JP5973416B2 - Advertising device, prediction method, and prediction program - Google Patents

Description

  The present invention relates to an advertising device, a prediction method, and a prediction program.

  In recent years, advertisement distribution via the Internet has been actively performed. For example, advertising content related to a company, a product, or the like is displayed at a predetermined position on a web page. Such advertisement content is, for example, an icon such as a still image, a moving image, or text data, and a hyperlink to a web page provided by the advertiser (hereinafter referred to as an advertiser page) is provided. When an advertisement content is clicked by the user, an advertisement page is displayed. The number of times the advertising content is displayed on the web page is an example of the advertising effect, and is referred to as the number of impressions.

  As a technique related to such advertisement distribution, a technique for predicting the number of impressions of advertisement content has been proposed. In this technique, a Gompertz curve is formed based on the past bid price and the actual value of the number of impressions. In this technology, the advertising content is classified into one of the dense area, discrete area, and limit area on the Gompertz curve according to the bid price presented by the advertiser, and the number of impressions depends on the classified area. Predict.

JP 2012-203575 A

  However, until now, there has not been proposed a technique for individually predicting the number of impressions of a plurality of advertising contents for which an advertising budget is set in a lump. This point will be specifically described below.

  An advertising device that distributes advertising content is sometimes referred to as an advertising fee budget (hereinafter referred to as “advertising budget”) charged by the advertiser when the advertising content is clicked by the advertiser. ) Etc. are accepted. The advertising device distributes the advertising content to the terminal device that displays the web page. Here, the advertiser may perform a campaign activity for a new product or a new service using advertisement distribution by an advertising device. Specifically, the advertiser intends to improve the advertising effect by submitting a plurality of advertising contents related to the new product or the like to the advertising device. At this time, the advertiser may set the advertising budget for a plurality of advertising contents used for the campaign activity in a lump rather than individually setting the advertising budget for each advertising content used for the campaign activity.

  On the other hand, until now, there has not been proposed a technique for individually predicting the advertising effect of a plurality of advertising contents for which an advertising budget is set in a lump. For example, in the above-described conventional technology, it is not based on the premise that an advertisement budget is collectively set for a plurality of advertisement contents, and no technique for individually predicting the advertisement effect of the plurality of advertisement contents is proposed.

  Examples of the advertising effect described above include the number of impressions of advertisement content and the number of clicks indicating the number of times the advertisement content has been clicked by the user.

  The present application has been made in view of the above, and provides an advertising device, a prediction method, and a prediction program capable of individually predicting the advertising effect of a plurality of advertising contents for which an advertising budget is collectively set. With the goal.

  The advertising device according to the present application acquires a receiving history including an advertising budget that is a budget for an advertising fee set in a batch for a plurality of advertising contents, and a distribution history including advertising effects of previously distributed advertising contents. Based on the advertisement effect included in the distribution history acquired by the acquisition means and the acquisition means, the digestion ratio, which is the ratio of the advertisement budget that is digested by the plurality of advertisement contents, out of the advertisement budget, Prediction means for individually predicting the upper limit of the advertisement effect in the plurality of advertisement contents according to the ratio.

  According to one aspect of the embodiment, there is an effect that the advertising effect of a plurality of advertising contents for which the advertising budget is set in a lump can be predicted individually.

FIG. 1 is a diagram illustrating an example of a prediction process according to the first embodiment. FIG. 2 is a diagram illustrating a configuration example of the advertisement distribution system according to the first embodiment. FIG. 3 is a diagram illustrating a configuration example of the advertising device according to the first embodiment. FIG. 4 is a diagram illustrating an example of the advertisement content storage unit according to the first embodiment. FIG. 5 is a diagram illustrating an example of a distribution history storage unit according to the first embodiment. FIG. 6 is a flowchart illustrating a prediction processing procedure performed by the advertising device according to the first embodiment. FIG. 7 is a diagram illustrating an example of a prediction process according to the modification. FIG. 8 is a diagram illustrating an example of impression number estimation processing by the advertising device according to the second embodiment. FIG. 9 is a diagram illustrating a configuration example of the advertising device according to the second embodiment. FIG. 10 is a diagram illustrating an example of a prediction result by the prediction unit according to the second embodiment. FIG. 11 is a diagram illustrating an example of a prediction result by the prediction unit according to the second embodiment. FIG. 12 is a diagram illustrating an example of impression number estimation processing by the advertising device according to the modification. FIG. 13 is a diagram illustrating an example of an impression number estimation process performed by the advertising device according to the modification. FIG. 14 is a diagram illustrating an example of a prediction result by the prediction unit according to the modification. FIG. 15 is a diagram illustrating an example of a prediction result by the prediction unit according to the modification. FIG. 16 is a diagram illustrating an example of a prediction result screen provided by the providing unit according to the modification. FIG. 17 is a diagram illustrating an example of a prediction result by the prediction unit according to the modification. FIG. 18 is a diagram illustrating an example of a prediction result by the prediction unit according to the modification. FIG. 19 is a hardware configuration diagram illustrating an example of a computer that realizes the function of the advertising device.

  Hereinafter, a mode for carrying out an advertising device, a prediction method, and a prediction program according to the present application (hereinafter referred to as “embodiment”) will be described in detail with reference to the drawings. Note that the advertising device, the prediction method, and the prediction program according to the present application are not limited by this embodiment. In the following embodiments, the same portions are denoted by the same reference numerals, and redundant description is omitted.

(First embodiment)
[1. (Prediction process)
First, an example of the prediction process according to the first embodiment will be described with reference to FIG. FIG. 1 is a diagram illustrating an example of a prediction process according to the first embodiment. FIG. 1 shows an example in which advertising effect prediction processing is performed by the advertising device 100. In the following, an example is shown in which the advertising device 100 predicts the number of impressions as an example of the advertising effect.

  The advertising device 100 shown in FIG. 1 performs processing for receiving advertising content from the advertiser device 20 used by the advertiser and distributing the received advertising content. Then, the advertising device 100 stores in the distribution history storage unit 122 a distribution history related to advertisement content distributed in the past (hereinafter may be referred to as “distributed advertisement content”). For example, the distribution history storage unit 122 stores, as the distribution history, the number of clicks, the price per click, the number of impressions, the CTR (Click Through Rate), and the like of the distributed advertisement content.

  Further, it is assumed that the advertiser of the advertiser apparatus 20 shown in FIG. 1 submits a plurality of advertisement contents related to the new product or the like to the advertising apparatus 100 for the purpose of campaign activities for new products and new services. Then, the advertiser shall collectively set an advertising budget for a plurality of advertising contents. In the following, a plurality of advertisement contents for which an advertisement budget is set in a lump may be collectively referred to as a “campaign advertisement group”.

  Under such a premise, in the example of FIG. 1, the advertising device 100 accepts a prediction request for predicting the advertising effect in the campaign advertisement group C10 from the advertiser device 20 (step S11). Specifically, the advertising device 100 receives a prediction request from the advertiser device 20 together with the campaign advertisement group C10 and the advertisement budget BU10 set in the campaign advertisement group C10 in a lump. The campaign advertisement group C10 includes advertisement contents C11, C12, C13, and C14.

  In this case, the advertising device 100, based on the distribution history stored in the distribution history storage unit 122, for each advertising content included in the campaign advertisement group C10, the upper limit value of the number of impressions constrained by the advertising budget BU10 (hereinafter referred to as the advertising history BU10). , Sometimes referred to as “maximum impressions”). That is, the advertising device 100 individually predicts the maximum number of impressions of the advertising contents C11, C12, C13, and C14.

  Specifically, it is assumed that the distribution history storage unit 122 illustrated in FIG. 1 stores a distribution history regarding the campaign advertisement group C1 including the combination of the advertisement contents C11, C12, C13, and C14. That is, the advertising device 100 has distributed the campaign advertisement group C1 similar to the campaign advertisement group C10 in the past.

  Here, it is also conceivable to use the actual value of the number of impressions stored in the distribution history storage unit 122 as the prediction result. However, the distribution conditions for the campaign advertisement group C1 may differ from the distribution conditions for the campaign advertisement group C10. For example, the advertisement budget of the campaign advertisement group C1 may be different from the advertisement budget of the campaign advertisement group C10. In addition, the distribution period of the campaign advertisement group C1 may be different from the distribution period of the campaign advertisement group C10. Since the number of impressions of the advertisement content varies depending on the advertisement budget and the delivery period, when the actual value of the number of impressions is simply used as the prediction result, the advertisement contents C11, C12, C13, and C14 restricted by the advertisement budget BU10. It is difficult to predict the maximum number of impressions.

  Therefore, the advertising device 100 according to the first embodiment uses the advertising budget C11, C12, C13, and C14 based on the distribution history of the campaign advertisement group C1 stored in the distribution history storage unit 122. The ratio of BU10 (hereinafter sometimes referred to as “digestion ratio”) is obtained (step S12). Specifically, the advertising device 100 determines the amount of the advertising budget BU10 that is consumed when the advertising content C11 is clicked, the amount of the advertising budget BU10 that is consumed when the advertising content C12 is clicked, and the advertising content. A ratio between the amount of the advertising budget BU10 that is consumed when C13 is clicked and the amount of the advertising budget BU10 that is consumed when the advertising content C14 is clicked is obtained as a digestion ratio.

