JP6576404B2 - Advertisement granting device, advertisement granting method, and advertisement granting program - Google Patents

Description

  The present invention relates to an advertisement giving device, an advertisement giving method, and an advertisement giving program.

  In recent years, with the rapid development of the Internet, advertisements have been actively provided via the Internet. As for advertisement grant, there are those that display advertisement information such as images and catch phrases that display companies and products at a predetermined position on the web page, and transition to the advertiser's web page when such advertisement information is clicked is there.

In recent years, a technique has been proposed in which an advertisement is effectively recognized by a user. For example, Patent Literature 1 discloses a terminal device that can select and provide an advertisement that matches a user's hobbies and preferences to the user.
In order to display an advertisement that attracts the user's interest, it is also effective to provide an advertisement (hereinafter referred to as a “recommendation advertisement”) that is related to the article that the user is browsing. In this case, for example, a characteristic search word is specified from the article text, and the advertisement server is searched using the specified search word to determine an advertisement related to the search word.

JP 2015-22561 A

Giving recommendation advertisements as described above requires time for advertisement search, and the processing load on the advertisement server side is large, so there are limited articles to which recommendation advertisements are given. On the other hand, since new articles are published every day, there are an enormous number of articles that can be subject to advertisements, and there are cases where advertisements have a valid period. Even if the article is given, it is necessary to give the recommended advertisement again when the validity period of the advertisement expires.
In this way, there are a large number of articles to which advertisements can be assigned, but there are only a limited number of articles to which recommendation advertisements can be assigned. Therefore, candidates for articles to receive recommendation advertisements are selected from a large number of articles. There is a need to. However, simply selecting a candidate for an article to which a recommendation advertisement is given at random may not lead to an advertising effect.

  The present invention has been made in view of such circumstances, and an advertisement granting apparatus, an advertisement granting method, and an advertisement capable of selecting an article having a relatively high advertising effect as a candidate for an article to which a recommendation advertisement is given. The purpose is to provide grant programs.

In order to solve the above problems, the present invention employs the following means.
In the first aspect of the present invention, a recommendation evaluation value is calculated for each of a plurality of articles, and based on the recommendation evaluation value, an article candidate to which a recommendation advertisement that is an advertisement related to the posted content of the article is selected is selected. Article selection means, and recommendation advertisement determination means for determining a recommendation advertisement related to the posted content of the article for all or some of the article candidates selected by the article selection means, The selection means calculates the recommendation evaluation value for the target article for which the target evaluation is calculated. Display number evaluation value for the predicted number of advertisements displayed for a predetermined period of time and whether the target article is published or a predetermined time after the article is published Based on the recommendation evaluation value of each of the target articles calculated by the calculation means, the calculation means for calculating the recommendation evaluation value of the target article using at least one of the profitability evaluation values related to the predicted profit for the predetermined period And a selection means for selecting candidate articles to which the recommendation advertisement is to be given from among the plurality of articles.

According to a second aspect of the present invention , there is provided an advertisement granting method executed by a computer , wherein a recommendation evaluation value is calculated for each of a plurality of articles, and the advertisement is related to the posted content of the article based on the recommendation evaluation value. An article selection step of selecting candidate articles to which a certain recommendation advertisement is given, and a recommendation advertisement related to the posted content of the article for all or some of the article candidates selected by the article selection step. A recommended advertisement determination step to determine, the article selection step, the article evaluation of the target article for calculating the recommendation evaluation value or the browsing number evaluation value related to the predicted cumulative browsing number in a predetermined period after the article issuance, The number of impressions for the predicted advertisement display count for a predetermined period from a predetermined time after the article issuance or article issuance of the target article, And calculating the recommendation evaluation value of the target article using at least one of the profitability evaluation value related to the predicted profit from the predetermined time after the issue of the target article or after the issue of the article, and the calculation A selection step of selecting candidate articles to which the recommended advertisement is to be assigned from among the plurality of articles based on the recommendation evaluation value of each of the target articles calculated in the process.

  According to a third aspect of the present invention, a recommendation evaluation value is calculated for each of a plurality of articles, and candidate articles to which a recommendation advertisement that is an advertisement related to the posted content of the article is selected based on the recommendation evaluation value. Causing the computer to execute an article selection process and a recommendation advertisement determination process for determining a recommendation advertisement related to the posted content of the article for all or some of the article candidates selected in the article selection process In the advertisement granting program, the article selection processing includes the article count of the target article for which the recommendation evaluation value is calculated, or the browsing count evaluation value related to the predicted cumulative browsing count in a predetermined period after the article issuance. Impression rating value for forecasted ad impressions for a given period from the time the article is published or after the article is published, And calculating the recommendation evaluation value of the target article using at least one of the profitability evaluation values related to the predicted revenue from a predetermined time after the article issuance or post-issue of the target article, and the calculation And a selection process for selecting a candidate for an article to which the recommended advertisement is to be assigned from among a plurality of the articles based on a recommendation evaluation value of each of the target articles calculated in the process.

