value r. calculating an index value S _I using equation (6), further added to the above formula (1). From this, the calculation part 133 makes it easy to extract the advertising content having high affinity with the article content Y as a distribution target by increasing the influence by the cosine similarity cos _Y.

Further, when the pattern 2 is used, the calculation unit 133 calculates the index value S _i by the following equation (7).

Index value S _I = (r ² cos _i−2 + rcos _i−1 + r ³ cos _{i + 1} + r ⁴ cos _{i + 2} ) + rcos _Y (7)

The expression in parentheses corresponds to the above expression (3). In addition, as shown in Expression (6) and Expression (7), the index value S _i is calculated using the cosine similarity, but the category similarity may be used.

[2. (Data flow)
Next, the advertisement extraction process by the advertisement extraction system 2 according to the second embodiment will be described using FIG. 10 while clarifying the data flow. FIG. 10 is a diagram illustrating an example of advertisement extraction processing by the advertisement extraction system 2 according to the second embodiment.

  First, it is assumed that a web page W1 including web pages W1a and W1b is displayed on the user terminal 10. In such a state, the user terminal 10 acquires, as an operation log, a user operation on the web page W1 and a display mode of the web page W1 that is changed according to the user operation (step S31).

  Here, it is assumed that the user terminal 10 transmits a request for the web page W1c to be connected under the web page W1b and the operation log acquired in step S31 to the optimization server 20 according to the user operation (step S32). When receiving the request for the web page W1c, the optimization server 20 transmits a content request for acquiring the article content included in the web page W1c to the content server 30 and acquires the advertisement content included in the web page W1a. The advertisement request and the operation log are transmitted to the advertisement apparatus 200 (step S33).

  Upon receiving the content request from the optimization server 20, the content server 30 determines the arrangement order of the distribution target article content and the distribution target article content (step S34). In addition, the content server 30 generates a list 3 that lists the determined information. The list 3 shown in FIG. 10 indicates that the content server 30 has decided to arrange the distribution of article contents C13 to C18 to be distributed in a vertical row such as the 15th, 16th,... 21st from the top. In addition, an example is shown in which the content server 30 associates text information such as a title and a description with each article content to be distributed.

  The content server 30 transmits the list 3 generated in this way to the advertising device 200 (step S35). Further, as shown in List 3, the content server 30 associates the data of the article content to be distributed with the arrangement order, and transmits it to the optimization server 20 (step S36).

  On the other hand, when the request reception unit 132 receives the advertisement request and the operation log from the optimization server 20, the prediction unit 236 of the advertising device 200 refers to the operation log and predicts the content that the user is interested in. (Step S34). For example, the predicting unit 236 calculates the moving speed at which the web page W1 is moved, and when the moving speed is equal to or lower than a predetermined speed, the user selects the content located near the center of the display screen of the user terminal 10. Predict the content you are interested in. For example, the prediction unit 236 may predict the content selected by the user as content that the user is interested in. Here, it is assumed that the prediction unit 236 predicts that the user is interested in the article content C11.

  In addition, the calculation unit 133 of the advertising device 200 refers to the list 3 received from the optimization server 20 by the request reception unit 132, has the distribution candidate advertisement content, the article content displayed around the arrangement position L3, and the interest. An index value indicating the degree of association with the article content C11 that is predicted to be calculated is calculated. Further, the calculation unit 133 calculates an advertisement evaluation value that is the sum of the index value and the eCPM. Then, the extraction unit 134 compares the calculated advertisement evaluation values of the distribution candidates, and extracts the advertisement content to be distributed (step S36). Here, it is assumed that the extraction unit 134 has extracted the advertising content AD2.

  Then, the distribution unit 135 of the advertising device 200 transmits the extracted advertising content AD2 to the optimization server 20 (step S37). The optimization server 20 generates the web page W1c using the article content C13 to C18 received from the content server 30 and the advertisement content AD2 received from the advertising device 200 (step S38). Then, the optimization server 20 delivers the generated web page W1c to the user terminal 10 (step S39). The user terminal 10 displays the web page W1c connected on the display screen as the web page W1c below the web page W1b already displayed.