  In the case of the example in FIG. 1, the advertising device 100 can determine the digestion ratio using the distribution history because the combination of the advertising contents C11 to C14 has been distributed in the past as the campaign advertisement group C1. Although this point will be described in detail later, the advertising device 100 can obtain the digestion ratio based on the number of clicks, the unit price of clicks, the number of impressions, and the like stored in the distribution history storage unit 122, for example.

  Then, the advertising device 100 individually predicts the maximum number of impressions of each advertising content (that is, advertising content C11, C12, C13, and C14) included in the campaign advertisement group C10 according to the digestion ratio (step S13). Specifically, the advertising contents C11 to C14 are not distributed the same number of times by the advertising device 100. For example, some of the advertising contents C11 to C14 are distributed preferentially over other advertising contents according to their characteristics (for example, unit cost per click, CTR, etc.). Some advertising content is harder to deliver. Since the advertising content with a higher number of distributions has a higher proportion of spending the advertising budget, the above-described digestion rate can be said to indicate a tendency of the number of times of distribution in consideration of the characteristics of each advertising content. For this reason, the advertising device 100 can predict the maximum number of impressions of each advertising content individually according to the consumption ratio of the advertising budget BU10.

  Subsequently, the advertising device 100 provides the prediction result to the advertiser device 20 (step S14). For example, the advertising device 100 provides the maximum number of impressions of the advertising content C11, C12, C13, and C14 predicted in step S13. Thereby, the advertiser can grasp the maximum number of impressions of the advertising contents C11, C12, C13, and C14 in advance. As a result, the advertiser can submit the campaign advertisement group C10 to the advertising device 100 after setting an advertising budget such that the maximum number of impressions of each advertising content reaches the target value.

  As described above, the advertising device 100 according to the first embodiment does not simply use the distribution result of the campaign advertisement group stored as the distribution history, but based on the digest ratio of the advertising budget, Estimate the maximum number of impressions individually. For this reason, the advertising device 100 can predict the maximum number of impressions of each advertisement content even in a campaign advertisement group in which an advertisement budget is set in a lump.

[2. Configuration of advertisement distribution system)
Next, the configuration of the advertisement distribution system according to the first embodiment will be described with reference to FIG. FIG. 2 is a diagram illustrating a configuration example of the advertisement distribution system 1 according to the first embodiment. As shown in FIG. 2, the advertisement distribution system 1 includes a terminal device 10, an advertiser device 20, a distribution device 30, and an advertising device 100. The terminal device 10, the advertiser device 20, the distribution device 30, and the advertising device 100 are connected via a network N so as to be communicable by wire or wirelessly. The advertisement distribution system 1 illustrated in FIG. 2 may include a plurality of terminal devices 10, a plurality of distribution devices 30, and a plurality of advertisement devices 100.

  The terminal device 10 is an information processing device such as a desktop PC (Personal Computer), a notebook PC, a tablet terminal, a mobile phone, or a PDA (Personal Digital Assistant). For example, the terminal device 10 obtains a web page from the delivery device 30 by accessing the delivery device 30 and displays the obtained web page on a display device (for example, a liquid crystal display). Further, when the web page includes an advertising space, the terminal device 10 acquires the advertising content from the advertising device 100 by accessing the advertising device 100, and displays the acquired advertising content on the web page. However, the present invention is not limited to this example, and the terminal device 10 may acquire a web page including advertisement content from the distribution device 30. In this case, the distribution device 30 distributes to the terminal device 10 a web page in which the advertising content distributed by the advertising device 100 is incorporated.

  The advertiser device 20 is an information processing device used by an advertiser who requests the advertising device 100 to distribute an advertisement. The advertiser device 20 submits advertising content to the advertising device 100 in accordance with an operation by the advertiser. For example, the advertiser apparatus 20 is an advertisement corresponding to a still image, a moving image, text data, a URL (Uniform Resource Locator) for accessing a web page provided by an advertiser server managed by the advertiser, or the like. Content is submitted to the advertising device 100.

  The advertiser may use the advertiser device 20 to request the agency to submit the advertising content without submitting the advertising content to the advertising device 100. In this case, it is the agency that submits the advertising content to the advertising device 100. In the following, the term “advertiser” is a concept that includes not only the advertiser but also the agency, and the term “advertiser device” includes the agency device used by the agency as well as the advertiser device. It is assumed to be a concept.

  The distribution device 30 is a Web server or the like that provides a web page to the terminal device 10. The distribution device 30 is, for example, various web pages related to news sites, auction sites, weather forecast sites, shopping sites, finance (stock price) sites, route search sites, map providing sites, travel sites, restaurant introduction sites, web blogs, and the like. I will provide a.

  The advertising device 100 is a server device that distributes advertising content submitted from the advertiser device 20. As described above, the advertising device 100 distributes the advertising content to the terminal device 10 when accessed from the terminal device 10. Further, the advertising device 100 distributes the advertising content to the distribution device 30 when accessed from the distribution device 30.

[3. Configuration of distribution device]
Next, the configuration of the advertising device 100 according to the first embodiment will be described with reference to FIG. FIG. 3 is a diagram illustrating a configuration example of the advertising device 100 according to the first embodiment. As illustrated in FIG. 3, the advertising device 100 includes a communication unit 110, an advertising content storage unit 121, a distribution history storage unit 122, and a control unit 130. The advertising device 100 includes an input unit (for example, a keyboard and a mouse) that receives various operations from an administrator who uses the advertising device 100, and a display unit (for example, a liquid crystal display) that displays various types of information. You may have.

[3-1. Communication department)
The communication unit 110 is realized by, for example, a NIC (Network Interface Card). The communication unit 110 is connected to the network N in a wired or wireless manner, and transmits / receives information to / from the terminal device 10, the advertiser device 20, and the distribution device 30 via the network N.

[3-2. (Storage section)
The advertisement content storage unit 121 and the distribution history storage unit 122 are realized by, for example, a semiconductor memory device such as a RAM (Random Access Memory) and a flash memory, or a storage device such as a hard disk and an optical disk.

[3-3. Advertising content storage unit]
The advertising content storage unit 121 stores the advertising content submitted from the advertiser device 20. Here, FIG. 4 shows an example of the advertising content storage unit 121 according to the first embodiment. In the example shown in FIG. 4, the advertisement content storage unit 121 includes “advertiser ID”, “campaign ID”, “delivery period”, “advertisement budget”, “advertisement content”, “bid price”, “cost per click ( CPC: Cost Per Click), “cumulative billing amount”, “targeting condition”, and “CTR”.

  The “advertiser ID” indicates identification information for identifying the advertiser or the advertiser device 20. “Campaign ID” indicates identification information for identifying a campaign advertisement group in which an advertisement budget is set in a lump. “Distribution period” is information set when an advertiser submits advertisement content, and indicates a period during which the advertisement content is distributed. “Advertising budget” is information set when an advertiser submits advertising content, and indicates an upper limit value of an advertising fee (that is, a cumulative billing amount) paid by an advertiser to an advertisement distributor.

  “Advertising content” indicates the advertising content submitted from the advertiser apparatus 20. The advertising content storage unit 121 stores, for example, advertising content that has been distributed in the past by the advertising device 100 but will not be distributed in the future, advertising content that is currently being distributed, advertising content that is scheduled to be distributed in the future, and the like. The In the example of FIG. 4, an example in which conceptual information such as “C11” and “C12” is stored in “advertising content” is shown. However, in practice, still images, moving images, text data, URLs, Alternatively, a file path name indicating the storage location is stored.

  “Bid price” indicates the unit price of the advertising fee set when the advertiser submits the advertising content. “Cost per click” is the unit price (for example, the unit is “yen”) of the amount charged by the advertisement distributor (for example, the administrator of the advertising device 100) to the advertiser when the advertisement content is clicked once by the user. It corresponds to. The “bid price” and the “click unit price” may be the same. Here, the “click unit price” is determined by an auction based on the “bid price” of a plurality of bidders. For example, the “cost per click” of the “n” bidder is the bid price of the “n + 1” bidder. Such a method for determining the unit price of click is called GSP (Generalized Second Price Auction) or the like.

  The “cumulative billing amount” indicates an advertisement fee charged to the advertiser at the present time, and corresponds to a value obtained by multiplying the number of times the advertisement content is clicked by the click unit price. The “targeting condition” is information set when the advertiser submits the advertisement content, and indicates the condition of the user who is the distribution target of the advertisement content. For example, the “targeting condition” stores the user attribute of the user to whom the advertising content is to be distributed.

  “CTR” indicates whether or not advertisement content has been selected (for example, clicked) by the user of the terminal device 10, and corresponds to a value obtained by dividing the number of times of selection (that is, the number of clicks) by the number of impressions. This “CTR” stores only the advertising content that has been distributed to the terminal device 10. The “CTR” of the advertisement content that has never been distributed to the terminal device 10 includes a predetermined fixed value, an average value of CTR for all advertisement contents, or the same advertisement category (for example, car, travel, etc.). The average value of CTR in all advertisement contents belonging to) may be stored.

  That is, FIG. 4 shows an example in which the advertiser identified by the advertiser ID “A10” submits the campaign advertisement group identified by the campaign ID “C1”. In FIG. 4, the campaign advertisement group with the campaign ID “C1” includes the advertisement contents C11 to C14, the distribution period “November 1, 2013 to December 31, 2013”, the advertisement budget “200,000. An example in which “circle” is set is shown.