  According to the present invention, it is possible to select an article having a relatively high advertising effect as a candidate for an article to which a recommended advertisement is given.

1 is a diagram illustrating an overall configuration of an advertisement distribution system according to an embodiment of the present invention. It is the figure shown about the hardware constitutions of the media server which concerns on one Embodiment of this invention, an advertiser server, and an advertisement delivery server. It is the functional block diagram which showed an example of the function with which the advertisement provision apparatus which concerns on one Embodiment of this invention is provided. It is a figure for demonstrating the calculation method of the recommendation evaluation value which concerns on one Embodiment of this invention. It is a figure for demonstrating the advertising effect by the difference in the browsing number characteristic of an article.

Hereinafter, an advertisement granting apparatus, an advertisement granting method, and an advertisement granting program according to an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a diagram showing an overall configuration of an advertisement distribution system according to the present embodiment. As shown in FIG. 1, the advertisement distribution system 1 includes a media (medium) server 10, an advertiser (advertisement agency) server 11, and an advertisement distribution server 12 as main components. In FIG. 1, one media server 10, advertiser server 11, and advertisement distribution server 12 are illustrated, but the number of these is not limited to the illustrated mode, and can be increased according to the situation as appropriate. Is possible. The media server 10, the advertiser server 11, and the advertisement distribution server 12 are communicably connected to each other via the network 15, and browsing from the user terminal 13 to various user terminals 13 that are also connected to the network 15. Provide services (for example, distribution of articles and advertisements) according to the request. An example of the network 15 is an Internet line, but another network line may be used.

  The media server 10 is a server of a mass media such as newspapers, magazines, radio broadcasts, television broadcasts, and a portal site, and various articles are uploaded every day. When the media server 10 receives an article browsing request from the user terminal 13 via the network 15, the media server 10 distributes the article that is the object of the browsing request toward the user terminal 13 via the network 15.

  The advertiser server 11 is a server that provides products and services to the user terminal 13. For example, a server owned by an advertiser such as a corporation that provides a product, service, etc. to consumers as an example, or an advertisement that conducts advertising related to a product or service of the corporation in response to a request from the corporation An example is a server owned by an agency. The advertiser server 11 stores various advertisements related to products and services of the corporation or the like in the advertisement database.

  The user terminal 13 is a so-called information processing apparatus such as a personal computer or a tablet terminal, and includes a communication unit for performing communication such as the Internet, an input unit for a user to input an input operation, and a display unit for displaying information. Have For example, the user terminal 13 accesses the media server 10 via the network 15 and browses articles (contents) uploaded to the media server 10 using a web browser that the user terminal 13 has. Further, the web browser of the user terminal 13 interprets an article acquired from the media server 10, for example, an article content described in HTML (HyperText Markup Language) or the like, and from the media server 10 to the advertisement distribution server 12 according to the interpretation. Send information about acquired articles and request advertisements.

For example, when the advertisement distribution server 12 receives article information and an advertisement request from the user terminal 13, the advertisement distribution server 12 retrieves advertisements assigned to the article from article advertisement information in an article advertisement database 32 (see FIG. 3) described later. The extracted advertisement is transmitted to the user terminal 13. Thereby, for example, an advertisement given to the article is displayed on the web page of the user terminal 13 together with the article requested to be browsed by the user.
Further, the advertisement distribution server 12 includes an advertisement granting device 20. The advertisement giving device 20 is a device that gives an advertisement to an article (for example, a newly arrived article) to which no advertisement is given. The advertisement information provided by the advertisement assignment device 20 is stored in the article advertisement database (FIG. 3) provided in the advertisement assignment device 20 in association with the article information, and used for advertisement distribution as described above. The details of the advertisement giving device 20 will be described later.

Next, the hardware configuration of the media server 10, the advertiser server 11, and the advertisement distribution server 12 according to the present embodiment will be described with reference to FIG. As shown in FIG. 2, the media server 10, the advertiser server 11, and the advertisement distribution server 12 are all so-called computer systems. For example, a CPU 21 and a ROM (Read) for storing a program executed by the CPU 21 and the like. (Only Memory) 22, a RAM (Random Access Memory) 23 functioning as a work area when executing each program, a hard disk drive (HDD) 24 as a mass storage device, a communication interface 25 for connecting to a network, a keyboard And an input unit 26 including a mouse and a display unit 27 including a liquid crystal display device for displaying data. These units are connected via a bus 28.
The storage medium for storing the program executed by the CPU 21 is not limited to the ROM 22 and may be another recording medium including a magnetic disk, a magneto-optical disk, a semiconductor memory, and the like.