  As described above, the advertising device 200 according to the second embodiment calculates an index value that further considers the degree of similarity with the content that the user is predicted to be interested in, and distributes the index value based on the calculated index value. Extract target advertising content. Thereby, since the advertising device 200 can make it easy to display advertising content having high affinity with, for example, article content that the user is interested in, the advertising effect can be enhanced.

(Third embodiment)
[1. (Advertising extraction process by terminal device)
In each of the above embodiments, the advertising devices 100 and 200 calculate the index value indicating the degree of association between the distribution candidate advertisement content and the content displayed around the arrangement position where the distribution candidate advertisement content is arranged, and the index An example of extracting advertisement content to be distributed based on the value has been described. However, such advertisement extraction processing may be performed by the user terminal 10 which is an example of a terminal device. Below, the advertisement extraction process by the user terminal 10 is demonstrated. In addition, since the advertisement extraction process by the user terminal 10 is the same as the advertisement extraction process performed by the advertisement apparatuses 100 and 200, detailed description thereof is omitted.

  The configuration of the advertisement extraction system 3 according to the third embodiment will be described with reference to FIG. FIG. 11 is a diagram illustrating a configuration example of the advertisement extraction system 3 according to the third embodiment. As shown in FIG. 11, the advertisement extraction system 3 includes a user terminal 10, an optimization server 20, a content server 30, and advertisement servers 50a to 50c. The user terminal 10, the optimization server 20, the content server 30, and the advertisement servers 50a to 50c are connected to be communicable by wire or wireless via a network. Note that the advertisement extraction system 3 illustrated in FIG. 11 may include a plurality of user terminals 10, a plurality of optimization servers 20, and a plurality of content servers 30. In the example shown in FIG. 11, the advertisement extraction system 3 includes three advertisement servers. However, the advertisement extraction system 3 is not limited to three, and the advertisement extraction system 3 includes more advertisement servers. It may be 3 or less.

  The user terminal 10 is a terminal device used by a user, and performs an advertisement extraction process based on the index value calculation performed by the advertising devices 100 and 200 described above.

  When receiving the advertisement request, the advertisement servers 50 a to 50 c extract the advertisement content from a predetermined storage unit in the own device, and distribute the extracted advertisement content to the optimization server 20.

  Further, as shown in FIG. 11, the content server 30 is used by the distributor t1, the advertisement server 50a is used by the distributor t2, the advertisement server 50b is used by the distributor t3, and the advertisement server 50c is Used by distributor t4. Further, the distributors t1 to t4 are different companies.

  Next, the advertisement extraction processing by the advertisement extraction system 3 according to the third embodiment will be described using FIG. 11 while clarifying the data flow.

  First, it is assumed that the user terminal 10 transmits a request for the web page W1a, which is an initially displayed portion of the web page W1, to the optimization server 20 in accordance with the user operation (step S40). When receiving the request for the web page W1a, the optimization server 20 transmits a content request for acquiring article content included in the web page W1a to the content server 30 (step S41), and sends the advertisement request to the advertisement servers 50a to 50a. 50c is transmitted to each (step S41).

  When the content server 30 receives the content request from the optimization server 20, for example, the content server 30 refers to a predetermined storage unit in its own device that stores various article content, and determines the distribution of the article content to be distributed and the article content to be distributed. The arrangement order is determined (step S42). Here, as shown in FIG. 11, the content server 30 determines that the article contents C1 to C6 to be distributed are arranged in a vertical row such that the first, second,... .

  In addition, when receiving the advertisement request from the optimization server 20, the advertisement servers 50 a to 50 c are distributed (displayed) to the user with reference to a predetermined storage unit in the own device storing various advertisement contents. The candidate advertisement content for distribution is extracted (step S42). Here, as shown in FIG. 11, it is assumed that the advertisement server 50a extracts the advertisement content AD11, the advertisement server 50b extracts the advertisement content AD12, and the advertisement server 50c extracts the advertisement content AD13. In order to simplify the description, an example in which each advertisement server extracts advertisement content one by one has been shown, but each advertisement server may extract a plurality of advertisement contents.