[3-4. Distribution history storage unit)
The distribution history storage unit 122 stores a distribution history related to advertising content distributed in the past by the advertising device 100 (that is, distributed advertising content). Here, FIG. 5 shows an example of the distribution history storage unit 122 according to the first embodiment. The distribution history storage unit 122 may be configured by a table in the database as in the example illustrated in FIG. 5, but may be a text file or the like in which the distribution history is written. In the example illustrated in FIG. 5, the distribution history storage unit 122 stores “distribution date / time”, “campaign ID”, “advertisement content”, “impression count”, “stop period”, “click count”, “CTR”, “ It has items such as “cost per click” and “targeting condition”.

  “Distribution date and time” indicates the date and time when the advertisement content is distributed. “Campaign ID” is the same as the campaign ID shown in FIG. “Advertising content” indicates the advertising content distributed by the advertising device 100. The “number of impressions” indicates the number of times the advertising content is distributed to the terminal device 10 by the advertising device 100. That is, the “number of impressions” indicates the number of times the advertising content is displayed on a web page or the like. The “stop period” indicates a period during which the advertisement apparatus 100 has stopped distributing advertisement content. Specifically, in the “stop period”, a distribution stop start time that is a start time of distribution stop and a distribution stop end time that is an end time of distribution stop are stored.

  “Number of clicks” indicates the number of times the user has clicked on the advertisement content. “CTR” is the same as the CTR shown in FIG. However, the CTR of the advertisement content storage unit 121 stores the ratio of the number of clicks to the total number of distributions, whereas the CTR of the distribution history storage unit 122 has the ratio of the number of clicks to the number of distributions at the distribution date and time. Remembered. The “CPC unit price” is the same as the CPC unit price shown in FIG. The “targeting condition” is the same as the targeting condition shown in FIG.

  Here, the advertiser may change the bid price or the targeting condition in the advertisement content being distributed. For this reason, the cost-per-click of the advertising content being distributed may also vary. The advertisement content storage unit 121 stores the latest cost-per-click and targeting conditions, while the distribution history storage unit 122 stores the cost-per-click and targeting conditions set at the distribution date and time.

  The distribution history storage unit 122 illustrated in FIG. 5 stores the distribution history of each advertising content distributed by the advertising device 100 every day. For example, as illustrated in FIG. 5, the distribution history storage unit 122 records different distribution histories of the campaign advertisement group C1 distributed on both “November 1, 2013” and “November 2, 2013”. Remember me. However, without being limited to this example, the distribution history storage unit 122 may store the distribution history every hour, every 12 hours, or every week.

  That is, FIG. 5 shows an example in which the advertisement contents C11 to C14 included in the campaign ID “C1” are distributed on “November 1, 2013”. In FIG. 5, the number of impressions of the advertising content C11 is “7000”, the stop periods are “12:30 to 12:50” and “19:30 to 19:50”, and the click An example in which the number is “28 times” and the CTR is “0.004” is shown. In this way, the advertisement content may be stopped from being delivered many times a day. In the stop period “12:30 to 12:50”, “12:30” indicates a distribution stop start time, and “12:50” indicates a distribution stop end time.

[3-5. Control unit)
Returning to the description of FIG. 3, the control unit 130, for example, various programs (an example of a prediction program) stored in a storage device inside the advertising device 100 by a CPU (Central Processing Unit), an MPU (Micro Processing Unit), or the like. Is realized by using the RAM as a work area. The control unit 130 is realized by an integrated circuit such as an application specific integrated circuit (ASIC) or a field programmable gate array (FPGA).

  As shown in FIG. 3, the control unit 130 includes a reception unit 131, a distribution unit 132, a submission reception unit 133, an acquisition unit 134, a prediction unit 135, and a provision unit 136, which will be described below. Realize or execute information processing functions and actions. Note that the internal configuration of the control unit 130 is not limited to the configuration illustrated in FIG. 3, and may be another configuration as long as the information processing described below is performed. Further, the connection relationship between the processing units included in the control unit 130 is not limited to the connection relationship illustrated in FIG. 3, and may be another connection relationship.

[3-6. (Receiver)
The receiving unit 131 receives an advertisement content acquisition request from the terminal device 10 or the distribution device 30. For example, the receiving unit 131 receives an advertisement content acquisition request by an HTTP (Hypertext Transfer Protocol) request or the like.

  The device that transmits the advertisement content acquisition request to the receiving unit 131 differs depending on the web page distributed by the distribution device 30. For example, when a web page in which a URL for accessing the advertising device 100 is embedded is distributed to the terminal device 10, the receiving unit 131 receives an advertising content acquisition request from the terminal device 10. When a web page in which advertisement content is already embedded is distributed to the terminal device 10, the receiving unit 131 receives an advertisement content acquisition request from the distribution device 30.

[3-7. (Delivery Department)
The distribution unit 132 distributes any of the advertisement contents stored in the advertisement content storage unit 121 when the reception unit 131 receives the advertisement content acquisition request. Specifically, the distribution unit 132 extracts, from the advertisement content stored in the advertisement content storage unit 121, the advertisement content whose current date and time is included in the distribution period as the distribution candidate advertisement content. Then, the distribution unit 132 determines the advertisement content to be distributed according to the CPC unit price, CTR, and targeting conditions from among the advertisement candidate advertisement contents.

  More specifically, the distribution unit 132 preferentially distributes the advertising content having a higher click unit price (that is, bid price) among the distribution candidate advertising contents. In addition, the distribution unit 132 preferentially distributes the advertisement content having a higher CTR among the distribution candidate advertisement contents. In addition, the distribution unit 132 preferentially distributes the advertising content whose targeting condition matches the user attribute of the terminal device 10 that is the distribution destination among the distribution candidate advertising content. For example, the distribution unit 132 distributes the advertisement content having the highest multiplication result or addition result of “CPC unit price” and “CTR” among the advertisement contents whose targeting conditions and user attributes match.

  The distribution unit 132 may use the CTR predicted by a predetermined method instead of using the actual CTR itself stored in the advertisement content storage unit 121. For example, the distribution unit 132 may use a CTR predicted based on a type of advertisement content distributed in the past, a type of web page on which the advertisement content is displayed, a past CTR, or the like.

  Then, when the advertising content is distributed, the distribution unit 132 updates the “number of impressions”, “cost per click”, and “targeting condition” in the distribution history storage unit 122. In addition, when the advertisement content is clicked by the user of the terminal device 10, the distribution unit 132 receives a click notification indicating that the user has clicked from the terminal device 10. In this case, the distribution unit 132 updates the “cumulative billing amount” and “CTR” in the advertisement content storage unit 121 and updates the “number of clicks” and “CTR” in the distribution history storage unit 122 based on the click notification. To do.

  In addition, the distribution unit 132 performs a process of stopping the distribution of the advertising content for each advertising content according to the advertising budget and the accumulated billing amount. For example, the distribution unit 132 calculates the advertising budget per day by dividing the advertising budget by the number of days in the distribution period, and the advertising content is calculated according to the advertising budget consumption rate so that the advertising budget is gradually consumed. Stop distribution for a certain period of time. Then, the distribution unit 132 updates the “stop period” in the distribution history storage unit 122 when the distribution stop processing of the advertising content is performed. For example, the distribution unit 132 stores a distribution stop start time that is a time when the distribution stop process is started and a distribution stop end time that is a time when the distribution stop process is ended in the “stop period” of the distribution history storage unit 122. .

[3-8. (Submission Reception Department)
The submission accepting unit 133 accepts the submission of the advertising content from the advertiser device 20 and stores the accepted advertising content in the advertising content storage unit 121. Specifically, the submission accepting unit 133 accepts the submission of the advertising content together with the delivery period, the advertising budget, the bid price, and the targeting condition, and associates the submitted advertising content with the delivery period, the advertising budget, the bid. The price and targeting condition are stored in the advertising content storage unit 121. At this time, the submission reception unit 133 also stores the advertiser ID corresponding to the advertiser who is the submission source of the advertisement content in the advertisement content storage unit 121.

  Further, the submission accepting unit 133 accepts submission of a campaign advertisement group from the advertiser device 20. Specifically, the submission receiving unit 133 receives a plurality of advertisement contents forming a campaign advertisement group, one distribution period set in the campaign advertisement group, and one advertisement budget. However, the submission accepting unit 133 accepts a bid price and a targeting condition for each advertisement content included in the campaign advertisement group. In this case, the submission receiving unit 133 advertises the distribution period, the advertising budget, each advertising content, the bid price and the targeting condition for each advertising content in association with the campaign ID for identifying the campaign advertising group. It is stored in the content storage unit 121.

  In addition, the submission receiving unit 133 receives a prediction request from the advertiser device 20 to predict the advertisement effect in the campaign advertisement group scheduled to be submitted. For example, the submission receiving unit 133 receives a prediction request including a campaign advertisement group scheduled to be submitted and an advertisement budget. In addition, for example, the submission receiving unit 133 receives a prediction request including a targeting condition and a bid price for each advertising content in addition to a campaign advertisement group and an advertising budget scheduled to be submitted. Then, when receiving a prediction request from the advertiser device 20, the submission receiving unit 133 instructs the acquisition unit 134 to predict the advertisement effect.