Various programs are stored in the ROM 22, and the CPU 21 reads out the programs from the ROM 22 to the RAM 23 and executes them to realize various functions.
Note that the media server 10, the advertiser server 11, and the advertisement distribution server 12 do not necessarily have the display unit 27. In addition, the media server 10, the advertiser server 11, and the advertisement distribution server 12 may be configured to be connected to the input unit 26 via a network or the like and be remotely operable.

Next, the advertisement giving apparatus 20 according to the first embodiment of the present invention will be described in detail.
FIG. 3 is a functional block diagram illustrating an example of functions provided in the advertisement granting device 20. As shown in FIG. 3, the advertisement granting apparatus 20 includes, for example, an article database 31, an article advertisement database 32, an article selection unit 33, and a recommended advertisement determination unit 34 as main components.
Various processes described later realized by these units are realized, for example, when the CPU 21 reads out and executes an advertisement granting program stored in the ROM 22 into the RAM 23.

The article database 31 stores information on a plurality of articles to which no advertisement is given. For example, in the article database 31, the latest articles issued from each medium (medium) every day are registered and accumulated.
The article advertisement database 32 stores article advertisement information in which article information and advertisement information attached to the article are associated with each other.

The article selection unit 33 selects candidate articles to which a recommendation advertisement, which is an advertisement related to the posted content of the article, is selected from among a large number of articles stored in the article database 31.
Here, in order to give a recommended advertisement, it is necessary to specify a search word and search the advertiser server 11 based on the search word, which takes time. For this reason, it is practically impossible to give a recommendation advertisement to all articles stored in the article database 31. Therefore, although it is necessary to select the article candidate to which the recommendation advertisement is to be given, it is preferable to select the article to which the recommendation advertisement is to be given from the aspect of the advertisement effect. From this point of view, the article selection unit 33 according to the present embodiment is an article candidate that gives a recommendation advertisement to an article that can be expected to have a larger number of advertisement display (impression: imp) and can be expected to have a higher profitability. We are going to select as.
Specifically, the article selection unit 33 in the present embodiment calculates a recommendation evaluation value by using the display count evaluation value related to the predicted advertisement display count in the article and the profitability evaluation value related to the predicted profit of the article, and this recommendation evaluation value. Article candidates that give recommendation advertisements are selected using as an index.

  As illustrated in FIG. 3, the article selection unit 33 includes a calculation unit 41 and a selection unit 42. For example, the calculation unit 41 repeatedly acquires a single article as a target article from a large number of articles stored in the article database 31 and repeatedly calculates a recommendation evaluation value of the acquired target article. More specifically, the calculation unit 41 uses the display count evaluation value related to the predicted advertisement display count in the predetermined period from the issue of the target article and the profitability evaluation value related to the predicted revenue in the predetermined period from the issue of the target article. The recommendation evaluation value of each article is calculated. Here, the “predetermined period” can be arbitrarily set. For example, it is preferable to set in consideration of a general advertisement insertion period. For example, the predetermined period is set to a period of 5 days or more. Details of the calculation method of the display count evaluation value and the profitability evaluation value will be described later.

  Based on the recommendation evaluation value of each article (target article) calculated by the calculation unit 41, the selection unit 42 selects candidate articles to which a recommended advertisement is to be assigned from among a plurality of articles stored in the article database 31. To do. For example, the selection unit 42 selects a predetermined number of articles in order from an article having a higher recommendation evaluation value as a candidate.

  The recommendation advertisement determination unit 34 gives a recommendation advertisement to all or some of the article candidates selected by the article selection unit 33. For example, the recommended advertisement determination unit 34 sequentially extracts articles from the selected article candidates in descending order of the recommendation evaluation value, analyzes the contents of the extracted articles, and specifies a search word for giving an advertisement. The advertiser server 11 is searched using the specified search word, and an advertisement related to the search word is acquired from the advertiser server 11 as a recommended advertisement. It should be noted that there are no particular limitations on the recommended advertisement search method and the like, and it is possible to appropriately adopt known techniques.

  Further, the recommended advertisement determination unit 34 determines the information of the recommended advertisement (for example, identification information for specifying the recommended advertisement) and the information of the recommended article to which the recommended advertisement is assigned (for example, to specify the article). The article advertisement information in which the identification number, the article content, the search word extracted from the article body, etc.) are associated is stored in the article advertisement database 32.

  Next, the display count evaluation value, profitability evaluation value, and recommendation evaluation value calculated by the calculation unit 41 will be described in detail.