  What kind of advertisement content is extracted may be according to the extraction logic of each of the advertisement servers 50a to 50c. For example, the advertising server 50a and the advertising server 50b extract the advertising content based on the eCPM, and the advertising server 50b extracts the advertising content based on the bid price. The advertising content may be extracted. On the other hand, the logic for extracting the advertising content may be unified.

  Next, the content server 30 associates the data of the article content to be distributed with the arrangement order and transmits it to the optimization server 20 (step S43). Further, each of the advertisement servers 50a to 50c also transmits the advertisement content data of the distribution candidates extracted to the optimization server 20 (step S43). And the optimization server 20 transfers the various data received from the content server 30 and advertisement server 50a-50c to the user terminal 10 (step S44).

  When the user terminal 10 receives data from the optimization server 20, the user terminal 10 calculates an index value indicating a degree of association between the distribution candidate advertisement content and the article content displayed around the arrangement position where the distribution candidate advertisement content is arranged. To do. Also, the user terminal 10 extracts display target advertisement content to be actually displayed to the user from the distribution candidate advertisement content based on the calculated index value (step S45).

  The advertisement extraction process will be specifically described. For example, when seven pieces of content are arranged in a vertical row, it is assumed that it is determined in advance to insert the advertising content at the arrangement position L10 (fourth) which is an intermediate position. As a result, the user terminal 10 calculates the similarity between each of the distribution candidate advertisement contents and the article contents C2 to C5 from the arrangement position L10 to the top and bottom, and uses the sum of the calculated similarities as an index value. calculate.

  For example, the user terminal 10 calculates four similarities between the advertisement content AD11 and each of the article contents C2 to C5, and adds up the calculated four similarities so that the advertisement content AD11 and the arrangement position L10 are surrounded. An index value with the displayed article contents C2 to C5 is calculated. Then, the user terminal 10 calculates the advertisement evaluation value of the advertisement content AD11 by adding the calculated index value and the eCPM of the advertisement content AD11.

  Further, the user terminal 10 calculates four similarities between the advertisement content AD12 and each of the article contents C2 to C5, and adds up the calculated four similarities to thereby calculate the advertising content AD12 and the article contents C2 to C5. And the index value is calculated. Then, the user terminal 10 calculates the advertisement evaluation value of the advertisement content AD12 by summing the calculated index value and the eCPM of the advertisement content AD12. Further, the user terminal 10 calculates the advertisement evaluation value in the same manner for the advertisement content AD13.

  And a user terminal extracts the thing with the highest advertisement evaluation value as an advertising content of a display object by comparing each advertisement evaluation value calculated in this way, for example. For example, if the user terminal 10 extracts the advertisement content AD13 as a display target from the distribution candidate advertisement contents AD11 to 13, the user terminal 10 creates a web page W1a from the extracted advertisement content AD13 and the distribution target article contents C1 to C6. Generate and display on the display screen.

  In the advertisement extraction system 3 illustrated in FIG. 11, an example is shown in which the optimization server 20 mediates data transmission / reception between the user terminal 10 and the content server 30 and data transmission / reception between the user terminal 10 and the advertisement servers 50a to 50c. However, the advertisement extraction system 3 does not necessarily need to have the optimization server 20. In such a case, for example, the user terminal 10 transmits a content request directly to the content server 30 and transmits an advertisement request directly to each of the advertisement servers 50a to 50c. The user terminal receives data directly from the content server 30 and the advertisement servers 50a to 50c.

  In this way, the advertisement extraction process is performed not on the server side but on the user terminal 10 side, so that, for example, even in a system in which advertisement server devices by a plurality of distributors are mixed as shown in FIG. It is possible to realize the advertisement extraction process.