[3-9. (Acquisition Department)
The acquisition unit 134 acquires the distribution history from the distribution history storage unit 122 when instructed by the submission accepting unit 133 to predict the advertising effect. Specifically, the acquisition unit 134 identifies a campaign ID including a combination of advertising content included in the campaign advertisement group to be predicted from the distribution history storage unit 122, and distributes history corresponding to the identified campaign ID (for example, Impressions, clicks, CTR, cost-per-click, targeting criteria, etc.). For example, it is assumed that the distribution history storage unit 122 is in the state shown in FIG. In addition, it is assumed that the advertisement contents C11, C12, C13, and C14 are included in the campaign advertisement group to be predicted. In this case, the acquisition unit 134 specifies the campaign ID “C1” stored in the distribution history storage unit 122, and acquires the distribution history corresponding to the campaign ID “C1”. Note that the acquisition unit 134 does not include all distribution histories stored in the distribution history storage unit 122, but only distribution histories whose distribution date and time are included in a predetermined period (for example, a period from the current date to a predetermined number of days in the past). May be obtained.

[3-10. (Predictor)
The prediction unit 135 predicts the advertising effect of the campaign advertisement group received by the submission receiving unit 133 using the distribution history acquired by the acquisition unit 134. Specifically, the prediction unit 135 obtains a digest ratio of the advertisement budget that is consumed by each advertisement content included in the campaign advertisement group, and predicts the maximum number of impressions of each advertisement content as an advertisement effect according to the determined digest ratio. To do.

  Hereinafter, the prediction process by the prediction unit 135 will be specifically described. First, as a premise, “CPC (cost per click)” is estimated based on a predetermined estimation formula using “bid price” as a variable, based on information related to advertising content stored in the advertising content storage unit 121. . For example, the relationship between “CPC” and “bid price (Bid)” is expressed by the following equation (1). In the following formula (1), “F” represents a predetermined estimation function.

  Subsequently, the relationship between “advertisement budget”, “CPC (cost per click)”, and “number of clicks” is expressed by the following equation (2).

  Further, the “click number” is represented by the following formula (3) including “CTR” and “impression number (Imps)”.

  That is, by substituting the above formula (3) into the above formula (2), the number of impressions (Imps) is expressed by the following formula (4).

  That is, the maximum number of impressions in the advertising content is the value on the right side of the above equation (4). Therefore, when the maximum number of impressions is “max (Imps)”, the following formula (5) is established.

  Thus, the maximum number of impressions varies depending on the advertising budget. That is, the prediction unit 135 can predict the maximum number of impressions of the campaign advertisement group by substituting the advertisement budget set for the campaign advertisement group at once into the above equation (5). However, the prediction unit 135 cannot individually predict the maximum number of impressions of each advertisement content included in the campaign advertisement group only by using the above formula (5).

  Therefore, the prediction unit 135 according to the first embodiment obtains a digest ratio of an advertisement budget that is consumed by each advertisement content included in the campaign advertisement group. Then, the prediction unit 135 sets the advertising budget used by each advertising content among the advertising budgets set in a batch for the campaign advertising group as the advertising budget in the above formula (5), so that the maximum number of impressions for each advertising content is obtained. Predict.

More specifically, the prediction unit 135 counts the total number of clicks having the same campaign ID and the same advertising content in the distribution history acquired by the acquisition unit 134. Then, the prediction unit 135 multiplies the total number of clicks by the cost per click included in the distribution history, thereby calculating the total of the advertising budget consumed by the advertising content (hereinafter sometimes referred to as “cumulative budget”). Ask. The prediction unit 135 calculates a cumulative budget for each advertising content included in the campaign advertisement group to be predicted, using a distribution history corresponding to the advertising content. For example, N pieces of advertising content included in the campaign advertisement group to be predicted are set as advertising content _i (i = 1 to N), and among the CPCs included in the distribution history, the CPC corresponding to the advertising content _i is “CPC”. _i ”, and when the total number of clicks corresponding to the advertisement content _i is“ cumulative Click _i ”, the prediction unit 135 obtains the following equation (6) for i = 1 to N. In addition, when the bid accepting price of each advertisement content is accepted by the submission accepting unit 133, the predicting unit 135 may substitute the accepted bid price into “CPC _i ” in the following equation (6). .

Here, since the number of clicks is represented by the above formula (3), the prediction unit 135 may use the following formula (7) instead of the above formula (6). In this case, the prediction unit 135 counts the cumulative number of impressions (cumulative Imps _i ) that have the same campaign ID and the same advertising content in the distribution history acquired by the acquisition unit 134. Further, the prediction unit 135 sets the CTR included in the distribution history as “CTR _i ” in the following equation (7).

Then, the prediction unit 135 obtains the ratio of the advertising budget consumed by the advertising content included in the campaign advertisement group to be predicted by obtaining the ratio of the cumulative budget _i obtained by the above formula (6) or formula (7). Find a certain digestion ratio. Thereby, the prediction unit 135 can individually predict an advertisement budget to be consumed by each advertisement content among the advertisement budgets set in the campaign advertisement group to be predicted, based on the digest ratio. That is, the prediction unit 135 can predict the maximum number of impressions for each advertising content by substituting the advertising budget for each advertising content predicted in this way into the above equation (5). For example, the prediction unit 135 obtains the maximum number of impressions _i (max (Imps _i )) of the advertisement content _i included in the campaign advertisement group to be predicted by the following equation (8).

  Note that the predicting unit 135 can convert the above equation (8) into the following equations (9) and (10).

  By the way, the advertiser sets an advertising budget for a predetermined distribution period. For this reason, the prediction unit 135 predicts the maximum number of impressions in the distribution period set by the advertiser by substituting the advertising budget into the above equation (8). For example, when the advertiser sets an advertising budget for each day, the prediction unit 135 predicts the maximum number of impressions in a day. For example, when the advertiser sets a weekly advertisement budget or a monthly advertisement budget, the prediction unit 135 predicts the maximum number of impressions in one week or the maximum number of impressions in one month. However, the prediction unit 135 can obtain the daily advertising budget by dividing the advertising budget set by the advertiser by the number of days in the distribution period. Then, the prediction unit 135 can predict the maximum number of impressions per day by substituting the daily advertising budget into the equation (8).

[3-11. (Providing department)
Returning to the description of FIG. 3, the providing unit 136 provides the advertiser device 20 with the prediction result predicted by the prediction unit 135. For example, the providing unit 136 provides the advertiser device 20 with the maximum number of impressions for each advertising content included in the campaign advertisement group to be predicted.

  The advertiser device 20 can formally submit the campaign advertisement group after receiving the prediction result from the providing unit 136. For example, the advertiser can grasp the predicted value of the maximum number of impressions by transmitting the campaign advertisement group to be submitted, the advertising budget, etc. to the advertising device 100 before officially submitting the campaign advertisement group. After that, campaign ads can be formally submitted.

[4. (Prediction processing procedure)
Next, the procedure of the prediction process by the advertising device 100 according to the first embodiment will be described with reference to FIG. FIG. 6 is a flowchart illustrating a prediction processing procedure performed by the advertising device 100 according to the first embodiment.

  As illustrated in FIG. 6, the submission receiving unit 133 determines whether an advertisement effect prediction request in the campaign advertisement group has been received from the advertiser device 20 (step S <b> 101). And when the acceptance request | requirement part 133 has not received the prediction request | requirement (step S101; No), it waits until it receives a prediction request | requirement.

  On the other hand, when a prediction request is received by the submission receiving unit 133 (step S101; Yes), the acquisition unit 134 includes a campaign including a combination of advertisement contents identical to the campaign advertisement group to be predicted from the distribution history storage unit 122. The distribution history of the advertisement group is acquired (step S102). In step S <b> 101, the submission receiving unit 133 receives a campaign advertisement group to be submitted, an advertisement budget, and the like from the advertiser device 20.

  Subsequently, the prediction unit 135 obtains a digest ratio of each advertisement content included in the campaign advertisement group to be predicted based on the distribution history acquired by the acquisition unit 134 (step S103). Then, the prediction unit 135 predicts the maximum number of impressions of each advertisement content based on the digestion ratio (step S104).

  Then, the providing unit 136 provides the prediction result by the prediction unit 135 to the advertiser device 20 that is the transmission source of the prediction request received in Step S101 (Step S105).

[5. effect〕
As described above, the advertising device 100 according to the first embodiment includes the submission reception unit (corresponding to an example of a reception unit) 133, the acquisition unit 134, and the prediction unit 135. The submission accepting unit 133 accepts an advertisement budget that is a budget for an advertisement fee set in a batch for a plurality of advertisement contents. The acquisition unit 134 acquires a distribution history including the number of impressions of the distributed advertisement content distributed in the past (corresponding to an example of the advertisement effect). Based on the number of impressions included in the distribution history acquired by the acquisition unit 134, the prediction unit 135 is based on the ratio of the advertising budget that is consumed by a plurality of advertising contents out of the advertising budget received by the submission receiving unit 133. A certain digest ratio is obtained, and the maximum number of impressions (corresponding to an example of the upper limit of the advertisement effect) in a plurality of advertisement contents is individually predicted according to the obtained digest ratio.

  Thereby, the advertising device 100 according to the first embodiment can predict the maximum number of impressions of each advertisement content even in the campaign advertisement group in which the advertisement budget is set in a lump.

[6. (Modification)
The advertising device 100 described above may be implemented in various different forms other than the above embodiment. Therefore, in the following, another embodiment of the advertising device 100 will be described.

[6-1. (Predicted advertising effect)
In the first embodiment, the example in which the advertising device 100 predicts the maximum number of impressions as an example of the advertising effect is shown. However, the advertising device 100 may predict other advertising effects instead of the maximum number of impressions. For example, the advertising device 100 predicts the upper limit number of clicks constrained by the advertising budget (hereinafter, may be referred to as “maximum click number”) for each advertising content included in the campaign advertisement group scheduled to be submitted. May be. In this case, the acquisition unit 134 and the prediction unit 135 can predict the maximum number of clicks in the advertisement content scheduled to be submitted by using the number of clicks instead of the number of impressions in the processing described above.

