JP6436951B2 - ADVERTISEMENT APPARATUS, ADVERTISEMENT APPARATUS PROCESSING METHOD, AND PROGRAM - Google Patents

Description

  The present invention relates to an advertisement giving device, a processing method for the advertisement giving device, and a program.

  Nowadays, articles posted on websites play a major role as advertising media for many companies. Therefore, in order to convey the contents of their products and services more correctly, to acquire more customers, and to increase customer satisfaction, many companies have the contents of articles posted on their websites. Wants to optimize advertisements that direct viewers to their website.

  In general, access analysis is important for optimizing articles and advertisements posted on a website. Access analysis is based on data such as the number of articles viewed on a website (the number of page views, hereinafter also referred to as the number of PV), the presence or absence of specific actions by viewers of each article, etc. This refers to a method for improving the configuration of articles to be posted, the contents of articles posted on a Web page, advertisement contents, and the like. There are many tools for performing this access analysis. As a general one, for example, there is an access analysis service called Google (registered trademark) Analytics (Google Analytics) provided free of charge online by Google.

  In Google Analytics, it is possible to acquire various data necessary for improving the efficiency of the net business, such as measurement of advertising effectiveness, confirmation of movement history within the user's site, and search keyword analysis. Therefore, by using such an access analysis service, you can understand the visitor's desire by grasping the trend of the site visitor, know the popular and unpopular articles in the site, the problematic articles, It is expected to improve visitor satisfaction and increase the number of views (number of PV) by improving. It also contributes to measures against SEO (Search Engine Optimization: a technique for configuring a Web page so that it appears higher in search results), which is a technique for obtaining a higher-level display at a search site.

  Then, when an advertisement related in content to an article browsed by the user is given, an advertisement related to the article is estimated and given. For example, in this case, a keyword included in the article body is used. The advertisement server is used to perform a search, and an advertisement to be given is determined therefrom. In particular, in the case of a configuration in which an advertisement server is connected via the Internet, the processing load on the advertisement server may be a problem.

  That is, since it is not possible to search for the keywords of all articles, it is not possible to give advertisements to all articles. Therefore, when there are an enormous number of articles, it is necessary to limit which article the advertisement is given to. In other words, it is necessary to determine which articles should be given advertisements. Although the similarity based on the keyword has been described here, the same applies to the case where the similarity of the browsing user layer is searched for the advertisement server and processed.

  In addition, there are advertisements that are limited in duration, and for example, there are advertisements that have a validity period of only one day, so that it is necessary to search for advertisements for all articles every day. In addition, even if the article is a past article, if a new article related to the past article is read, the past article may be read again accordingly. And there will be an infinite number of articles. For this reason, the processing load problem of the advertisement server is likely to become obvious.

  Thus, since the number of articles to be searched is enormous, it is impossible to give an advertisement to all articles in the first place. Also, simply because the number of browsing is large, it does not mean that an advertisement should be given, and depending on the content of an article, it may not be appropriate to give an advertisement in the first place.

  In Patent Document 1, in order to estimate with high accuracy what information the user is really interested in based on the lifestyle and behavior pattern of the user of the Internet site, the advertisement information provided to the user is When provided on the display screen, a click expectation value that is a numerical value indicating an index of the probability that the detailed information of the advertisement information is browsed is adopted as the priority, for example, the advertisement information with the first and second priority A technique for providing only the user with the information is disclosed.

  Further, Non-Patent Document 1 uses a Google Analytics API (Application Programming Interface) API to obtain information on a session (a series of operations and communications from connection / login to disconnect / logoff) on a daily Web page. A method of sorting extracted data in descending order of the number of page views and analyzing client session trends is disclosed.

JP 2008-097545 A

Takanori Kurata, “Useful for advanced analysis! Introducing frequently used axes / indexes and segments in Google Analytics API” [online], December 15, 2015, Web person Forum, [July 25, 2016 search ] Internet <URL: http://web-tan.forum.impressrd.jp/e/2015/12/15/21593>

  However, the technique described in Patent Document 1 stores the search site accessed by the user, the search keyword, the product information of the purchased product in the product purchase procedure, and various personal information in the advertisement information management server in association with the identification information of the user terminal. Based on the user history information accumulated in the advertisement information management server, the advertisement information that should be provided to the user terminal is extracted, and the advertisement information that is expected to exert a high advertising effect on the user, for example, In the advertisement management server, it is possible to preferentially display the advertisement information estimated to have a high click expectation value on the user terminal and adopt a configuration according to the processing load without using the Internet. No problem of processing load such as the load problem of the advertisement server as described above occurs.

  In addition, in the technique described in Non-Patent Document 1, it is conceivable to give an advertisement preferentially from higher-level session information among session information sorted in descending order of the number of page views. On the other hand, when an advertisement is given, if the number of sessions is large, there is a problem that the processing load of the advertisement giving server increases and the processing takes time.

  Furthermore, when a keyword is extracted from an article, an advertisement server is searched based on the keyword, and an advertisement corresponding to the article is determined, the time and the number of times required to perform a search using the advertisement server are limited. At the same time, when the size of the advertisement information such as video is large, the problem is that the optimal advertisement cannot be given to all articles because the communication line with the advertisement server is limited. easy.

  Therefore, the present invention has been made in view of the above-described conventional problems, and is capable of easily estimating which article should have an optimal advertisement with a small processing load. It is an object to provide a processing method and a program.

  In order to solve the above-mentioned problem, the advertisement granting apparatus according to the present invention described in claim 1 is an advertisement granting apparatus that grants an advertisement to a predetermined article from among a plurality of articles viewed by a user. A number of browsing means for acquiring the number of browsing that each of a plurality of articles is browsed by the user in a predetermined period, and whether to give an advertisement to the predetermined article, depending on the number of browsing An advertisement granting determination unit that determines probabilistically in response, and an advertisement granting process execution unit that executes a process of granting an advertisement to the predetermined article determined to be given an advertisement by the advertisement granting determination unit It is characterized by.

  Moreover, in order to solve the said subject, the advertisement provision apparatus in this invention of Claim 6 is an advertisement provision apparatus which provides an advertisement with respect to a predetermined article from the several articles browsed by the user. , A click number acquisition means for acquiring the number of clicks clicked in a predetermined period by the user for each of the advertisements assigned to the plurality of articles, and whether or not to give an advertisement to the predetermined article. , An advertisement grant determining unit that probabilistically determines according to the number of clicks, and an ad grant process for executing a process of granting an advertisement to the predetermined article determined to be given an advertisement by the ad grant determining unit Execution means.