[2. Configuration of terminal device]
Next, the configuration of the user terminal 10 according to the third embodiment will be described with reference to FIG. FIG. 12 is a diagram illustrating a configuration example of the user device 10 according to the third embodiment. As illustrated in FIG. 12, the user device 10 includes a communication unit 11, an input unit 12, an output unit 13, and a control unit 14.

  The communication part 11 is implement | achieved by NIC etc., for example. The communication unit 11 is connected to the network by wire or wireless, and transmits / receives information to / from the optimization server 20, the content server 30, and the advertisement servers 50a to 50c.

  The input unit 12 is an input device that receives various operations from the user. For example, the input unit 12 is realized by a keyboard, a mouse, an operation key, or the like. The output unit 13 is a display device for displaying various information. That is, the output unit 13 corresponds to the display screen of the user terminal 10. For example, the output unit 13 is realized by a liquid crystal display or the like. In addition, when a touch panel is employ | adopted for the user terminal 10, the input part 12 and the output part 13 are integrated.

  Returning to FIG. 12, the control unit 14 is realized by, for example, a CPU, an MPU, or the like executing various programs stored in the storage device inside the user apparatus 10 using the RAM as a work area. The control unit 14 is realized by an integrated circuit such as an application specific integrated circuit (ASIC) or a field programmable gate array (FPGA).

  As shown in FIG. 12, the control unit 14 includes a request unit 15, a calculation unit 16, an extraction unit 17, and a display control unit 18, and realizes or executes information processing functions and operations described below. To do. The internal configuration of the control unit 14 is not limited to the information illustrated in FIG. 14, and may be other configurations as long as the configuration performs information processing described later. Further, the connection relationship between the processing units included in the control unit 14 is not limited to the connection relationship illustrated in FIG. 12, and may be another connection relationship.

  The request unit 15 transmits a request for acquiring predetermined content to a predetermined server. In addition, the request unit 15 receives content and candidate advertisement contents (distribution candidates) displayed as a list together with the content. In the example of FIG. 13, the request unit 15 transmits a request for the web page W <b> 1 a to the optimization server 20. Further, the request unit 15 receives article content C1 to C6 displayed as a list as the web page W1a and advertisement content AD11 to AD13 as distribution candidates from the optimization server 20.

  The calculation unit 16 calculates an index value indicating the degree of association between the advertisement content as a distribution candidate and the article content displayed around the arrangement position where the advertisement content is arranged. For example, the calculation unit 16 calculates the similarity with the advertisement content of the distribution candidate for each content displayed around the arrangement position, and uses the calculated similarity to display the content group displayed around the arrangement position. Calculate an index value with the advertising content of the distribution candidate. Further, the calculation unit 16 calculates the advertisement evaluation value by adding the calculated index value and the evaluation value (for example, eCPM) of the advertisement content of the distribution candidate. That is, the calculation unit 16 is a processing unit corresponding to the calculation unit 133 of the advertising device 100.

  Based on the index value calculated by the calculation unit 133, the extraction unit 17 extracts from the advertisement content to be distributed from the advertisement content to be displayed displayed on the output unit 13. For example, the extraction unit 134 extracts the advertisement content of the delivery candidate having the highest advertisement evaluation value calculated using the index value by the calculation unit 133 as a display target. That is, the calculation unit 16 is a processing unit corresponding to the extraction unit 134 of the advertising device 100.

  The display control unit 18 controls display of the content requested by the request unit 15 on the output unit 13. For example, the display control unit 18 creates the web page W1a from the article content and the advertisement content received by the request unit 15, and displays the created web page W1a on the output unit 13. For example, when a user operation is performed on the web page W1a, the display control unit 18 changes the display mode of the web page W1a according to the user operation.

  Note that the advertisement extraction process performed by the control unit 14 described above may be realized by, for example, JavaScript (registered trademark). When the above-described extraction process is performed by a predetermined application (for example, an application for displaying the web page W1), the control unit 14 functions as, for example, an application control unit that controls a predetermined application.

(Other embodiments)
The first to third embodiments described above may be implemented in various different forms other than the above embodiment. Accordingly, other embodiments will be described below.