  Specifically, the maximum number of clicks (max (Click)) is expressed by the following equation (11) according to the above equation (2).

That is, similarly to the above formula (8), the prediction unit 135 calculates the maximum number of clicks (max (Click) of the advertising content _i (i = 1 to N) by the following formula (12) in which the digestion ratio of the advertising budget is considered _i )) can be predicted.

  Note that the prediction unit 135 can convert the equation (12) into the following equations (13) and (14) based on the equation (6).

  In this way, the advertising device 100 may predict the maximum number of clicks for each advertising content and provide the prediction result to the advertiser device 20. Alternatively, the advertising device 100 may predict both the number of impressions and the maximum number of clicks, and provide the prediction result to the advertiser device 20.

[6-2. (Delivery results)
In the first embodiment, an example has been shown in which a combination of advertisement contents included in a campaign advertisement group to be predicted has been distributed in the past as one campaign advertisement group. However, when a part of the combination of the advertisement contents is distributed as the campaign advertisement group among the plurality of advertisement contents included in the campaign advertisement group to be predicted, the advertising device 100 determines the maximum number of impressions for each advertisement content. (Or the maximum number of clicks) can be predicted. Specifically, the advertising device 100 can predict the maximum number of impressions (or the maximum number of clicks) for each advertising content when the prediction target campaign advertisement group satisfies the following two conditions.

<Condition 1> Each advertisement content included in the campaign advertisement group to be predicted has been distributed as a campaign advertisement group together with any of the other advertisement contents included in the campaign advertisement group to be predicted <Condition 2 > In condition 1, each campaign advertisement group distributed in the past includes at least one advertisement content identical to that of another campaign advertisement group

  This point will be described with reference to FIG. FIG. 7 is a diagram illustrating an example of a prediction process according to the modification. Here, it is assumed that a campaign advertisement group including the advertisement contents C11, C12, C13, and C14 is a prediction target. As illustrated in FIG. 7, the distribution history storage unit 122 stores a distribution record related to the campaign advertisement group C2 including the advertisement contents C11 and C12 and a distribution record related to the campaign advertisement group C3 including the advertisement contents C12, C13, and C14. .

  In the case of this example, each of the advertisement contents C11 to C14 included in the campaign advertisement group to be predicted has been distributed as a campaign advertisement group together with any of the advertisement contents C11 to C14. For this reason, the campaign advertisement group to be predicted satisfies <Condition 1> described above. The campaign advertisement groups C1 and C2 include the same advertisement content C12. For this reason, the campaign advertisement group to be predicted satisfies <Condition 2> described above.

  Here, it is assumed that the prediction unit 135 obtains “2: 1” as the digestion ratio between the advertisement content C11 and the advertisement content C12 included in the campaign advertisement group C2 using the above formula (7). In addition, it is assumed that the prediction unit 135 calculates “2: 1: 1” as the digest ratio of the advertisement content C12, the advertisement content C13, and the advertisement content C14 included in the campaign advertisement group C3. In this case, the prediction unit 135 can obtain the digest ratio “4: 2: 1: 1” of the advertisement contents C11 to C14 by using the advertisement content C12 as a reference. Thereby, the prediction unit 135 can individually predict the maximum number of impressions of the advertisement contents C11 to C14 by applying the digestion ratio “4: 2: 1: 1” to the above formula (8). Similarly, the prediction unit 135 can individually predict the maximum number of clicks of the advertisement contents C11 to C14 by applying the digestion ratio “4: 2: 1: 1” to the equation (12).

(Second Embodiment)
In the said 1st Embodiment, the example which estimates the maximum impression number of the advertising content contained in the campaign advertisement group of prediction object was shown. However, the number of impressions decreases when the winning bid price (i.e., the cost per click) is low. This is because advertisement content with a higher winning bid price is preferentially distributed. In other words, the number of impressions of the advertisement content may not reach the maximum number of impressions predicted in the first embodiment when the bid price is low. In the second embodiment, an example in which the number of impressions corresponding to the bid price is predicted in addition to the maximum number of impressions will be described. In addition, since the structure of the advertisement delivery system 1 which concerns on 2nd Embodiment is the same as that of the example shown in FIG. 2, description is abbreviate | omitted below.

[1. (Prediction process)
The prediction process according to the second embodiment is performed by the advertising device 200. The advertising device 200 preferentially distributes the advertising content having a higher bid price. Further, the advertising device 200 stops the distribution of the advertising content for a certain period according to the advertising budget consumption rate so that the advertising budget set for the advertising content is gradually consumed.

  In addition, the advertising device 200 predicts the relationship between the bid price and the number of impressions for each advertising content included in the campaign advertisement group to be predicted. Specifically, the advertising device 200 includes a distribution history storage unit 122 as with the advertising device 100. Then, the advertising device 200 generates a prediction model for predicting the number of impressions of each advertisement content included in the campaign advertisement group to be predicted based on the distribution history stored in the distribution history storage unit 122.

  In the first embodiment, an example has been described in which the advertising device 100 uses the distribution results in the campaign advertisement group to be predicted in order to obtain the digestion ratio. On the other hand, since the advertising device 200 according to the second embodiment generates a prediction model indicating the tendency of the number of impressions corresponding to the bid price, not only the delivery results of the campaign advertisement group to be predicted, The distribution results of advertisement contents are also used.

  Here, the distributed advertisement content whose distribution history is stored in the distribution history storage unit 122 includes a distributed advertisement content that has not been stopped by the advertising device 200 and a distributed advertisement that has been stopped. Content. Since the distributed advertisement content that has never been stopped is not stopped in the first place, there is no possibility that it has been distributed more than the number of impressions stored in the distribution history storage unit 122. On the other hand, the distributed advertisement content that has been stopped may actually be distributed more than the number of impressions stored in the distribution history storage unit 122 unless the advertising budget is limited.

  That is, the number of impressions stored in the distribution history may be suppressed due to factors such as the advertising budget. Therefore, the advertising device 200 is not always able to predict the number of impressions corresponding to the bid price with high accuracy when the actual value of the number of impressions is used. For example, since an advertisement content with a higher bid price is preferentially distributed, the number of impressions may increase as the bid price increases. However, as described above, the number of impressions of advertisement content whose distribution has been stopped is suppressed. For this reason, when the actual value of the number of impressions is simply used, the advertising device 200 may predict the number of impressions such that the increase rate decreases when the bid price is higher than a predetermined amount. When such a prediction result is provided to the advertiser, the advertiser recognizes that the number of impressions does not increase even if the bid price is set higher than a predetermined amount. As a result, even an advertiser who desires a large number of impressions does not set a high bid price, so the administrator of the advertising device 200 may lose the opportunity to obtain a high amount of advertising revenue.

  Therefore, the advertising device 200 according to the second embodiment makes it possible to predict the relationship between the bid price and the number of impressions with high accuracy by performing the processing described below.

  Specifically, the advertising device 200 acquires from the distribution history storage unit 122 the number of impressions of distributed advertising content that has never been stopped. Further, the advertising device 200 acquires the number of impressions and the stop period of the distributed advertisement content that has been stopped from the distribution history storage unit 122. Then, the advertising device 200 estimates the number of impressions when the distribution is not stopped, based on the number of impressions and the stop period for the distributed advertisement content that has been stopped.

  Here, the impression number estimation processing by the advertising device 200 will be described with reference to FIG. FIG. 8 is a diagram illustrating an example of impression number estimation processing by the advertising device 200 according to the second embodiment. FIG. 8 shows the relationship between the daily budget consumption rate and time for one advertisement content C11. The digestion allowable line B11 shown in FIG. 8 indicates the allowable value of the budget digestion rate at each time. For example, the advertising device 200 assigns a budget that can be consumed per day to the advertising content C11, and sets a budget digest line B11 that gradually digests the allocated budget. Moreover, the digest performance line R11 shown in FIG. 8 shows the actual budget digest rate of the advertising content C11.

  In this example, the advertising device 200 determines whether or not the current budget spending rate exceeds the budget spending line B11 every time a predetermined time has elapsed. Then, the advertising device 200 stops the distribution of the advertising content C11 for a predetermined time (for example, 5 minutes, 10 minutes, 20 minutes, etc.) when the budget consumption rate exceeds the budget consumption line B11. In the example of FIG. 8, the advertising device 200 stops the distribution of the advertising content C11 in the periods “t1” and “t2”. In the example of FIG. 8, it is assumed that the actual value of the number of impressions of the advertising content C11 is “N”. In other words, the advertisement content C11 is not distributed during the periods “t1” and “t2”, and is distributed “N” times by the advertising device 200. In this case, the distribution history storage unit 122 stores the number of impressions “N” and stop periods “t1” and “t2” as the distribution history of the advertisement content C11.

  In such a case, the advertising device 200 assumes that the distribution of the advertising content C11 is not stopped in the stop periods “t1” and “t2”, and calculates the number of impressions of the advertising content C11 when the distribution is not stopped. Predict. For example, the advertising device 200 proportionally distributes the number of impressions to the stop periods “t1” and “t2” according to the ratio of the stop periods “t1” and “t2” to one day (24 hours). Specifically, assuming that the total time of the suspension periods “t1” and “t2” is T [seconds], the advertising device 200 uses the following expression (15) to calculate the advertising content C11 when there is no delivery suspension. The number of impressions “Imps” is predicted. In addition, “86400” shown in the following formula (15) is the number of seconds in one day (24 hours).