  Furthermore, in order to solve the said subject, the processing method of the advertisement provision apparatus in this invention of Claim 7 WHEREIN: The advertisement provision apparatus which provides an advertisement with respect to a predetermined article from the several articles browsed by the user A method of obtaining the number of browsing that each of the plurality of articles is browsed in a predetermined period by the user, and whether to give an advertisement to the predetermined article, Including a step of probabilistically determining according to the number of viewings, and a step of executing a process of giving an advertisement to the predetermined article determined to give an advertisement by the step of probabilistically determining It is characterized by.

  In order to solve the above problem, a program that is executed by the computer of the advertisement granting device according to the present invention described in claim 8 grants an advertisement to a predetermined article from among a plurality of articles viewed by the user. Whether or not to give an advertisement to the predetermined article, a process of acquiring the number of browsing that each of the plurality of articles is browsed by the user in a predetermined period, and a computer of the advertisement granting device, A process that executes a process that probabilistically determines according to the number of viewings, and a process that executes a process that assigns an advertisement to the predetermined article that is determined to be given an advertisement by the process that determines the probability It is characterized by that.

  ADVANTAGE OF THE INVENTION According to this invention, the advertisement provision apparatus, the processing method of an advertisement provision apparatus, and a program which can estimate easily which articles should be provided with the optimal advertisement with a small processing load are obtained.

It is a figure which shows the whole structure of the advertisement provision system containing the advertisement provision server in this embodiment. It is a figure explaining an example of the hardware constitutions of the media server which comprises the advertisement provision system in this embodiment, an advertisement provision server, an advertiser (advertisement agency) server, and a user terminal. It is a block diagram explaining an example of the function which the advertisement grant server in this embodiment has. It is a figure explaining the process performed in the advertisement provision process target memory | storage part which the advertisement provision server in the advertisement provision system on the conventional extension line has. It is a figure explaining the process performed while it is sent to the advertisement provision process target memory | storage part from the advertisement unassignment memory | storage part which the advertisement provision server in the advertisement provision system in this embodiment has. It is a flowchart explaining the operation | movement which the advertisement provision server in the advertisement provision system in this embodiment calculates prediction PV and measurement PV. It is a flowchart explaining the operation | movement which the advertisement provision server in the advertisement provision system in this embodiment calculates prediction PV and measurement PV. It is a flowchart explaining the operation | movement which the advertisement provision server in the advertisement provision system in this embodiment calculates prediction PV and measurement PV. It is a flowchart explaining the operation | movement which the advertisement provision server in the advertisement provision processing system in this embodiment performs advertisement provision with respect to an article.

  Next, embodiments for carrying out the present invention will be described in detail with reference to the drawings. In addition, in each figure, the same code | symbol is attached | subjected to the part which is the same or it corresponds, The duplication description is simplified thru | or abbreviate | omitted suitably. Hereinafter, although this embodiment is described, the present invention is not limited to the embodiment described below. In addition, although embodiment described below has mentioned and demonstrated the advertisement provision system by which the advertisement provision server, the advertiser (advertisement agency) server, and the user terminal were each comprised by 1 unit | set, as a specific example. The advertisement granting server, the advertiser (advertisement agency) server, and the user terminal can be applied to an advertisement granting system including a plurality of each. In the following description, the target advertisement is not only a still image advertisement or a video advertisement displayed on the display screen of the user terminal in association with an article viewed using a web browser on the user terminal. This includes banner advertisements linked from

  First, the overall configuration of an advertisement granting system including an advertisement granting server in the present embodiment will be described. FIG. 1 is a diagram showing an overall configuration of an advertisement granting system including an advertisement granting server in the present embodiment.

  In FIG. 1, media servers 101, 102,..., 10N (hereinafter also simply referred to as media server 100), advertisement grant server 200, advertiser (advertisement agency) server 300, and user via network 400 Terminals 500 are connected so as to communicate with each other. The network 400 will be described using an Internet line as a specific example, but may be another network line.

  The media server 100 is a server possessed by mass media such as newspapers, magazines, radio broadcasts, television broadcasts and the like, which is a means for transmitting information to a large number of unspecified recipients, and a server possessed by a portal site. Has been uploaded. When receiving a request for browsing an article using the Web browser function of the user terminal 500 from the user terminal 500 via the network 400, the media server 100 sends an article for which the browsing request is made via the network 400 to the user terminal. Deliver to 500.

  The advertiser (advertising agency) server 300 is a server owned by an advertiser such as a corporation that provides a consumer with products, services, etc., or a product of the corporation in response to a request from the corporation. , A server of an advertising agency that advertises services. The advertiser (advertisement agency) server 300 uploads and stores various advertisements related to products and services of advertisers such as corporations.

  The advertisement granting server 200 is the core of the advertisement granting system in the present embodiment. When receiving a browsing request from the user terminal 500 for a predetermined article uploaded to the media server 100, the advertisement granting server 200 is optimal for the article. The server acquires an advertisement from the advertiser (advertising agency) server 300 and transmits the advertisement to the user terminal 500 together with the article viewed by the user 550. The determination of whether or not is performed in a probabilistic manner according to the number of views of the article. This point will be described later.

  The user terminal 500 accesses the media server 100 via the network 400, and browses an article uploaded to the media server 100 using its own Web browser. The user 550 browses the article. Display means for operation, operation means for operation by the user 550, and the like.

  The media server 100, the advertisement granting server 200, and the advertiser (advertisement agency) server 300 may be any device having any information processing function as long as it has the above-described functions. Further, the user terminal 500 has been described by taking a personal computer (hereinafter also referred to as a PC) as a specific example, but any device having any information processing function may be used as long as it has the above-described functions. good.

  Next, an example of the hardware configuration of the media server, the advertisement grant server, the advertiser (advertisement agency) server, and the user terminal that configure the advertisement grant system in the present embodiment will be described. FIG. 2 is a diagram illustrating an example of a hardware configuration of a media server, an advertisement grant server, an advertiser (advertisement agency) server, and a user terminal that configure the advertisement grant system according to the present embodiment.