[1. List information)
In each of the above embodiments, the advertisement extraction process has been described using the web page W1 as an example of list information in which a plurality of contents are arranged in a vertical row. However, the list information according to the present application may be information in which a plurality of contents are arranged in the vertical direction and the horizontal direction as in the web page W2a shown in FIG. Hereinafter, an index value calculation process for extracting the advertisement content arranged at the arrangement position L20 of the advertisement content included in the web page W2a will be described using the advertising device 100.

  FIG. 13 is a diagram illustrating an example of advertisement extraction processing according to another embodiment. For example, the web page W2a shown in FIG. 13 conceptually shows a part of the web page W2. Here, it is assumed that the web page W2a is a portion of the web page W2 that is initially displayed.

  In such a case, for example, the request receiving unit 132 receives a list indicating the web page W2a from the content server 30. For example, the request reception unit 132 receives a list in which display positions of article content to be distributed and text information are associated with each other. The display position of the article content to be distributed is indicated by, for example, the arrangement order and direction.

  Assuming that the arrangement position L20 of the advertisement content, i.e., the arrangement order is i-th, in the example of FIG. 13, the request receiving unit 132 “article content A1: i-3th (left)”, “article content A2: i-3”. “(Th) (right)”, “article content A3: i-2th (right)”, “article content A4: i-2th (middle)”,...

  As described above, when the list is received by the request receiving unit 132, the calculation unit 133 determines, for example, the article content from the arrangement position L20 to the upper and lower two pieces as the article content around the arrangement position L20, and sets the article content around the arrangement position L20. For each article content, the degree of similarity with the distribution candidate advertisement content I is calculated. In other words, in the example of FIG. 13, the calculation unit 133 includes the advertisement content I as a distribution candidate, the i-2th (left) article content A3, the i-2th (middle) article content A4, and the i-2th. Article content A5, i−1th (left) article content A6, i−1th (right) article content A7, i + 1th (left) article content A8, i−1th (right) article content A9, The degree of similarity between each of i + 2 (left) article content A10 and i + 2 (right) article content A11 is calculated. Here, the cosine similarity is described as an example of the similarity, but it may be a category similarity.

FIG. 13 illustrates an example in which the calculation unit 133 calculates the cosine similarity cos _A3 as the similarity between the advertisement content I and the i-2th (left) article content A3. Further, an example is shown in which the calculation unit 133 calculates the cosine similarity cos _A4 as the similarity between the advertisement content I and the i-2th (medium) article content A4. Similarities with other article contents are also as shown in FIG.

  In addition, the calculation unit 133 adds a predetermined weight value to each calculated similarity. In pattern 1, the calculation unit 133 calculates the distance between the arrangement position L20 and the display position of each article content, and multiplies the similarity with the article content closer to the arrangement position L20 by a higher weight value. For example, the calculation unit 133 calculates the length of a straight line connecting the midpoint P20 of the arrangement position L20 and the midpoint of the article content display position as the distance between the arrangement position L20 and the article content display position.

  The midpoint P20 of the arrangement position L20 corresponds to an intersection connecting the diagonal lines of the advertisement area indicated by the arrangement position L20. Further, the article content A3 will be described as an example. The midpoint of the display position of the article content A3 corresponds to the intersection point connecting the diagonal lines of the content area indicated by the display position “i-2th (left)”. Note that such intersection information (for example, coordinates) may be included in the list and transmitted from the content server 30, or the advertising device 100 may have information on each intersection in advance.

  Then, the calculating unit 133 calculates a length D3 of a straight line connecting the midpoint P20 of the arrangement position L20 and the midpoint P3 of the display position of the article content A3. In this way, the calculation unit 133 calculates the length of a straight line that connects the midpoint P20 of the arrangement position L20 and the midpoint of the display positions of each article content around the arrangement position L20. Since it is possible to determine that the article content having a shorter straight line length is closer to the arrangement position L20, the calculating unit 133 multiplies the similarity with the article content having a shorter straight line length by a higher weight value. .