  In this way, the advertising device 200 does not stop the distribution of the already-distributed advertising content whose distribution history is stored in the distribution history storage unit 122 and whose distribution has been stopped. Estimate the number of impressions at.

  Then, the advertising device 200 uses the actual value of the number of impressions of the distributed advertisement content that has never been stopped and the estimated value of the number of impressions of the already-distributed advertisement content that has been stopped. Is generated. That is, this prediction model is a model for predicting the relationship between the bid price of the advertisement content and the number of impressions when the distribution is not stopped by the advertisement budget.

  Then, the advertising device 200 predicts the relationship between the bid price and the number of impressions for each advertising content included in the prediction target campaign advertisement group using the prediction model. That is, since the advertising device 200 uses a prediction model on the assumption that distribution is not stopped, the advertising device 200 predicts a relationship in which the number of impressions continues to increase as the bid price increases. However, in practice, since the advertising budget is set for the campaign advertisement group to be predicted, the number of impressions of each advertisement content may be less than or equal to the maximum number of impressions. Therefore, the advertising device 200 predicts the maximum number of impressions of each advertising content by the same prediction process as the advertising device 100 according to the first embodiment.

  Then, the advertising device 200 provides the advertiser with the relationship between the bid price predicted using the prediction model and the number of impressions, and the maximum number of impressions predicted by the same processing as the advertising device 100. However, the advertising device 200 predicts the number of impressions corresponding to the bid price using the prediction model when the bid price of each advertisement content is received. And advertising device 200 makes the number of impressions small among prediction results of the number of impressions, and the maximum number of impressions a prediction result.

  As described above, the advertising device 200 according to the second embodiment does not simply use the actual value of the number of impressions stored as the distribution history, but uses the estimated value of the number of impressions considering whether or not the distribution is stopped. To predict the number of impressions corresponding to the bid price. For this reason, the advertising device 200 can predict the number of impressions with high accuracy even when the distribution history of the distributed advertising content that has been stopped is used.

  For example, the advertising device 200 can prevent the prediction of the number of impressions such that the increase rate decreases when the bid price becomes higher than a predetermined amount by estimating the number of impressions as described with reference to FIG. . Therefore, the advertising device 200 can make the advertiser recognize that the higher the bid price is, the more the number of impressions is, by providing the advertiser with the prediction result. As a result, the advertising device 200 can set an expensive bid price for an advertiser who desires a huge number of impressions, and can thus prevent losing an opportunity to obtain a large amount of advertising revenue.

  In the above example, the advertising device 200 may predict the number of impressions without using the distribution history of the distributed advertising content that has been stopped. However, when the amount of distributed advertisement content that has been stopped for distribution is enormous, the number of distribution histories used for prediction processing is reduced. For this reason, it is difficult for the advertising device 200 to predict the number of impressions with high accuracy when only the distribution history of the distributed advertising content that has never been stopped is used.

  Hereinafter, the configuration and processing procedure of the advertising device 200 described above will be described. In the following, a process for predicting the number of impressions in consideration of the characteristics of the advertisement content and the like in addition to the actual value and the estimated value of the number of impressions will be described. Moreover, since the prediction process of the maximum number of impressions by the advertising device 200 is the same as the example shown in the first embodiment, a description thereof will be omitted below. That is, in the following, prediction processing using a prediction model will be mainly described.

[2. Configuration of distribution device]
Next, the configuration of the advertising device 200 according to the second embodiment will be described with reference to FIG. FIG. 9 is a diagram illustrating a configuration example of the advertising device 200 according to the second embodiment. As illustrated in FIG. 9, the control unit 230 of the advertising device 200 includes an acquisition unit 234, an estimation unit 237, a prediction unit 235, and a provision unit 236, and functions and functions of information processing described below. Realize or execute. The internal configuration of the control unit 230 is not limited to the configuration illustrated in FIG. 9, and may be another configuration as long as the information processing described below is performed. Further, the connection relationship between the processing units included in the control unit 230 is not limited to the connection relationship illustrated in FIG. 9, and may be another connection relationship.

[2-1. (Acquisition Department)
The acquisition unit 234 acquires various types of information from the distribution history storage unit 122 when instructed by the submission accepting unit 133 to predict the advertisement effect. At this time, the acquisition unit 234 acquires a distribution history for each distribution date and each advertising content from the distribution history storage unit 122. Hereinafter, one distribution history (number of impressions, stop period, number of clicks, CTR, cost per click, and combination of targeting conditions) specified for each distribution date and time and each advertising content may be referred to as a “distribution history record”. . Note that the acquisition unit 234 does not include all the distribution history records stored in the distribution history storage unit 122, but a distribution history in which the distribution date and time is included in a predetermined period (for example, a period from the current date to a past number of days). Only records may be acquired.

[2-2. Estimator)
The estimation unit 237 uses the number of impressions acquired by the acquisition unit 234 and the stop period to estimate the number of impressions of the distributed advertisement content when the distribution is not stopped. Specifically, the estimation unit 237 extracts, from the distribution history records acquired by the acquisition unit 234, the distribution history record corresponding to the distributed advertisement content that has been stopped as a processing target. Then, the estimation unit 237 estimates the number of impressions when the distribution is not stopped for each distribution history record to be processed. For example, the estimation unit 237 estimates the number of impressions using the above equation (15).

  An example of the estimation process performed by the estimation unit 237 will be described. Here, it is assumed that the distribution history storage unit 122 is in the state shown in FIG. In this case, the acquisition unit 234 has a distribution history record corresponding to the distribution date and time “November 1, 2013” and the advertisement content C11, the impression number “7000”, and the stop period “12:30 to 12:50, 19:30 to 19:50 "is acquired. In this example, the total time in the stop period is “2400 seconds (40 minutes)”. Therefore, T in the above formula (15) is “2400”, and N in the above formula (15) is “7000”. For this reason, the estimation unit 237 estimates the number of impressions of the advertisement content C11 when the distribution is not stopped as “7200”. Similarly, the estimation unit 237 includes the number of impressions of the advertisement content C12 having the distribution date “November 1, 2013” and the number of impressions of the advertisement contents C11 and C12 having the distribution date “November 2, 2013”. Is estimated.

[2-3. (Predictor)
The prediction unit 235 uses the distribution history record acquired by the acquisition unit 234 and the number of impressions estimated by the estimation unit 237 to predict the number of impressions in each advertisement content included in the campaign advertisement group to be predicted. Specifically, the prediction unit 235 generates a prediction model for predicting the relationship between the bid price and the number of impressions. Further, the prediction unit 235 predicts the maximum number of impressions restricted by the advertising budget, similarly to the prediction unit 135 according to the first embodiment. Hereinafter, the prediction process by the prediction unit 235 will be described. Hereinafter, a case where the prediction unit 235 predicts the number of impressions per day will be described as an example.

[2-3 (1). (Prediction model generation process)
First, prediction model generation processing by the prediction unit 235 will be described. The prediction unit 235 performs a regression analysis using the number of impressions included in the distribution history record as a dependent variable (objective variable) and various information included in the distribution history record as an independent variable (explanatory variable), whereby the bid price and the number of impressions are determined. A prediction model for predicting the relationship with is generated. Specifically, the prediction unit 235 assigns various types of information to the following equation (16) for each distribution history record acquired by the acquisition unit 234.

In the above formula (16), “Imps ₁ ” indicates the number of impressions included in the distribution history record. “C” indicates an intercept (constant term). “A” represents a constant. Also, “eCPM (effective Cost Per Mill)” is an example of an index value indicating the value of the advertising content, and the multiplication result of “CPC (CPC)” and “CTR” included in the distribution history record is “ This corresponds to a value obtained by further multiplying 1000 ”. That is, “eCPM” = “CPC · CTR · 1000”.

“X _m ” (m = 1, 2,..., N) indicates whether or not the advertisement content characteristic is assigned and the advertisement content has the characteristic. For example, it is assumed that the targeting condition “male” is assigned to “x _m ”. In this case, the prediction unit 235 substitutes “1” for “x _m ” when the distribution history record includes the targeting condition “male”. On the other hand, when the targeting condition “male” is not included in the distribution history record, the prediction unit 235 substitutes “0” for “x _m ”. Further, the present invention is not limited to this example, and “x _m ” may be assigned an advertisement content category, title, or the like. In this case, “0” or “1” is substituted into “x _m ” based on the category, title, etc. of the advertising content corresponding to the distribution history record. “W _m ” (m = 1, 2,..., N) indicates a coefficient of “x _m ”.

Here, when the distribution history record corresponding to the distributed advertisement content that has been stopped for distribution is a processing target, the prediction unit 235 sets the estimation unit 237 to “Imps ₁ ” in the above equation (16). Substitute the number of impressions estimated by. For example, in the example of FIG. 5, the prediction unit 235 substitutes various information included in the distribution history record corresponding to the distribution date “November 1, 2013” and the advertisement content C11 into the above formula (16). . In this case, the prediction unit 235 substitutes “7200” estimated by the estimation unit 237 as in the above example, instead of “7000” in “Imps ₁ ” of the equation (16).