  In FIG. 2, the server functioning as the media server 100, the advertisement granting server 200, and the advertiser (advertisement agency) server 300 and the PC functioning as the user terminal 500 in this embodiment are a server or a PC as a hardware configuration. A CPU (Central Processing Unit) (control unit) 401 is provided for controlling the overall operation. In addition, a storage unit such as a ROM (Read Only Memory) 402 and a RAM (Random Access Memory) 403 that stores various data and various programs is provided. Further, a power supply unit 407 that supplies power to the server or the PC is provided.

  In addition, the server or PC user has a function of operating the server or PC such as a keyboard or a touch panel composed of an LCD (Liquid Crystal Display), and displays data stored in the ROM 402 and RAM 403 of the server or PC. An operation / display unit 405 having a function is provided. The server or PC includes a communication unit 406 that performs communication with a server or PC provided outside via the network 400. Furthermore, an HDD (Hard Disk Drive) 404 for storing application programs installed in the server or PC is provided.

  The CPU 401, ROM 402, RAM 403, HDD 404, operation / display unit 405, communication unit 406, and power supply unit 407 are connected by a bus 408. Various processes in the server or the PC are realized by the CPU 401 loading and executing various programs stored in the ROM 402. However, the present invention is not limited to this, and these functions can also be realized by using circuits such as individual hardware.

  Next, an example of the function of the advertisement giving server in this embodiment will be described. FIG. 3 is a block diagram for explaining an example of the function of the advertisement granting server in the present embodiment.

  In FIG. 3, the media server communication interface unit 201 performs communication with the media server 100 via the network 400. The advertiser (advertisement agency) server communication interface unit 202 performs communication with the advertiser (advertisement agency) server 300 via the network 400. Communication between the media server communication interface unit 201 and the advertiser (advertisement agency) server communication interface unit 202 and the network 400 may be wired or wireless. In the following description, it is assumed that the period of advertisements given to articles is one day (24 hours).

  The article browsing number acquisition unit 203 acquires the number of browsing when the user 550 accesses the media server 100 and browses an article uploaded to the media server 100 using the Web browser of the user terminal 500. Specifically, the number of articles viewed on the previous day (number of page views, hereinafter referred to as the number of PV) is acquired, and information on whether or not an article to which no advertisement has been given as of today has been acquired. To do. Details will be described later.

  The advertisement grant lower limit PV number determination unit 204 determines the minimum PV number (advertisement grant lower limit PV number) of articles to which an advertisement can be given. For example, if the number of articles that can be given an advertisement (the number of articles that can be given an advertisement) is A, and the number of articles that can be given an advertisement after the start of giving an advertisement to the article is a, the number of articles Among the articles sorted in descending order by the number of PV of the previous day's articles acquired by the browsing number acquisition unit 203, the number of PV of the Ath article (the number of articles that can be given advertisements) from the article having the highest number of PV is B. As (PV), this is determined as the advertisement grant lower limit PV number.

  The predicted PV number calculation unit 205 uses the average attenuation rate of articles other than articles newly uploaded to the media server 100 today (hereinafter also referred to as new articles) as the predicted browsing number (hereinafter also referred to as predicted PV number). Is to be calculated. That is, the predicted number of browsing refers to the number of browsing predicted that articles other than new articles will be browsed by the user 550 in the future. Since the period of advertisements given to articles is one day, the ratio of the total number of PVs of the previous day to the total number of PVs of the previous day is obtained for a certain article ((total number of PVs of the previous day / the day before the previous day). The total number of PV) * 100), and this is the predicted PV (C%). Note that the above-described advertisement grant lower limit PV number (B (PV)) and predicted PV (C%) may be used for a certain period of time once calculated, and need not be calculated every day.

  The predicted PV correction unit 206 corrects the calculated predicted PV (C%). For example, the predicted PV of an article uploaded up to the previous day (hereinafter, also referred to as an old article) is the number of actually viewed views of the old article (hereinafter, also referred to as the actual PV number) calculated on the previous day calculated by an actual PV number calculation unit 207 described later. C%), that is, when the previous day is today, the number of PVs in the old article is normally C% attenuated from the actual PV of the previous day.

  However, if the previous day's actual PV was less than or equal to the predicted PV (C%), it was determined that the original number of predicted PVs was too high, and it was down to about 90% of the predicted PV (C%). Make corrections to correct. In addition, when the article that did not give the previous day advertisement has the number of PVs to which the previous day advertisement should be given, that is, when the actual PV exceeds the predicted PV (C%), the original predicted PV number is low. It is determined that the amount is too high, and correction is made to correct upward to about 120% of the predicted PV (C%).

  The actual PV number calculation unit 207 measures and stores the number of PVs of a new article and an old article in a current predetermined period. The actual PV calculated by the actual PV number calculation unit 207 is used by the prediction PV correction unit 206 and the advertisement assignment determination unit 208 described later. The advertisement assignment determination unit 208 determines whether to give an advertisement to an article, according to the number of article views acquired by the article view number acquisition unit 203, specifically, the number of article views. It will be decided. This probabilistic determination method refers to whether or not an advertisement is given to each article according to the number of articles viewed acquired by the article browsing number acquisition unit 203. Based on the predicted PV (C%) calculated by the unit 205 and the actual PV calculated by the actual PV number calculation unit 207, a process of determining probabilistically is performed. This point will be described later.

  The advertisement giving process execution unit 209 executes a process for giving an advertisement to an article determined to be given an advertisement by the advertisement grant determining unit 208. In the advertisement granting system on the conventional extension described below, the advertisement granting processing target storage unit 210 is a process for granting an article to which an advertisement has not yet been given (hereinafter also referred to as an advertisement-unassigned article). In the advertisement granting system according to the present embodiment described below, an article determined to give an advertisement to the article by the advertisement granting determination unit 208 (hereinafter referred to as advertisement granting). Is also referred to as a target article.) Is stored as a target article that performs a process of giving an advertisement.

  The advertisement non-assignment storage unit 211 stores an article to which an advertisement has not yet been assigned (hereinafter also referred to as an advertisement non-attachment article) as an advertisement non-attachment article in the advertisement assignment system according to the present embodiment described below. It is a storage unit newly provided in the advertisement grant system in the present embodiment. Further, the overall control unit 212 controls the entirety of each block constituting the advertisement granting server 200.