In the example of FIG. 13, since the distance between the arrangement position L20 and the display positions of the (i−1) th (left) advertisement content A6 and the (i−1) th (right) advertisement content A7 is the shortest, the article contents A6 and A7 Is at a position closest to the arrangement position L20. Therefore, calculation unit 133, the similarity _{cos A7} of similarity _{cos A6} and advertising content I and article content A7 with advertising content I and article content A6, multiplied by the highest weight value r (0.9).

In the example of FIG. 13, since the distance between the arrangement position L20 and the display position of the (i + 1) th (left) article content A8 and the (i + 1) th (right) article content A9 is the next shortest, the article contents A8 and A9 are It is the second closest position to the arrangement position L20. Therefore, calculation unit 133, the similarity _{cos A9} of similarity _{cos A8} and advertising content I and article content A9 with advertising content I and article content A8, higher weight values to the second ^r 2 (0.81) Multiply

In this way, the calculation unit 133 sums up the respective degrees of similarity multiplied by the weight value, thereby indicating an index value indicating the degree of association between the advertisement content I and two article content groups from the arrangement position Lx up and down. to calculate the S _I. Specifically, the calculation unit 133 can calculate the index value S _I by the following equation (8).

Index value S _I = r ⁴ cos _A3 + r ³ cos _A4 + r ⁴ cos _A5 + rcos _A6 + rcos _A7 + r ² cos _A8 + r ² cos _A9 + r ⁵ cos _A10 + r ⁵ cos _A11

Then, calculating unit 133, based on the index value S _I calculated using Equation (8), calculates an advertisement evaluation value. For example, the calculation unit 133 calculates the total value of eCPM _I , which is the evaluation value of the advertising content I, and the index value S _I calculated using Expression (8) as the advertising evaluation value E _I. In addition, the extraction unit 134 compares the advertisement evaluation values E _I for the respective distribution candidate advertisement contents, and extracts the advertisement contents to be distributed.

  Next, the pattern 2 will be described. Pattern 2 is different from Pattern 1 in the way of assigning weight values. Specifically, the calculation unit 133 multiplies the similarity with the article content positioned above the arrangement position L20 by the higher weight value when multiplying the similarity with the article content near the arrangement position L20 by a higher weight value. .

For example, the calculation unit 133 includes the i−1th (left) article content A6 and the i−1th (right) article content A7 that are closest to the arrangement position L20 and located upward in the arrangement position L20. The similarity cos _A6 and cos _A7 , which are similarities to the advertisement content I, are multiplied by the highest weight value r (0.9). In addition, the calculation unit 133 sets the similarity cos _A4 between the advertisement content I and the i-2th (medium) article content A4 that is next close to the arrangement position L10 and in the upper direction of the arrangement position L10. Multiply by the second highest weight value r ² (0.81).

In this way, the calculation unit 133 sums up the respective degrees of similarity multiplied by the weight value, thereby indicating an index value indicating the degree of association between the advertisement content I and the article content group from the arrangement position L10 up and down two pieces. to calculate the S _I. Specifically, the calculation unit 133 can calculate the index value S _I by the following equation (9).

Index value S _I = r ³ cos _A3 + r ³ cos _A4 + r ³ cos _A5 + rcos _A6 + rcos _A7 + r ⁴ cos _A8 + r ⁴ cos _A9 + r ⁵ cos _A10 + r ⁵ cos _A11 + r ⁵ cos

Then, calculating unit 133, based on the index value S _I calculated using Equation (9), calculates the advertisement evaluation value. For example, the calculation unit 133 calculates the total value of eCPM _I , which is the evaluation value of the advertising content I, and the index value S _I calculated using Expression (9) as the advertising evaluation value E _I. In addition, the extraction unit 134 compares the advertisement evaluation values E _I for the respective distribution candidate advertisement contents, and extracts the advertisement contents to be distributed.