[1. (Impression count evaluation value)
First, as described above, the display count evaluation value is an evaluation value obtained based on a result of predicting how often an advertisement given to the target article is displayed in a predetermined period from the issue of the target article. Yes, for example, as shown in FIG. 4, the browsing number evaluation value related to the predicted cumulative browsing number in the predetermined period from the issue of the target article, and the display probability evaluation value related to the display probability of the display frame in the predetermined period from the issue of the target article It is calculated using and. Specifically, the display count evaluation value can be obtained by multiplying the browsing count evaluation value described later and the display probability evaluation value.

[1-1. (Browsing number evaluation value)
For example, the number of articles viewed is greatly influenced by the media issuing the article. For example, articles distributed by high-profile media and major media tend to be more likely to be read by users than articles distributed by low-profile media and small-scale media. Therefore, in this embodiment, the browsing number evaluation value for each medium is calculated from the browsing number result data of each medium in the past predetermined period. In the present embodiment, PV (page view) or access count can be used as the browse count.

As the browsing number evaluation value, for example, by statistically processing the browsing number actual data in the past predetermined period of the issuer media of the target article, the cumulative number of browsing in the predetermined period is predicted from the article publication of the target article. It can be adopted as the evaluation value for the number of browsing articles. Since various methods are known for predicting future trends from past performance data, those techniques may be used as appropriate.
Also, if the process is simplified, for example, the latest number of browsing per unit time or the number of browsing on the day is acquired for each medium, and the total number of browsing obtained by integrating these browsing numbers over a predetermined period is adopted as the browsing evaluation value. May be. Alternatively, the average number of browsing per day is acquired for each medium from the number of browsing in the most recent predetermined period (for example, several months, several days, etc.), and the total browsing number obtained by integrating this average browsing number over the predetermined period is the browsing evaluation value. May be adopted.

  What is important here is that, as described above, the browsing evaluation value is based on a value obtained by predicting the cumulative browsing number in a predetermined period from the issue of the target article. For example, it is assumed that there are two articles a and b having browsing number characteristics as shown in FIG. In this case, if only the peak value of the number of browsing is evaluated, the evaluation of article a is higher than that of article b. However, when considering the advertising effect, it is preferable to evaluate the cumulative number of browsing in a predetermined period from the issue of the article rather than the peak value of the number of browsing. Thus, in this embodiment, since the estimated cumulative number of articles viewed in a predetermined period from the issue of the article is evaluated, the peak value is higher than that of an article in which the peak value is conspicuous but the cumulative number of articles viewed in the predetermined period is relatively small. Articles that are not conspicuous but have a relatively large cumulative number of views in a predetermined period can be prioritized.

  In the present embodiment, the browsing number evaluation value described above is calculated in advance for each medium, and browsing number information in which the browsing number evaluation value is associated with the media information is prepared in advance, and the browsing number information is used. The browsing number evaluation value corresponding to the publishing medium of each article is obtained. Note that the browsing number information creation process may be performed, for example, as a pre-process in which the calculation unit 41 performs the calculation process of the recommendation evaluation value, or may be performed at any time at a predetermined timing. Thus, by obtaining the browsing number evaluation value of each article using the browsing number information in which the media and the browsing number evaluation value are associated with each other, the processing load can be reduced and the processing time can be shortened. It becomes.

In addition, the total number of browsing varies depending on the category of articles even in the same media. For example, if a media strong in politics distributes articles about fashion, the number of articles viewed on politics shows a stable high value, while the number of articles viewed on fashion cannot be expected to be so high. For example, articles related to politics often show the browsing characteristics shown in article a in FIG. 5, while articles related to fashion may show the browsing characteristics shown in article b in FIG. In this way, the characteristics and peak value of the number of browsing differ depending on the category of the article.
Therefore, it is preferable to calculate the browsing number evaluation value in consideration of not only the media but also the category of the article. In this case, for example, if the browsing number actual data in the past predetermined period of the media is further divided into categories of articles, the browsing number evaluation data for each category of articles in each medium is obtained using the browsing number actual data for each category. Good. In this case, browsing number information in which browsing number evaluation values are associated with each medium and each category of articles is prepared in advance.

[1-2. (Display probability evaluation value)
The display probability evaluation value is an evaluation value obtained by predicting the display probability of the article display frame. This display probability evaluation value can also be evaluated for each medium in the same way as the browsing number evaluation value described above. For example, the display probability evaluation value is calculated for each medium using the display probability actual data of the advertising space in the past predetermined period of each medium. For example, as the display probability evaluation value, an average value of the display probability of the advertisement space in the past predetermined period, an evaluation value obtained by processing the display probability display data of the advertisement space using a statistical method, etc. Can do.