  The media server communication interface unit 201, the advertiser (advertisement agency) server communication interface unit 202, the article browsing number acquisition unit 203, the advertisement grant lower limit PV number determination unit 204, the predicted PV number calculation unit 205, and the predicted PV correction unit 206 The measured PV number calculation unit 207, the advertisement assignment determination unit 208, the advertisement assignment process execution unit 209, the advertisement assignment process target storage unit 210, and the advertisement non-attachment storage unit 211 are connected by a bus 215.

  Next, the process performed in the advertisement provision process target memory | storage part which the advertisement provision server in the advertisement provision system on the conventional extension line has is demonstrated. FIG. 4 is a diagram for explaining processing performed in the advertisement grant processing target storage unit of the advertisement grant server in the advertisement grant system on the conventional extension line.

  As described above, since the new article is estimated to have the highest probability of being read among the articles uploaded to the media server 100, an advertisement has been given for the time being. Of the articles uploaded to the media server 100, articles other than the new article are sorted in descending order by the number of PVs on the previous day of the same article, from the highest PV number, that is, from articles that are often read. It was supposed to give advertisements in order.

  Therefore, even if a past article with a low PV number until the previous day is read today and the PV number suddenly rises, the increase in today's PV number is not reflected until the next day. Advertising is not given. In other words, the PV number of an article the day before was very small, but when the number of PV increased when it was read suddenly the day before, an advertisement was given to the article today because the PV number of the previous day is large. Will be. Therefore, compared with the actual increase in the number of PV, the advertisement is slightly delayed for the article.

  In FIG. 4, when a new article is uploaded to the media server 100, the new article PV = 1. Since this new article is an article to which no advertisement is attached, it means that the number of PV of the article is gradually increased when stored in the advertisement assignment processing target storage unit 210. In other words, the advertisement grant processing target storage unit 210 stores the article ID and the actual PV calculated by the actual PV number calculation unit 207 in association with each other. And the process which provides an advertisement with respect to the article which has the maximum number of PV among the articles stored in the advertisement provision process object memory | storage part 210 after time T was performed.

  Then, when the advertisement giving process ends, the association between the article having the maximum number of PVs stored in the advertisement giving process target storage unit 210 and the article ID of the article is deleted. In addition, all of the tables 450 in which article IDs and measured PVs stored in the advertisement assignment process target storage unit 210 are periodically associated with each other at the timing of starting the process of assigning advertisements to articles having the maximum number of PVs. Is deleted. For example, assuming that the advertisement cycle in the conventional advertisement grant system on the extension line is one day (24 hours), the article ID and the measured PV stored in the advertisement grant processing target storage unit 210 are obtained at a predetermined time every day. All the associated tables 450 are deleted.

  As described above, in the conventional advertisement granting system on the extension line, if the article is an article to which no advertisement is attached at the time when the article is browsed, the article may be an article to be given an advertisement. Further, after the time T, it is only necessary to give advertisements in descending order from articles having the maximum number of PV, but it is also necessary to perform processing that reduces the processing load of the advertisement assignment server 200.

  However, there are articles (hereinafter also referred to as “instant high PV articles”) in which the number of PV drops immediately, although the number of PV increases instantaneously. Then, since it is meaningless to give an advertisement to an article that is no longer read after time T (for example, when the peak of the PV number of the instantaneous high PV article has passed after time T), For example, if the storage state of the advertisement grant processing target storage unit 210 is initialized periodically (for example, every hour) and the interval of the time T is shortened (for example, 1 minute), the instantaneous high PV article It becomes possible to give an advertisement to.

  However, some articles can earn only a few PVs every day but have a certain number of PVs over a long period of time (hereinafter also referred to as long seller articles). Then, if the interval of time T is shortened as described above, an advertisement can be given to the instantaneous high PV article, but an advertisement cannot be given to the long seller article. That is, there is a case where it is desired to give an advertisement to an article having a certain number of PVs, and how to set the time T becomes a problem.

  For example, if the advertisement is given when the PV number reaches 10,000 by increasing the time T, the user will be allowed to view the article without giving the advertisement until the PV number reaches 10,000. The effect of the advertisement cannot be obtained. In other words, if the time T is long, the effect of the advertisement cannot be obtained. Therefore, if it is originally intended to give an advertisement when a certain number of PVs are measured, as described above, the PV number measurement time is shortened. Then, it becomes impossible to give an advertisement to a long seller article. In other words, if the PV measurement time is short, articles other than the instantaneous high PV article are not given advertisements and are discarded.

  Therefore, if the measurement time of the number of PV is long, the number of articles discarded without giving an advertisement increases, and the timing for picking up an article to which an advertisement should be originally given is lost. If it does so, an advertisement can be given to a long seller article, but if a momentary high PV article continues, most of these articles will be thrown away without being given an advertisement.

  As described above, in the conventional advertisement grant system on the extension line, by varying the time T, for example, since the 100 PV number was earned per hour, the advertisement was given, and because the 100 PV number was earned for 30 minutes, the advertisement The number of PVs to which advertisements are given varies due to load fluctuations processed by the advertisement granting server 200. That is, there is a problem that the determination process for determining whether or not to provide an advertisement to an article, that is, selecting whether or not to provide an advertisement depending on the number of PVs is affected by external factors such as time T. It is.

  Next, a process performed while being sent from the advertisement non-assignment storage unit included in the advertisement assignment server in the advertisement assignment system in the present embodiment to the advertisement assignment processing target storage unit will be described. FIG. 5 is a diagram illustrating a process performed while the advertisement granting server in the advertisement granting system according to the present embodiment has a non-assignment storage unit sent to an advertisement granting process target storage unit.

  In FIG. 5, if an article is an article to which an advertisement has not yet been given (hereinafter also referred to as an advertisement-unassigned article), the article is stored in the advertisement-unassigned storage unit 211. Then, the advertisement non-assignment storage unit 211 stores a table 650 in which the article ID of the article to which no advertisement is attached, the actual PV of the article, and the predicted PV are associated with each other. Here, the article ID is an identifier for identifying each of the advertisement unassigned articles. The actual PV is calculated by the actual PV number calculation unit 207 of the actual PV of the advertisement unassigned article corresponding to the article ID. The predicted PV is calculated by the predicted PV number calculation unit 205 for the predicted PV of the advertisement unassigned article corresponding to the article ID. As described above, the prediction PV is calculated by (total number of PV on the previous day / total number of PV on the previous day) * 100 = C%.