[2. Similarity calculation using image information)
In the above embodiment, the calculation unit 133 calculates the cosine similarity between the feature word extracted from the text of the advertisement content as the distribution candidate and the feature word extracted from the text of the article content displayed around the arrangement position. Indicated. However, the calculation unit 133 is not limited to this example, and the calculation unit 133 extracts the feature words extracted from the image data corresponding to the distribution candidate advertisement content and the feature words extracted from the image data corresponding to the article content displayed around the arrangement position. The similarity may be calculated.

[3. Configuration of advertising device]
In each of the above embodiments, an example has been described in which the advertising device 100 or 200 performs an advertisement extraction process using the list received from the content server 30. However, the content server 20 and the advertising device 100 may be integrated as the advertising device 100. In addition, the content server 20 and the advertising device 200 may be integrated as the advertising device 200.

[4. program〕
Further, the advertising devices 100 and 200 according to the above-described embodiments are realized by a computer 1000 having a configuration as shown in FIG. 14, for example. Hereinafter, the advertisement apparatus 100 will be described as an example. FIG. 11 is a hardware configuration diagram illustrating an example of a computer 1000 that implements the functions of the advertising device 100. The computer 1000 includes a CPU 1100, RAM 1200, ROM 1300, HDD 1400, communication interface (I / F) 1500, input / output interface (I / F) 1600, and media interface (I / F) 1700.

  The CPU 1100 operates based on a program stored in the ROM 1300 or the HDD 1400 and controls each unit. The ROM 1300 stores a boot program executed by the CPU 1100 when the computer 1000 is started up, a program depending on the hardware of the computer 1000, and the like.

  The HDD 1400 stores programs executed by the CPU 1100, data used by the programs, and the like. The communication interface 1500 receives data from other devices via the communication network 50 and sends the data to the CPU 1100, and transmits the data generated by the CPU 1100 to other devices via the communication network 50.

  The CPU 1100 controls an output device such as a display and a printer and an input device such as a keyboard and a mouse via the input / output interface 1600. The CPU 1100 acquires data from the input device via the input / output interface 1600. Further, the CPU 1100 outputs the generated data to the output device via the input / output interface 1600.

  The media interface 1700 reads a program or data stored in the recording medium 1800 and provides it to the CPU 1100 via the RAM 1200. The CPU 1100 loads such a program from the recording medium 1800 onto the RAM 1200 via the media interface 1700, and executes the loaded program. The recording medium 1800 is, for example, an optical recording medium such as a DVD (Digital Versatile Disc) or PD (Phase change rewritable disk), a magneto-optical recording medium such as an MO (Magneto-Optical disk), a tape medium, a magnetic recording medium, or a semiconductor memory. Etc.

  For example, when the computer 1000 functions as the advertising device 100 according to the first embodiment, the CPU 1100 of the computer 1000 realizes the function of the control unit 130 by executing a program loaded on the RAM 1200. The HDD 1400 stores data in the advertisement information storage unit 120. The CPU 1100 of the computer 1000 reads these programs from the recording medium 1800 and executes them, but as another example, these programs may be acquired from other devices via the communication network 50.

  Further, for example, when the computer 1000 functions as the advertising device 200 according to the second embodiment, the CPU 1100 of the computer 1000 implements the function of the control unit 430 by executing a program loaded on the RAM 1200.

[5. Others]
Of the processes described in the above embodiment, all or part of the processes described as being automatically performed can be performed manually, or all of the processes described as being performed manually or A part can be automatically performed by a known method. In addition, the processing procedures, specific names, and information including various data and parameters shown in the document and drawings can be arbitrarily changed unless otherwise specified.

  Further, each component of each illustrated apparatus is functionally conceptual, and does not necessarily need to be physically configured as illustrated. In other words, the specific form of distribution / integration of each device is not limited to that shown in the figure, and all or a part thereof may be functionally or physically distributed or arbitrarily distributed in arbitrary units according to various loads or usage conditions. Can be integrated and configured.

  Moreover, each embodiment mentioned above can be combined suitably in the range which does not contradict a process content.

  As described above, some of the embodiments of the present application have been described in detail based on the drawings. It is possible to implement the present invention in other forms with improvements.