In the present embodiment, the display probability evaluation value described above is calculated in advance for each medium, and display probability information in which the display probability evaluation value is associated with the media information is prepared in advance, and the display probability information is used. The display probability evaluation value of the target article is obtained. Note that the display probability information creation processing may be performed, for example, as pre-processing in which the calculation unit 41 performs calculation processing for the recommendation evaluation value, or may be performed at any time at a predetermined timing. Thus, by obtaining the display probability evaluation value of each target article using the display probability information in which the media and the display probability evaluation value are associated, the processing burden can be reduced and the processing time can be shortened. It becomes possible.
Note that the display probability evaluation value is also influenced by the category as much as the browsing number evaluation value described above. Therefore, the display probability evaluation value may be calculated for each medium and each article category.

[2. (Profitability evaluation value)
The profitability evaluation value is an evaluation value related to predicted advertising revenue obtained from the advertisement in a predetermined period from the issue of the article when a recommended advertisement is given to the target article.
The profitability evaluation value can be evaluated for each medium. For example, the revenue evaluation value is calculated for each medium using the actual revenue of each medium in the past predetermined period. As the profit evaluation value, for example, an actual profit rate per impression in each medium, an actual profit per advertisement space, or the like can be adopted. Note that, similarly to the above-described browsing number evaluation value and the like, the profit evaluation value may be calculated for each medium and each article category in order to be influenced by the category.

  In the present embodiment, the above-described profitability evaluation value is calculated in advance for each medium, and profitability information in which the profitability evaluation value is associated with media information is prepared in advance, and this profitability information is used. , Get a profitability assessment of the target article. Note that the profitability creation process may be performed, for example, as a preprocess in which the calculation unit 41 performs a calculation process of the recommendation evaluation value, or may be performed at any time at a predetermined timing. Thus, by obtaining the profitability evaluation value of each target article using the profitability information in which the media and the profitability evaluation value are associated with each other, the processing burden can be reduced and the processing time can be shortened. It becomes possible.

[3. (Recommendation evaluation value)
As shown in FIG. 4, the recommendation evaluation value is calculated using the display count evaluation value and the profitability evaluation value described above. For example, a predetermined arithmetic expression having the impression number evaluation value and the profitability evaluation value as parameters is held, and the recommendation evaluation value is obtained by substituting the display number evaluation value and the profitability evaluation value into the calculation expression. It is possible to calculate.
For example, the recommendation evaluation value may be a value obtained by simply multiplying the display count evaluation value and the profitability evaluation value, or may be a weighted average value obtained by multiplying these by a predetermined weighting coefficient. What is necessary is just to set suitably about the arithmetic formula which calculates a recommendation evaluation value according to which of the frequency | count of display and profitability is emphasized according to operation | use.

  Next, the advertisement giving process executed by the advertisement giving apparatus 20 according to the present embodiment will be briefly described mainly with reference to FIGS. 3 and 4. The advertisement granting process described below is repeatedly performed at a predetermined timing, for example, the timing when the article database 31 is updated.

  First, as preprocessing, the browsing number evaluation value, the display probability evaluation value, and the profitability evaluation value described above at a predetermined timing are calculated for each medium, and the browsing number information in which the media information and each evaluation value are associated with each other. Display probability information and profitability information are created and stored in a predetermined storage area. As described above, these pieces of information may be created in consideration of the category of the article.

  Next, the calculation unit 41 sequentially reads articles stored in the article database 31 as target articles, and calculates a recommendation evaluation value for the read target articles. Specifically, the browsing number evaluation value, the display probability evaluation value, and the profitability evaluation value associated with the publisher media (and article category) of the target article from the browsing number information, display probability information, and profitability information, respectively. get. Subsequently, the display number evaluation value of the target article is obtained by multiplying the acquired browsing number evaluation value and the display probability evaluation value. Further, the recommendation evaluation value of the target article is calculated by substituting the profitability evaluation value and the display frequency evaluation value into a predetermined arithmetic expression. By repeatedly performing such processing for each article, a recommendation evaluation value is calculated for the article stored in the article database 31.

  In the present embodiment, as described above, the evaluation value is calculated according to the medium or the medium and the category. Therefore, the same recommendation evaluation value is calculated for articles published from the same media or articles of the same media and the same category. From this, articles stored in the article database 31 in advance may be classified by media (and by category), and a recommendation evaluation value may be assigned to each by the above-described method. Thereby, it is possible to efficiently perform the calculation process of the recommendation evaluation value.

  When the recommendation evaluation value is calculated for each article in this way, the selection unit 42 extracts a predetermined number of articles in descending order of the recommendation evaluation value, and selects the extracted articles as candidate articles to which a recommendation advertisement is to be assigned. To do. The selected article is registered, for example, in a queue of a recommendation engine that gives a recommendation advertisement. At this time, the higher the recommendation evaluation value, the higher the position.