Similarly, the prediction unit 235 assigns various types of information to the above equation (16) for each distribution history record acquired by the acquisition unit 234. That is, when 1000 distribution history records are acquired by the acquisition unit 234, the prediction unit 235 generates 1000 of the above formula (16) into which various information is substituted. Then, the prediction unit 235 performs regression analysis (for example, multiple regression analysis) on the plurality of formulas (16) generated for each distribution history record, whereby “C”, “A”, “ w _m ”(m = 1, 2,..., n) is obtained. Then, the prediction unit 235 converts the above equation (16) after substituting the obtained “C”, “A”, and “w _m ” into the following equation (17).

Furthermore, since “eCPM” included in the equation (17) = “CPC · CTR · 1000”, the prediction unit 235 converts the equation (17) into the following equation (18). In the following formula (18), “C”, “A”, and “w _m ” are written, but actually, the predicting unit 235 uses “C”, “A”, Substitute for “w _m ”.

Then, the prediction unit 235 substitutes “CTR” and “x _m ” of Expression (18) for each advertisement content included in the campaign advertisement group to be predicted. For example, when the advertising content to be predicted has been distributed in the past, the prediction unit 235 substitutes the average value of the past CTR and the like into “CTR” in the above equation (18). Moreover, the prediction part 235 substitutes the value estimated by the predetermined | prescribed method to "CTR" of said Formula (18), when the advertising content of prediction object has not been delivered in the past. For example, the prediction unit 235 identifies the delivered advertisement content whose category is the same as the advertisement content to be predicted, and substitutes the average value of CTR or the like in the identified delivered advertisement content for “CTR” in the above equation (18). To do.

Further, for example, the prediction unit 235 sets “x _m ” (m = 1, 2) in the above formula (18) based on the targeting condition of the advertising content to be predicted, the category or title of the advertising content to be submitted, ,..., N) is substituted with “0” or “1”. For example, it is assumed that the targeting condition “male” is assigned to “x _m ” and the targeting condition received by the submission receiving unit 133 includes “male”. In this case, the prediction unit 235 assigns “1” to “x _m ”. On the other hand, in this example, it is assumed that “male” is not included in the targeting condition received by the submission receiving unit 133. In this case, the prediction unit 235 substitutes “0” for “x _m ”.

Here, “CPC (Cost Price)” and “Bid Price (Bid)” are expressed by the above formula (1). That is, when the above equation (1) is substituted into the above equation (18), the unknowns in the above equation (18) are “Imps ₁ (number of impressions) on the left side and“ Bid (bid price) ”on the right side. It becomes. In this manner, the prediction unit 235 generates a prediction model for predicting the relationship between the bid price and the number of impressions for each advertising content included in the campaign advertisement group to be predicted.

  In this example, since the number of impressions per day is stored in the distribution history storage unit 122, the above equation (18) is a prediction model for predicting the number of impressions per day. Note that the advertising model accepted by the submission accepting unit 133 is not considered in the prediction model represented by the above equation (18).

[2-3 (2). (Prediction process)
Subsequently, similarly to the prediction unit 135 according to the first embodiment, the prediction unit 235 predicts the maximum number of impressions for each advertising content included in the campaign advertisement group to be predicted. Then, the prediction unit 235 uses a smaller number of impressions as a prediction result among the number of impressions predicted by the prediction model and the maximum number of impressions predicted by the above equation (8). That is, the prediction unit 235 sets the result of the following formula (19) as the prediction result for each advertising content.

  For example, it is assumed that an advertisement budget is accepted by the submission accepting unit 133 but a bid price is not accepted. In this case, the prediction unit 235 uses both the relationship between the bid price represented by the above equation (18) and the number of impressions and the maximum number of impressions calculated by the above equation (5) as prediction results.

  In addition, for example, it is assumed that a bid price (Bid) is received in addition to the advertising budget by the submission receiving unit 133. In this case, the prediction unit 235 substitutes CPC obtained from the above equation (1) into the above equation (18). And the prediction part 235 makes a prediction result the small number of impressions among the prediction value of the said Formula (18), and the maximum number of impressions calculated | required by the said Formula (8) based on the said Formula (19).

  Here, FIG.10 and FIG.11 shows an example of the prediction result by the prediction part 235 which concerns on 2nd Embodiment. FIG. 10 shows a graph corresponding to the above equation (16) as a reference. FIG. 11 shows a graph corresponding to the equation (18).

  The vertical axis in FIG. 10 indicates the logarithm of the number of impressions. In addition, the horizontal axis of FIG. 10 indicates the value of “log (eCPM)” included in the right side of the equation (16). And the straight line graph g11 shown in FIG. 10 shows said Formula (16). Moreover, the straight line graph L11 shown in FIG. 10 shows the logarithm of the maximum number of impressions calculated by the above equation (8). In FIG. 10, a straight line graph L11 indicates the maximum number of impressions in one day. That is, the logarithm of the prediction result represented by the above equation (19) corresponds to the prediction result graph g12 indicated by a thick line in FIG. Here, it is assumed that the constant “A” included in the equation (16) is “0 <A <1”.

  The vertical axis in FIG. 11 indicates the number of impressions. That is, the vertical axis in FIG. 11 represents the predicted value of the number of impressions in one day. In addition, the horizontal axis of FIG. 11 indicates “Bid (bid price)” included in the right side of the equation (18) into which the equation (1) is substituted.

  The curve graph g21 shown in FIG. 11 corresponds to the prediction model represented by the above equation (18). A straight line graph L21 shown in FIG. 11 shows the maximum number of impressions calculated by the above equation (8). That is, the prediction result represented by the above formula (19) corresponds to the prediction result graph g22 indicated by a thick line in FIG.

  As the prediction price graph g22 indicated by a thick line in FIG. 11 increases, the number of impressions increases as the bid price increases, but reaches the maximum number of impressions when the bid price exceeds a certain amount. In other words, when the bid price set by the advertiser is below a certain amount, the number of impressions does not reach the maximum number of impressions.

[2-4. (Providing department)
Returning to the description of FIG. 9, the providing unit 236 provides the advertiser device 20 with the prediction result predicted by the prediction unit 235. For example, if the bid price is not accepted by the submission accepting unit 133, the providing unit 236 provides the advertiser device 20 with the prediction result g22 illustrated in FIG. For example, when the bid accepting price is accepted by the submission accepting unit 133, the providing unit 236 provides the advertiser device 20 with the number of impressions predicted by the above formula (19).

[3. (Modification)
The advertising device 200 described above may be implemented in various different forms other than the above embodiment. Therefore, in the following, another embodiment of the advertising device 200 will be described.

[3-1. Impression number estimation process)
In the above embodiment, as described with reference to FIG. 8, an example in which the estimation unit 237 estimates the number of impressions of the distributed advertising content when there is no distribution stop by proportionally distributing the number of impressions during the stop period. showed that. However, the estimation unit 237 may estimate the number of impressions of the distributed advertisement content when there is no distribution stop by another method.

  This point will be specifically described. In general, when the distribution mode of the advertisement content is different, the distribution of the number of impressions at each time (hereinafter referred to as “time distribution of the number of impressions”) is different for each distribution mode. For example, the time distribution of the number of impressions is not always constant and varies depending on the time zone. In general, the time distribution of the number of impressions varies depending on the day of the week. As an example, the number of impressions of advertisement content distributed on weekdays and the number of impressions of advertisement content distributed on holidays may be different even in the same time zone. Therefore, the advertising device 200 may store, for example, the time distribution of the number of impressions on weekdays and the time distribution of the number of impressions on holidays. Then, the estimation unit 237 may estimate the number of impressions of the distributed advertisement content based on the time distribution corresponding to the day of the week on which the distributed advertisement content is distributed among the time distributions.

  An example will be described with reference to FIGS. 12 and 13 are diagrams illustrating an example of impression number estimation processing by the advertising device 200 according to the modification. The graphs shown in FIGS. 12 and 13 show the average value of the number of impressions measured every second. However, the graph of FIG. 12 is formed based on the number of impressions measured on weekdays, and FIG. 13 is formed based on the number of impressions measured on holidays. For example, the advertising device 200 counts the number of advertisement contents distributed by the distribution unit 132 every second, and divides the counting result by the total number of advertisement contents that can be distributed to thereby determine the number of impressions in one advertisement content ( That is, the average value of the number of distributions) is calculated. The advertising device 200 accumulates the time distribution of the number of impressions shown in FIGS. 12 and 13 by performing such processing separately on weekdays and holidays.

In this case, the acquisition unit 234 acquires the time distribution of the number of impressions accumulated by the advertising device 200 (see FIG. 12 and FIG. 13) when the estimation unit 237 performs the impression number estimation process. Here, it is assumed that the estimation unit 237 sets the distributed advertisement content distributed on weekdays as a processing target. At this time, the estimation unit 237 refers to the time distribution of the number of impressions corresponding to weekdays (FIG. 12) among the time distribution of the number of impressions acquired by the acquisition unit 234. Then, it is assumed that the distributed advertisement content to be processed is stopped at the distribution stop start time t11 shown in FIG. 12 and is restarted at the distribution stop end time t12. Furthermore, it is assumed that the distributed advertisement content to be processed is stopped at the distribution stop start time t13 shown in FIG. 12, and is restarted at the distribution stop end time t14. In this case, the estimation unit 237 calculates the total number of impressions in the stop period t3 (= t12−t11). Then, the estimation unit 237 calculates the total number of impressions per day (24 hours) by adding all the impressions per second based on the graph shown in FIG. Then, the estimation unit 237 calculates a ratio X ₁ % of the total number of impressions in the stop period t3 with respect to the total number of impressions per day. Similarly, the estimation unit 237 calculates the total number of impressions in the stop period t4 (= t14−t13). Then, the estimation unit 237 calculates a ratio X ₂ % of the total number of impressions in the suspension period t4 with respect to the total number of impressions in one day. Then, the estimation unit 237 performs processing according to the following equation (20), where N is the actual value of the number of impressions of the distributed advertising content, and X is the sum of the ratios X ₁ and X ₂ (= X ₁ + X ₂ ). Estimate the number of impressions for the targeted ad content.