  In addition, the “section (module, unit)” described above can be read as “means” or “circuit”. For example, the determination unit can be read as determination means or a determination circuit.

DESCRIPTION OF SYMBOLS 1 Advertisement extraction system 2 Advertisement extraction system 3 Advertisement extraction system 10 User terminal 14 Control part 16 Calculation part 17 Extraction part 20 Optimization server 30 Content server 100 Advertising apparatus 120 Advertisement information storage part 130 Control part 133 Calculation part 134 Extraction part 200 Advertisement Device 230 Control unit 236 Prediction unit

Claims (12)

A calculation unit that calculates an index value indicating a degree of association between the distribution candidate advertisement content displayed together with the content and the content displayed around the arrangement position where the distribution candidate advertisement content is arranged;
An advertising device comprising: an extraction unit that extracts advertisement content to be distributed from the distribution candidate advertisement content based on the index value calculated by the calculation unit. The calculation unit calculates a similarity with the advertisement content of the distribution candidate for each content displayed around the arrangement position, and uses the calculated similarity to display a group of contents displayed around the arrangement position The advertising device according to claim 1, wherein an index value indicating the degree of association between the advertisement content and the distribution candidate advertisement content is calculated. The advertising device according to claim 2, wherein the calculation unit weights higher the degree of similarity with content displayed at a position closer to the arrangement position among contents displayed around the arrangement position. The advertisement according to claim 2, wherein the calculation unit weights the higher the similarity with the content displayed above the arrangement position among the contents displayed around the arrangement position. apparatus. A prediction unit for predicting content that the user is interested in among the displayed content and advertisement content;
The advertisement according to any one of claims 2 to 4, wherein the calculation unit weights higher the similarity with the content predicted to be interesting to the user by the prediction unit. apparatus. The calculation unit calculates a similarity between information extracted from text included in the advertisement content of the distribution candidate and information extracted from text included in the content displayed around the arrangement position. The advertising device according to any one of claims 2 to 5. The calculation unit calculates a similarity between a category to which the distribution candidate advertisement content belongs and a category to which the content displayed around the arrangement position belongs. The advertising device described in 1. The calculation unit calculates an index value indicating the relevance by summing the similarities,
The extraction unit extracts advertisement content to be distributed from the distribution candidate advertisement content based on the sum of the index value calculated by the calculation unit and the evaluation value of the distribution candidate advertisement content. The advertising device according to any one of claims 2 to 7. An advertisement extraction method executed by a computer,
A calculation step of calculating an index value indicating a degree of association between the distribution candidate advertisement content displayed together with the content and the content displayed around the arrangement position where the distribution candidate advertisement content is arranged;
An extraction method comprising: an extraction step of extracting an advertisement content to be distributed from the distribution candidate advertisement content based on the index value calculated in the calculation step. A calculation procedure for calculating an index value indicating a degree of association between the distribution candidate advertisement content displayed together with the content and the content displayed around the arrangement position where the distribution candidate advertisement content is arranged;
An advertisement extraction program causing a computer to execute an extraction procedure for extracting advertisement content to be distributed from the advertisement candidate advertisement content based on the index value calculated by the calculation procedure. A reception unit that receives the content and advertisement content that is a candidate for distribution displayed together with the content;
A calculation unit that calculates an index value indicating a degree of association between the distribution candidate advertisement content received by the reception unit and the content displayed around an arrangement position where the distribution candidate advertisement content is arranged;
A terminal device, comprising: an extraction unit that extracts advertisement content to be displayed from the distribution candidate advertisement content based on the index value calculated by the calculation unit. An acceptance procedure for accepting content and candidate advertisement content to be listed together with the content;
A calculation procedure for calculating an index value indicating a degree of association between the distribution candidate advertisement content received by the reception procedure and the content displayed around the arrangement position where the distribution candidate advertisement content is arranged;
A terminal control program that causes a terminal device to execute an extraction procedure for extracting advertising content to be displayed from the distribution candidate advertising content based on the index value calculated by the calculation procedure.

Images