  Next, a process performed by the processing unit (advertisement grant determining unit) 250 of the advertisement grant server 200 will be described. The processing unit (advertising grant determination unit) 250 grants an advertisement to the non-adverted article every time the non-advertised article stored in the unadvertised storage unit 211 is viewed by the user. Whether to do this is determined based on the predicted PV stored in the table 650.

  At this time, it is estimated that the advertisement unassigned article whose predicted PV exceeds the predetermined value will continue to be browsed and earn the number of PV. Therefore, since it is considered that an advertisement should be given with a high probability, the advertisement should be stored in the advertisement grant processing target storage unit 210 with a high probability.

  That is, the probability that it is determined that an advertisement will be given increases as the advertisement PV is sufficiently large compared to the advertisement grant lower limit PV number (B (PV)). Thus, even if the measured PV is an early stage below a predetermined value, the advertisement is stored in the advertisement grant processing target storage unit 210 with a high probability and an advertisement is given.

  Furthermore, an advertisement-unassigned article with a large number of actual PVs can be repeatedly probabilistically determined using the actual PV even if the predicted PV is below the advertisement grant lower limit PV number (B (PV)). As a result, the probability that an advertisement is given increases. Therefore, by performing the repeated probabilistic judgment process using the measured PV, the advertisement is stored in the advertisement giving process target storage unit 210 with a high probability, and the advertisement is given.

  In other words, in the present embodiment, it is determined whether to add an advertisement to an article in consideration of the actual PV in addition to the predicted PV. The measured PV is, for example, the speed within a certain time, and the predicted PV can be compared to the distance that the accumulated PV of today will be about this. In other words, the predicted PV is based on the concept that the number of browsing was large on the previous day, so it is likely to be large today, or that much PV will be earned today. Note that the actual PV is preferably cleared within a certain period of time, such as being cleared from time to time, or always counting the number of browsing for a certain period of time.

  It has been described that the larger the actual PV and the predicted PV, the higher the probability, the higher the actual PV and the predicted PV are stored in the advertisement grant processing target storage unit 210. However, the actual PV and the predicted PV have a certain size or more. In such a case, a stochastic stepwise storage method may be used, such as storing from the advertisement non-assignment storage unit 211 to the advertisement assignment processing target storage unit 210 with a certain probability or more. If there is a prediction that it is estimated that a lot of PV will be earned today because it earns a lot of PV the previous day, it is stored in the advertisement grant processing target storage unit 210 from the advertisement non-assignment storage unit 211 with a high probability.

  When the predicted PV is small but the measured PV is large, the processing unit (advertising determination unit) 250 determines how many times to determine whether or not to store the advertising grant processing target storage unit 210 from the non-advertisement granting storage unit 211. Repeat repeatedly. That is, since the measured PV is high, if the advertisement has not been given, the process of adding to the advertisement non-assigned storage unit 211 is repeatedly executed. Then, the processing unit (advertising grant determination unit) 250 repeatedly determines whether to store, in the ad grant processing target storage unit 210, an unadvertised article that is repeatedly added to the non-advertisement storage unit 211. As a result, when the actual PV becomes large, the probability of being stored in the advertisement grant processing target storage unit 210 from the advertisement non-assignment storage unit 211 is relatively high.

  Accordingly, since only the articles that are determined to be given advertisements are stored in the advertisement granting processing target storage unit 210, the time T is set as in the conventional advertisement granting system on the extension line as described above. There is no need to go to the advertisement granting process target storage unit 210 later to get an article. In short, there is no need to wait for the article to be stored in the advertisement granting process target storage unit 210.

  Then, an article having a sufficiently large number of actually measured PVs as compared with the advertisement grant lower limit PV number (B (PV)) is stored in the advertisement grant processing target storage unit 210 with a high probability. Also, articles near the threshold value of the advertisement grant lower limit PV number (B (PV)) are rarely stored in the advertisement grant processing target storage unit 210, and are normally stored in the advertisement grant processing target storage unit 210. There is no.

  That is, an article with a sufficiently high predicted PV is easily stored in the advertisement grant processing target storage unit 210 at an early stage of browsing. On the other hand, an article with a lower predicted PV (an article near the threshold of the advertisement grant lower limit PV number (B (PV))) is difficult to be stored in the advertisement grant processing target storage unit 210 at an early stage from the start of browsing. Become. And about such an article with a lower predicted PV (an article near the threshold of the advertisement grant lower limit PV number (B (PV))), if there are many actual PVs, the decision to grant an advertisement is made many times. Become.

  Note that, as described above, the predicted PV may be appropriately corrected based on the measured PV. For example, there was a fact that an article that was predicted to be read 1000 times in 24 hours (predicted PV = 1000) was read 100 times in the first hour as the actual PV number gradually increased. In this case, it is determined that the predicted PV is incorrect, and for example, correction such as upward correction of the predicted PV to 1500 is performed.

  In the present embodiment, the advertisement granting determination unit 208 stores the advertisement grant target articles in the advertisement grant processing target storage unit 210 as long as the number of possible advertisement grants remains. That is, the advertisement giving process execution unit 209 can process at the maximum speed. That is, an advertisement may be given to all articles stored in the advertisement grant processing target storage unit 210. In short, being stored in the advertisement assignment processing target storage unit 210 means that an advertisement is always given to the article. Therefore, as long as it is stored in the advertisement grant process target storage unit 210, the advertisement grant process may be performed at the maximum speed.

  Therefore, the process of determining whether to give an advertisement to an article as in the above-described conventional advertisement giving system on the extension line, that is, selecting whether to give an advertisement based on the number of PV is time T The problem of being influenced by external factors is eliminated, and the speed at which advertisements are given to articles changes as compared with the conventional advertisement giving system on the extension line.

  However, while the actual PV is increased and the probability calculation of whether or not to store from the advertisement non-assignment storage unit 211 to the advertisement assignment processing target storage unit 210 is repeatedly performed, no advertisement is attached even if the article is read. As a result, PV that cannot obtain the effect of providing an advertisement will also appear, but the determination as to which article should be given an advertisement cannot be made.

  In this embodiment, when the predicted PV is much larger than the advertisement grant lower limit PV number (B (PV)), the advertisement is immediately before the actually measured PV starts to rise, for example, at an early stage such as several seconds. Stored from the unassigned storage unit 211 to the advertisement assignment processing target storage unit 210, an advertisement is given to the article.