  The recommendation advertisement determination unit 34 extracts articles in order from the article having the highest recommendation evaluation value among the article candidates selected by the selection unit 42 and determines a recommendation advertisement for the extracted article. Thereby, a recommendation advertisement is sequentially given to the articles registered in the queue, and the article advertisement information in which the article information and the recommendation advertisement information are associated is stored in the article advertisement database 32.

  As described above, according to the advertisement granting apparatus, the advertisement granting method, and the advertisement granting program according to the present embodiment, the display count evaluation value related to the predicted advertisement display count of the article in the predetermined period from the article issuance, and the predetermined from the article issuance. A recommendation evaluation value is calculated using the profitability evaluation value related to the predicted profit of the article in the period, and an article candidate to which a recommendation advertisement is given is selected using the recommendation evaluation value as an index. Thereby, it becomes possible to evaluate the display frequency and profitability in the whole predetermined period (for example, advertisement insertion period) from the time of article publication. As a result, from the viewpoint of the number of display times and profitability, it is possible to select an appropriate article with a higher advertising effect as an article candidate to which a recommendation advertisement is given. This effect is effective when, for example, an article to which a recommendation advertisement is given by evaluating only the PV of the article on the current day, or an article candidate is selected in consideration of only the past PV peak value. This effect cannot be obtained.

  Next, an advertisement giving apparatus according to the second embodiment of the present invention will be described. The advertisement granting apparatus according to the present embodiment has substantially the same configuration as that of the first embodiment described above, but the calculation method of the display count evaluation value by the calculation unit 41 is different. Hereinafter, description of points common to the first embodiment will be omitted, and different points will be mainly described.

  In the present embodiment, the display count evaluation value can be obtained by multiplying the browsing evaluation value and the display probability evaluation value, as in the first embodiment described above. In the present embodiment, the browsing evaluation value is calculated by using browsing record data from the time of article publication to the present in the target article, and past article browsing record data in the issuer medium of the target article.

[Browsing number evaluation value]
There are a wide variety of articles published by the media that publishes the target articles, and there are, of course, articles that are often read and articles that are not often read. In such a case, if the number of browsing evaluations of the target article is calculated using the total cumulative number of browsing in the past predetermined period for all articles without particularly distinguishing those articles, the prediction error becomes large. The prediction accuracy may be reduced.

  Therefore, in the present embodiment, browsing of articles having a high degree of similarity with the trend of the number of browsing results from the issue of the target article to the present in the target article from the browsing number actual data in the past predetermined period of the media issuing the target article. The result data is extracted, and the cumulative number of browsing of the target articles after the present is predicted using the extracted browsing result data. This makes it possible to exclude information on articles that have a low correlation with the target article, and expect improvement in the prediction accuracy of the number of views. In addition, about the browsing number from the article issue of the object article to the present, since the browsing results already exist, a track record value may be used.

Specifically, with respect to a plurality of articles issued by the publisher media in the past, the actual number of views from the article issuance until a predetermined period (for example, 5 days) is displayed for each predetermined unit period (for example, every hour, every 12 hours). For example, every day). Thereby, the distribution (viewing tendency) of the number of browsing of each article in the predetermined period from the issue of the article and the total number of browsing in the predetermined period can be grasped. Then, from the issue of the target article, extract articles similar to the distribution (viewing tendency) of the number of browsing results per fixed period, and statistically process the cumulative number of readings of the extracted articles. Predict the number of views in a given period. In this way, by using the past articles with similar browsing number distributions, predicting the number of target articles viewed in the future from the current time of the target articles, compared to the case of predicting the number of browsing using all articles The prediction accuracy can be improved.
In addition to articles that show a simple browsing trend that has increased since the publication of the article and then declined, there are also seasonal articles in which the number of browsing periodically increases in the order of week, month, year, etc. However, for example, by appropriately setting the predetermined period, it is possible to cover these characteristics.

[Display probability evaluation value]
Although the display probability evaluation value according to the first embodiment described above may be used as the display probability evaluation value, for example, the display probability of the advertisement display frame also depends on the installation position of the advertisement display frame. Therefore, the display probability evaluation value may be calculated in consideration of the installation position of the advertisement frame in the article.
As for the display probability, depending on the media, the scroll rate may vary depending on the article freshness. Therefore, for example, similarly to the above-described browsing number evaluation value in the present embodiment, the display probability from the article issuance to a predetermined period (for example, 5 days) is set for each predetermined unit period (for example, every hour, every 12 hours, Record every day, etc.), extract past similarity probability data with a high degree of similarity in consideration of the distribution of these display probabilities, and predict the display probability of the target article using the extracted display probability data It is good.

  Thus, when the browsing number evaluation value and the display probability evaluation value are calculated, the display number evaluation value is obtained by multiplying them.