  In addition, when the predicted PV and the advertisement grant lower limit PV number (B (PV)) are almost equal, the determination process such as whether or not to store the advertisement grant processing target storage unit 210 from the advertisement non-assignment storage unit 211 is performed many times. In the meantime, the advertisement grant processing target storage unit 210 will be stored. In short, the actual PV rises and is stored in the advertisement grant processing target storage unit 210 from the unadvertised storage unit 211 before it is certain that the article should be given an advertisement due to the increase in the actual PV. Is determined to be probabilistically high, an advertisement is given to the article. In this way, although there is not necessarily complete accuracy, it is possible to reduce the useless PV to which no advertisement is given within a range that is not a big mistake in many cases.

  For example, when the article posting period is 3 days (72 hours), the estimated PV (for example, one day) that will be read in the next 72 hours is estimated. Give. Next, if the predicted PV is sufficiently high as estimated from the past PV, it is predicted that the PV will be sufficiently high for the next 72 hours, so an advertisement is given to the article with a high probability. Next, although the past PV was not very high, for example, an article whose number of readings in the last minute has risen (the actual PV has risen) is given an advertisement to the article using the actual PV Whether to do so is determined stochastically, and how many times the trial is executed is determined based on the measured PV.

  In short, by using only the predicted PV and the actual PV, if the decision is suspended until it is determined whether or not to give the advertisement to the article, the article is browsed without the advertisement being given, and the effect of the advertisement is improved. Since PV that cannot be obtained is generated, it is avoided as much as possible, and the decision as to whether or not to add an advertisement to an article is made promptly and promptly.

  When the article is initialized after 72 hours, it returns to the state where no advertisement is given to the article again. In the long run, there is a continuing absence of advertisements for a large number of articles. If it is initialized 72 hours later, it will be judged whether it is better to give an advertisement in the next 72 hours at the timing when it is read again, or it is better not to give an advertisement because the PV number has already dropped. to continue.

  If the predicted PV is sufficiently higher than the advertisement grant lower limit PV number, the probability of storing from the advertisement unassigned storage unit 211 to the advertisement grant processing target storage unit 210 is determined by the value of the predicted PV (for example, if the forecast PV is 50%, It is stored in the advertisement grant processing target storage unit 210 from the advertisement non-assignment storage unit 211 once every two times.

  If the advertisement grant probability corresponding to the predicted number of PV is 0.1%, it is not easily stored in the advertisement grant processing target storage unit 210 from the advertisement non-assignment storage unit 211, but when viewed by a viewer many times. , And repeatedly calculating whether to store from the advertisement non-assignment storage unit 211 to the advertisement grant processing target storage unit 210, of which, from the advertisement non-assignment storage unit 211 to the advertisement grant processing target storage unit 210 Will come to be. In other words, although the calculation is based on the predicted PV, when the actual PV increases, the calculation process is repeated many times according to the actual PV number.

  For example, when the advertisement provision probability corresponding to the actual PV number is 0.1% and the actual measurement value is as predicted PV, the actual PV is small, so that the advertisement non-assignment storage unit 211 stores the advertisement provision process target storage unit 210. The calculation of whether or not to store is executed only occasionally. For example, since the calculation is performed only once per hour, it is difficult to easily store the advertisement from the non-advertisement storage unit 211 to the advertisement provision process target storage unit 210.

  However, unlike the predicted PV (0.1%), for example, the actual PV increases, and the calculation stored in the advertisement grant processing target storage unit 210 from the advertisement non-assignment storage unit 211 is calculated 100 times per hour. Then, the possibility of being stored in the advertisement grant processing target storage unit 210 from the advertisement non-assignment storage unit 211 is increased with a considerable probability.

  The advertisement unassigned storage unit 211 stores advertisement unassigned articles. When the measured PV = 5, the calculation of whether to store the advertisement from the unassigned storage unit 211 to the advertisement assignment process target storage unit 210 is executed five times. That is, a process for determining whether or not to store the advertisement granting target storage unit 210 from the advertisement non-assignment storage unit 211 by raising the probability that no advertisement given by subtracting the advertisement granting probability corresponding to the predicted number of PV from 1 to the fifth power Should be executed.

  In addition, the method of providing an advertisement using the advertisement providing system in the present embodiment is not limited to the advertisement providing server, and may be provided in a user terminal (client terminal). Because it is a user terminal (client terminal), an advertisement to be given to an article may be fetched from an advertiser (advertising agency) server when it is found that it has a certain probability.

  In addition, when the advertisement granting method using the advertisement granting system in the present embodiment is installed in a user terminal (client terminal), grouping may be performed for each genre to which an article belongs. For example, when the article browsed the previous day is related to sports news, the article to be browsed the next day is also calculated for the predicted PV and the measured PV for articles related to sports news. By doing in this way, an advertisement can be preferentially given to an article with high possibility of being interested for every user. On the other hand, when installed in the advertisement granting server, it is assumed that 1 million people browsed an article the previous day and 900,000 people read the article the next day. To ask.

  Next, the operation in which the advertisement giving server in the advertisement giving system in the present embodiment calculates the predicted PV and the measured PV will be described. 6, 7, and 8 are flowcharts for explaining operations in which the advertisement assignment server in the advertisement assignment system according to the present embodiment calculates the predicted PV and the measured PV.

  6, first, in the process of step S701, the advertisement grant lower limit PV number determination unit 204 of the advertisement grant server 200 sorts the previous day's PV numbers in descending order, and sets the A-th PV number that is the number of advertisement grants to B. Calculated as (PV) (advertising grant lower limit PV number).

  Next, in the process of step S702, the predicted PV number calculation unit 205 calculates (total PV number on the previous day / total PV number on the previous day) * 100 = C [%] for articles other than the new article, The predicted number of PV (average attenuation rate) is calculated. It is sufficient that steps S701 and S702 are calculated only once or once in several days.

  Next, in the process of step S703, the article browsing number acquisition unit 203 determines whether an article to which no advertisement is attached has been read. If it is determined that an unadvertised article has been read (step S703: Y), the process proceeds to step S704. If it is determined that an article to which no advertisement is attached has not been read (step S703: N), the process is terminated.

  Next, in the process of step S704, the advertisement assignment determination unit 208 determines whether or not to add an advertisement to an article to which no advertisement is attached based on the predicted PV calculated in the process of step S702. In the process of step S705, the article browsing number acquisition unit 203 determines whether or not an article to which no advertisement is attached is a new article. If it is determined that the article is a new article (step S705: Y), the process proceeds to step S706. If it is determined that the article is not a new article (step S705: N), the process proceeds to step S710.

  In the process of step S706, the predicted PV correction unit 206 adopts the maximum value, average value, etc. of the actual number of PVs of the new article on the previous day as the predicted PV. In the process of step S710, the advertisement assignment determination unit 208 determines whether or not the previous day advertisement has been provided. If it is determined that the previous day advertisement has been given (step S710: Y), the process proceeds to step S711. If it is determined that the previous day advertisement has not been made (step S710: N), the process proceeds to step S714.

  In the process of step S711, the actual PV number calculation unit 207 determines whether the previous day actual PV number is less than or equal to the predicted PV number. If it is determined that the actual PV number of the previous day is less than or equal to the predicted PV number (step S711: Y), the process proceeds to step S713, and the actual PV number of the previous day is not less than or equal to the predicted PV number (the actual PV number of the previous day is not the predicted PV number). If it is determined that the number exceeds, the process proceeds to step S712.

  In the process of step S712, the predicted PV correction unit 206 proceeds to the process of step S707 by regarding the predicted PV number as C% of the actual PV number of the previous day, that is, the ratio of the average attenuation rate. In the process of step S713, the predicted PV correction unit 206 corrects the predicted PV number downward to about 90% of the predicted PV number, and proceeds to the process of step S707.

  In the process of step S714, the advertisement assignment determination unit 208 determines whether or not the advertisement should have been given the previous day. If it is determined that the previous day advertisement should have been given (step S714: Y), the process proceeds to step S715. If it is determined that the previous day advertisement should not have been given (step S714: N), the process proceeds to step S712. To do.

  In the process of step S715, the predicted PV correction unit 206 upwardly corrects the predicted PV number to about 120% of the predicted PV number, and proceeds to the process of step S707. In the process of step S707, the predicted PV correction unit 206 compares the measured PV partial speed with the predicted PV partial speed. In the process of step S708, the predicted PV correction unit 206 determines whether correction of the predicted PV is necessary.

  In the process of step S708, if the predicted PV correction unit 206 determines that correction of the predicted PV is necessary (step S708: Y), the process proceeds to step S709, and correction of the predicted PV is not required (step S708: If it is determined as N), the process is terminated. In the process of step S709, the predicted PV correction unit 206 corrects the predicted PV and ends the process.

  Next, the operation in which the advertisement granting server in the advertisement granting processing system according to the present embodiment stores the advertisement non-assigned article from the advertisement non-assigned storage unit to the advertisement granting process target storage unit will be described. FIG. 9 is a flowchart illustrating an operation in which the advertisement granting server in the advertisement granting processing system according to the present embodiment grants an advertisement to an article.

  In FIG. 9, first, in the process of step S <b> 901, the article browsing number acquisition unit 203 of the advertisement assignment server 200 determines whether an article to which no advertisement is attached has been read. If it is determined that an article to which no advertisement is attached has been read (step S901: Y), the process proceeds to step S902. If the read article has already been given an advertisement (step S901: N), the process ends.

  In the process of step S902, the predicted PV number calculation unit 205 calculates an advertisement grant probability according to the predicted PV number of the article (article to which no advertisement is given). For example, the provision probability is obtained using a function that increases the provision probability as the number of predicted PVs increases. At this time, if the predicted PV is sufficiently larger than the advertisement grant lower limit PV number (B (PV)), a uniformly high probability is given, and if it is sufficiently small, a very small probability is uniformly given to grant the advertisement. In the case of the vicinity of the lower limit PV number (B (PV)), a high probability may be given only in the case of sufficient measured PV.

  In the process of step S903, the actual PV number calculation unit 207 measures the actual PV number. And the advertisement grant probability is corrected so that the grant probability increases as the actual PV increases. For example, the number obtained by subtracting the probability obtained in step S902 from 1 is raised to the power of the number of actually measured PVs, and subtraction from 1 is performed so that the advertisement assignment determination is performed by the amount of the actually measured PVs.

  In the process of step S904, a random number is generated, and advertisement assignment is determined from the random number and the advertisement assignment probability obtained in step S903. For example, if the advertisement grant probability is 1.5%, an integer random number between 1 and 1000 is generated. If it is determined that the advertisement is given if it is 15 or less, and if it is judged that no advertisement is given if it is 16 or more, good.

  The operation flow of the advertisement giving server in the present embodiment shown in FIGS. 6 to 9 can be executed by a program on a computer. That is, a CPU (control unit) 401 configuring the advertisement granting server 200 loads a program stored in the ROM 402. Then, each processing step of the program is performed sequentially.

  In addition, among the articles uploaded to the media server 100, an advertisement to be given to an article that has been decided to be given an advertisement extracts a keyword included in the article, and an advertiser (advertising agent) based on the keyword. It is obtained by searching the store 300 server. Further, the advertisement provided by the advertisement granting server 200 is acquired from an advertiser (advertisement agency) server 300 provided outside the advertisement granting server 200 that is its own device.

  As described above, in the present embodiment, the probability of storing from the advertisement non-assignment storage unit 211 to the advertisement assignment processing target storage unit 210 is determined by the value of the prediction PV. B (PV)) In order to determine whether or not to give an advertisement to articles below the vicinity, the probability that an advertisement will be placed in each article below the advertisement grant lower limit PV number (B (PV)) The calculation is executed for the number of times of actual measurement PV.

  That is, an advertisement should be given to each article below the number of advertisement grant lower limit PV (B (PV)) by the number of times proportional to the number of times actually viewed by the user, in other words, the number of times corresponding to the measured PV. The decision is made whether or not As a result, the process of determining whether to give an advertisement to an article, that is, the selection of whether to give an advertisement based on the number of PVs is not affected by external factors such as time T, and the instantaneous high PV Advertisements can be given to articles and long-seller articles.

  As described above, according to the present invention, it is possible to obtain an advertisement granting apparatus, a processing method for the advertisement granting apparatus, and a program capable of easily estimating which article should be given an optimal advertisement with a small processing load. It can be done.