  In addition, the profitability evaluation value also varies depending on the profitability (CPM) per advertisement display for each media, as well as the number of advertisement impressions and the freshness of the articles. It is also possible to calculate the profitability evaluation value in consideration of the elapsed time since then, the number of times of advertisement display, and the like. For example, the profitability evaluation value may be calculated using an arithmetic expression including a parameter related to an elapsed period from when the target article is issued and a parameter related to the number of advertisement display times.

  As mentioned above, although this invention was demonstrated using the said embodiment, the technical scope of this invention is not limited to the range as described in the said embodiment. Various changes or improvements can be added to the above-described embodiment without departing from the gist of the invention, and embodiments to which the changes or improvements are added are also included in the technical scope of the present invention. Moreover, you may combine the said embodiment suitably.

  For example, in the above embodiment, the recommendation evaluation value is calculated using the display count evaluation value and the profitability evaluation value. However, the present invention is not limited to this example. For example, only the display count evaluation value, only the profitability evaluation value is used. It is good also as obtaining a recommendation evaluation value using. Moreover, it is good also as calculating a recommendation evaluation value only using a browsing number evaluation value, without using a display frequency evaluation value and a profitability evaluation value. As described above, the recommendation evaluation value may be calculated using at least one of the browsing number evaluation value, the display count evaluation value, and the profitability evaluation value.

  Further, when calculating the display count evaluation value, it is also possible to further predict the advertisement display count in a predetermined period from the issue of the article by further using the competition scale evaluation value related to the competition scale for the advertising space. For example, in recent years, there is a system called RTB (Real Time Bidding) in which a plurality of advertisers bid for one advertising space, and the advertiser with the highest bid amount acquires the advertising space. . In the case of a medium that provides an article provided with an advertising space using such a bidding method, it is necessary to predict the number of times the advertisement is displayed in consideration of the magnification of one advertising space. Therefore, taking into account articles with such bid-type advertising space, calculate the competitive magnification evaluation value for evaluating the competitive magnification of the advertising space in each medium, and further use this competitive magnification evaluation value to evaluate the number of display times. The value may be calculated.

  Further, in each of the above-described embodiments, the display count evaluation value is calculated by multiplying the browsing count evaluation value and the display probability evaluation value. It is good also as estimating the browsing total from the present time to the end of a predetermined period using a track record. In this case, for example, the number of display times in the past predetermined period of the issuer media of the target article is recorded per unit time, and among these display number results, the degree of similarity with the display number result of the target article is high, for example, Alternatively, articles having a degree of similarity greater than or equal to a predetermined value may be extracted, and the display count evaluation value may be calculated using the display count results of those articles.

  Further, in each of the above-described embodiments, the predicted cumulative number of views, the predicted advertisement display count, and the predicted revenue in a predetermined period from the publication of the article of the target article are calculated. Need not be. For example, the start point of the predetermined period may be a predetermined time after the publication of the article of the target article, the start point of the predetermined period may be set between the issue of the article and the present, or the start point of the predetermined period in the future from the present May be set. Further, when the start point of the predetermined period is set in the past from the present time, it is possible to use the actual value as the cumulative number of browsing from the start point to the present.

DESCRIPTION OF SYMBOLS 1 Advertisement delivery system 10 Media server 11 Advertiser server 12 Advertisement delivery server 13 User terminal 15 Network 20 Advertisement granting device 31 Article database 32 Article advertisement database 33 Article selection part 34 Recommendation advertisement determination part 41 Calculation part 42 Selection part

Claims (14)

An article selection means for calculating a recommendation evaluation value for each of a plurality of articles, and selecting an article candidate to give a recommendation advertisement that is an advertisement related to the content of the article based on the recommendation evaluation value;
A recommended advertisement determining means for determining a recommended advertisement related to the posted content of the article for all or a part of the article candidates selected by the article selecting means,
The article selection means is:
The article evaluation of the target article for which the recommendation evaluation value is calculated or the number-of-views evaluation value related to the predicted cumulative browsing number from a predetermined time after the article issuance, from the predetermined time after the article issuance or article issuance for the predetermined period The recommendation evaluation of the target article using at least one of the display frequency evaluation value related to the predicted advertisement display frequency and the profitability evaluation value related to the predicted profit from a predetermined time after the article issuance or post-issue of the target article. A computing means for computing a value;
Selection means for selecting a candidate article to which the recommended advertisement is to be assigned from among the plurality of articles based on the recommendation evaluation value of each of the target articles calculated by the calculation means;
An ad grant device comprising:   The advertisement granting apparatus according to claim 1, wherein the calculation unit calculates the browsing number evaluation value of the target article by using browsing number actual data of articles published in the past by the target article publisher.   The computing means computes the browse number evaluation value using the browse number actual data related to the category to which the target article belongs, which is the browse number actual data of articles published in the past by the publishing medium of the target article. The advertisement giving apparatus according to claim 1. The calculation means has browsing number information in which the browsing number evaluation value is associated with each media and each category of articles,
The advertisement granting apparatus according to claim 1, wherein the browsing number evaluation value associated with a category of the target article that is an issuer medium of the target article is acquired from the browsing number information.   The calculation means calculates a display probability evaluation value using display probability performance data of an advertisement in the medium that issues the target article, and displays the target article using the display probability evaluation value and the browsing number evaluation value. The advertisement giving apparatus according to any one of claims 1 to 4, wherein a number evaluation value is calculated.   The computing means computes the display count evaluation value using a competition magnification evaluation value related to a competition magnification with respect to an advertisement space in the publisher media of the target article, the display probability evaluation value, and the browsing number evaluation value. Item 6. The advertisement giving device according to Item 5.   The advertisement granting apparatus according to any one of claims 1 to 6, wherein the computing means computes the profitability evaluation value using profit performance data in a media from which the target article is issued.   The said calculating means is the profit performance data of the issuer | publisher media of the said target article, Comprising: The said profitability evaluation value is calculated using the said profit performance data regarding the category to which the said target article belongs. The advertisement provision apparatus in any one. The computing means has profitability information in which the profitability evaluation value is associated with each media and each category of articles,
The advertisement granting apparatus according to any one of claims 1 to 6, wherein the profitability evaluation value associated with a category of the target article, which is a publisher media of the target article, is acquired from the profitability information.   The calculation means analyzes the display tendency of the advertisement of the target article from the actual value of the number of advertisements displayed for a certain period from the issue of the target article to the present, and a plurality of items issued from the publisher media of the target article From the article, the article having the display tendency of the target article and a similarity equal to or higher than a predetermined value is extracted, and the extracted article display count result data is used to display the target article from the current time to the end of the predetermined period. The advertisement granting apparatus according to claim 1, wherein the display count is estimated by calculating the display count evaluation value.   The computing means analyzes the article browsing tendency of the target article from the actual value of the number of article browsing per fixed period from the issue of the target article to the present time, and a plurality of items issued from the publisher media of the target article From the article, an article having an article browsing tendency of the target article and a similarity equal to or higher than a predetermined value is extracted, and using the extracted article browsing number result data, from the current time in the target article to the end of the predetermined period The advertisement granting apparatus according to claim 1, wherein the browsing number evaluation value is calculated by predicting the browsing number.   The calculation means calculates the profitability evaluation value using an arithmetic expression including at least one of an elapsed time from the issue of the target article and a display count evaluation value of the target article as a parameter. The advertisement granting apparatus according to claim 11. An advertisement granting method executed by a computer,
An article selection step of calculating a recommendation evaluation value for each of a plurality of articles, and selecting a candidate article for giving a recommendation advertisement that is an advertisement related to the content of the article based on the recommendation evaluation value;
A recommended advertisement determining step for determining a recommended advertisement related to the posted content of the article for all or some of the article candidates selected by the article selecting step;
The article selection process includes:
The article evaluation of the target article for which the recommendation evaluation value is calculated or the number-of-views evaluation value related to the predicted cumulative browsing number from a predetermined time after the article issuance, from the predetermined time after the article issuance or article issuance for the predetermined period The recommendation evaluation of the target article using at least one of the display frequency evaluation value related to the predicted advertisement display frequency and the profitability evaluation value related to the predicted profit from a predetermined time after the article issuance or post-issue of the target article. A calculation step for calculating a value;
A selection step of selecting candidate articles to which the recommended advertisement is to be assigned from among the plurality of articles based on the recommendation evaluation value of each of the target articles calculated in the calculation step;
An advertisement granting method comprising: An article selection process for calculating a recommendation evaluation value for each of a plurality of articles, and selecting an article candidate to give a recommendation advertisement that is an advertisement related to the posted content of the article based on the recommendation evaluation value;
Advertisement granting program for causing a computer to execute a recommendation advertisement determination process for determining a recommendation advertisement related to the posted content of the article for all or some of the article candidates selected in the article selection process Because
The article selection process is:
The article evaluation of the target article for which the recommendation evaluation value is calculated or the number-of-views evaluation value related to the predicted cumulative browsing number from a predetermined time after the article issuance, from the predetermined time after the article issuance or article issuance for the predetermined period The recommendation evaluation of the target article using at least one of the display frequency evaluation value related to the predicted advertisement display frequency and the profitability evaluation value related to the predicted profit from a predetermined time after the article issuance or post-issue of the target article. A calculation process for calculating a value;
A selection process for selecting candidate articles to which the recommended advertisement is to be assigned from among the plurality of articles based on the recommendation evaluation value of each of the target articles calculated in the calculation process;
An ad grant program that includes

Images