  As mentioned above, although the embodiment of the present invention so far has been described based on the number of browsing of the article of the user, it may be determined by the number of advertisement clicks instead of the number of PV for each article. Depending on the type of advertisement, the effect is determined not by the number of times of display but by the number of times the viewer clicks on the advertisement (the number of clicks). In addition, there are articles that are difficult to be clicked on advertisements given from their contents. In the case of such an advertisement type, there may be cases where the number of PVs is large but the number of clicks is small, or vice versa. In that case, advertisements should be given to articles with more clicks than articles with more PVs. Furthermore, since the advertising unit price for each click is not constant, the advertising revenue obtained by multiplying the advertising unit price and the number of clicks may be used as a judgment material. In either case, it is possible to cope with a small modification of the method described so far.

  For example, assuming that when the average click rate is 1%, the click rate of articles with the number of views A is 2%, the article with the number of views A has a double click rate compared to the average click rate. Therefore, when compared with the same click rate, the number of browsing should be corrected by a factor of two. As described above, it is preferable to correct the predicted PV based on the results up to the previous day or the actually measured PV in a certain period of today using the ratio with the average click rate.

  As mentioned above, although embodiment of this invention has been described so far, embodiment of this invention is not limited to embodiment mentioned above. That is, other embodiments, additions, changes, deletions, and the like can be changed within the scope that can be conceived by those skilled in the art, and as long as the effects of the present invention are exhibited in any aspect, the scope of the present invention is included. It is included.

100, 101, 102,..., 10N Media server 200 Advertisement grant server 201 Media server communication interface unit 202 Advertiser (advertisement agency) server communication interface unit 203 Article browsing number acquisition unit 204 Advertisement grant lower limit PV number determination unit 205 Predicted PV number calculation unit 206 Predicted PV correction unit 207 Actual PV number calculation unit 208 Advertisement assignment determination unit 209 Advertisement assignment process execution unit 210 Advertisement assignment process target storage unit 211 Advertisement non-assignment storage unit 212 Overall control unit 215, 408 Bus 250 processing Part 300 Advertiser (advertising agency) server 400 Network 401 CPU (control part)
402 ROM
403 RAM
404 HDD
405 Operation / display unit 406 Communication unit 407 Power supply unit 450, 650 Table 500 User terminal 550 User

Claims (7)

An advertisement granting device that grants an advertisement to a predetermined article from among a plurality of articles viewed by a user,
A number-of-views acquisition means for acquiring the number of views viewed by each of the plurality of articles in a predetermined period by the user;
Whether or not to give an advertisement to the predetermined article, an advertisement grant determining means for probabilistically determining according to the number of browsing,
An advertisement giving process executing means for executing a process for giving an advertisement to the predetermined article determined to give an advertisement by the advertisement grant determining means;
Based on the past browsing number for each of the plurality of articles, a predicted browsing number calculating unit that calculates a predicted browsing number that is predicted to be viewed by the user in the future,
Only including,
The browsing number acquisition means determines whether or not an unadvertised article to which the advertisement is not attached has been read, and determines that the unadvertised article has been read. An advertisement granting apparatus for calculating an advertisement grant probability for giving an advertisement to an unadvertised article according to the predicted number of browsing of the unadvertised article . Based on the current number of browsing for each of the plurality of articles, further comprising an actual number of browsing calculation means for calculating an actual number of browsing, which is the number of browsing that each of the plurality of articles is actually browsed by the user,
The advertisement granting apparatus according to claim 1 , wherein the measured browsing number calculation unit corrects the calculated advertisement granting probability according to the measured browsing number of the unadvertised article. And further comprising: extraction means for extracting a keyword from the predetermined article to which the advertisement is attached; and search means for searching for an advertisement to be attached to the predetermined article based on the extracted keyword. The advertisement giving apparatus according to claim 1 or 2 . The advertisement providing apparatus according to any one of claims 1 to 3 , wherein the advertisement is acquired from a server provided outside the own apparatus. An advertisement granting device that grants an advertisement to a predetermined article from among a plurality of articles viewed by a user,
Click number acquisition means for acquiring the number of clicks each of the advertisements given to the plurality of articles clicked in a predetermined period by the user;
Whether to give an advertisement to the predetermined article, an advertisement grant determining means for probabilistically determining according to the number of clicks;
An advertisement giving process executing means for executing a process for giving an advertisement to the predetermined article determined to give an advertisement by the advertisement grant determining means;
Based on the number of past clicks on each of the plurality of articles, a predicted click number calculating means for calculating a predicted number of clicks that is predicted that each of the plurality of articles will be clicked by the user in the future;
Only including,
The click number acquisition means determines whether or not an advertisement unassigned article to which the advertisement is not attached is clicked, and determines that the advertisement unattached article is clicked, the predicted click number calculation means includes: An advertisement granting apparatus for calculating an advertisement grant probability for giving an advertisement to an unadvertised article according to the predicted number of clicks of the unadvertised article . A method of processing an advertisement granting device that grants an advertisement to a predetermined article from among a plurality of articles viewed by a user,
Acquiring each of the plurality of articles viewed in a predetermined period by the user;
A step of probabilistically determining whether to give an advertisement to the predetermined article according to the number of viewings;
Executing a process of giving an advertisement to the predetermined article determined to give an advertisement by the stochastic determination step;
Based on the past number of browsing for each of the plurality of articles, calculating a predicted number of browsing, which is the number of browsing predicted that each of the plurality of articles will be viewed by the user in the future,
Only including,
The step of obtaining the number of browsing determines whether or not an unadvertised article to which the advertisement is not given has been read, and if it is determined that the article to which no advertisement is given has been read, calculates the predicted number of browsing The process of calculating the advertisement grant probability of giving an advertisement to the unadvertised article is calculated according to the predicted number of browsing of the unadvertised article . In the computer of the advertisement granting apparatus that grants an advertisement to a predetermined article from among a plurality of articles viewed by the user,
A process of acquiring the number of browsing that each of the plurality of articles is browsed by the user in a predetermined period;
Whether to give an advertisement to the predetermined article, probabilistically determining according to the number of browsing,
A process of executing a process of giving an advertisement to the predetermined article determined to give an advertisement by the process of determining the probability;
Based on the past browsing number for each of the plurality of articles, a process of calculating a predicted browsing number that is a browsing number predicted to be viewed by the user in the future,
A program for executing,
The process of obtaining the number of browsing is to determine whether or not an unadvertised article to which the advertisement is not attached is read, and if it is determined that the article to which the advertisement is not attached is read, the predicted number of browsing is calculated. The process of calculating an advertisement grant probability for giving an advertisement to the unadvertised article according to the predicted number of browsing of the unadvertised article .

Images