  Further, for example, it is assumed that the estimation unit 237 sets a distributed advertisement content distributed on a holiday as a processing target. In this case, the estimation unit 237 refers to the time distribution of the number of impressions corresponding to a holiday (FIG. 13) among the time distribution of the number of impressions acquired by the acquisition unit 234. Then, it is assumed that the distributed advertisement content to be processed is stopped at the distribution stop start time t15 shown in FIG. 13 and restarted at the distribution stop end time t16. In this case, the estimation unit 237 calculates the total number of impressions in the stop period t5 (= t16−t15). Then, the estimation unit 237 calculates a ratio X% of the total number of impressions in the stop period t5 with respect to the total number of impressions per day. Then, the estimation unit 237 estimates the number of impressions of the processed advertisement content to be processed by the above formula (20), where N is the actual value of the number of impressions of the distributed advertisement content.

  In this way, the estimation unit 237 may estimate the number of impressions of the distributed advertisement content in consideration of the variation in the number of impressions according to the distribution mode of the advertisement content. Thereby, the estimation unit 237 can estimate the number of impressions of the distributed advertisement content when the distribution is not stopped with high accuracy. As a result, the prediction unit 235 can predict the number of impressions in the advertisement content scheduled to be submitted with high accuracy.

  In the above example, the day of the week on which the advertisement content is distributed has been described as an example of the distribution mode of the advertisement content. However, even if other delivery modes other than the day of the week are different, the time distribution of the number of impressions may be different. For example, the time distribution of the number of impressions may differ for each user attribute of the advertisement content delivery destination user. For example, although the number of impressions of advertisement content distributed using the user attribute “10's” as the targeting condition increases in the afternoon time zone, the advertisement distributed using the user attribute “60's” as the targeting condition The number of content impressions may increase during the morning hours. Therefore, the advertising device 200 may accumulate the time distribution of the number of impressions as illustrated in FIG. 12 or 13 for each user attribute of the advertising content distribution destination user. In this case, the estimation unit 237 estimates the number of impressions of the distributed advertisement content based on the time distribution corresponding to the user attribute of the distribution destination user of the distributed advertisement content among the time distributions.

[3-2. (Predicted advertising effect)
Moreover, in the said 2nd Embodiment, the advertising device 200 showed the example which estimates the number of impressions as an example of an advertising effect. However, the advertising device 200 may predict other advertising effects instead of the number of impressions. For example, the advertising device 200 may predict the number of clicks in each advertisement content included in the campaign advertisement group scheduled to be submitted. In this case, the acquisition unit 234, the estimation unit 237, and the prediction unit 235 can predict the number of clicks in the advertisement content scheduled to be submitted by using the number of clicks instead of the number of impressions in the processing described above.

  Specifically, the estimation unit 237 estimates the number of clicks when the distribution is not stopped in the stop period for each distribution history record to be processed among the distribution history records acquired by the acquisition unit 234. For example, the estimation unit 237 estimates the number of clicks by the method described with reference to FIG. 8 or the method described with reference to FIGS.

Further, the prediction unit 235 performs regression analysis using the following equation (21) instead of the above equation (16). In the following formula (21), “Click ₁ ” corresponds to the number of clicks acquired by the acquisition unit 234 or the number of clicks estimated by the estimation unit 237.

  And the prediction part 235 converts the said Formula (21) into following formula (22) and Formula (23) similarly to the said Formula (17) and (18).

  Then, the prediction unit 235 predicts the number of clicks in the advertising content scheduled to be submitted by using the following equation (24) instead of the above equation (19).

  Here, FIG.14 and FIG.15 shows an example of the prediction result by the prediction part 235 which concerns on a modification. FIG. 14 shows a graph corresponding to the above equation (21) for reference. FIG. 15 shows a graph corresponding to the above equation (23).

  The vertical axis in FIG. 14 indicates the logarithm of the number of clicks. In addition, the horizontal axis of FIG. 14 indicates the value of “log (eCPM)” included in the right side of the equation (21). And the linear graph g31 shown in FIG. 14 shows the said Formula (21). A straight line graph L31 shown in FIG. 14 shows the logarithm of the maximum number of clicks calculated by the above equation (12). That is, the logarithm of the prediction result represented by the above equation (24) corresponds to the prediction result graph g32 indicated by a thick line in FIG. Here, it is assumed that the constant “A” included in the formula (21) is “0 <A <1”.

  The vertical axis in FIG. 15 indicates the number of clicks. That is, the vertical axis in FIG. 15 indicates the predicted number of clicks in one day. Further, the horizontal axis of FIG. 15 indicates “Bid (bid price)” included in the right side of the equation (23) to which the equation (1) is substituted.

  And the curve graph g41 shown in FIG. 15 corresponds to the prediction model represented by the said Formula (23). A straight line graph L41 illustrated in FIG. 15 indicates the maximum number of clicks calculated by the above equation (12). That is, the prediction result graph represented by the above formula (24) corresponds to the prediction result graph g42 indicated by a thick line in FIG.

[3-3. Example of providing prediction results)
In the second embodiment, the providing unit 236 may provide the advertiser device 20 with the prediction result from the prediction unit 235 on a screen. FIG. 16 shows an example of a prediction result screen provided by the providing unit 236 according to the modification. As illustrated in FIG. 16, the providing unit 236 provides a prediction result screen W10 including an input field B11, prediction result frames P10 and P20, and a prediction graph frame P30. Although not shown in FIG. 16, the prediction result screen W10 includes an input field for inputting a campaign advertisement group to be submitted, and a bid price and a targeting condition for each advertisement content. Includes input fields.

  The input field B11 accepts an input of an advertisement budget set in a batch for the campaign advertisement group. The prediction result frame P10 indicates the prediction result of each advertisement content included in the campaign advertisement group. Here, it is assumed that the prediction unit 235 and the like predict the number of impressions and the number of clicks as the advertisement effect. In addition, it is assumed that a bid price for each advertising content is input by the advertiser. In this case, in the prediction result frame P10, the predicted number of clicks predicted by the prediction unit 235, the predicted value of costs, and the predicted value of impressions are displayed for each advertisement content. Note that the prediction unit 235 obtains a predicted cost value by multiplying the predicted number of clicks by the unit cost of click. In the example of FIG. 16, the providing unit 236 displays a predicted value such as the number of impressions in one month in the prediction result frame P10. Here, since the number of impressions per day is predicted by the prediction unit 235, the providing unit 236 obtains the number of impressions per month by multiplying the prediction result by the prediction unit 235 by “30” or the like.

  Here, when any of the advertisement contents displayed in the prediction result frame P10 is selected by the advertiser, the providing unit 236 displays the prediction result corresponding to the selected advertisement content as the prediction result frame P20 and the prediction graph. It is displayed on the frame P30. Specifically, the providing unit 236 displays, for each predetermined bid price, a list in which the predicted number of clicks, the predicted value of costs, and the predicted value of impressions are associated with each other in the prediction result frame P20. Further, the providing unit 236 may display an input field B12 for setting a bid price in the prediction result frame P20 as in the example of FIG.

  Further, as illustrated in FIG. 16, the providing unit 236 displays a graph indicating the relationship between the bid price and the number of impressions (for example, the prediction result graph g22 illustrated in FIG. 11) in the prediction graph frame P30.

  Thereby, the advertiser inputs the advertising budget in the input field B11, so that the advertising effect such as the number of impressions and the number of clicks for each advertising content included in the campaign advertisement group before officially submitting the campaign advertisement group. Can be grasped.

[3-4. (Prediction model)
Moreover, in the said embodiment, although the prediction model by the prediction part 235 was shown by the graph (For example, prediction result graph g22, prediction result graph g42) in FIG.11 and FIG.15, the shape of a graph is restricted to the said example. Absent. This point will be described with reference to FIGS. 17 and 18. 17 and 18 show an example of a prediction result by the prediction unit 235 according to the modification.

  As described above, when the constant “A” included in the equation (18) is “0 <A <1”, the curve graph g21 illustrated in FIG. 11 is obtained. However, when the constant “A” is “A> 1,” the prediction model represented by the above equation (18) is a curve graph g51 shown in FIG. That is, the prediction result represented by the above equation (19) corresponds to the prediction result graph g52 indicated by the thick line in FIG. Similarly, when the constant “A” is “0 <A <1”, the curve graph g41 shown in FIG. 15 is obtained, but when the constant “A” is “A> 1,” the above-described curve graph g41 is obtained. The prediction model represented by Expression (23) has the same shape as the curve graph g51 illustrated in FIG.

  Moreover, in the said embodiment, although the prediction part 235 showed the example which uses the said Formula (16) and Formula (21), the formula which the prediction part 235 uses is not restricted to this example. For example, the prediction unit 235 may use the following equation (25) having “log {log (eCPM)}” as an explanatory variable instead of the above equation (16). However, in the following formula (15), it is assumed that “eCPM> 1”.

In this case, the prediction unit 235 converts the equation (25) into the following equations (26) and (27), similarly to the equations (17) and (18). Note that CPC = F ⁻¹ (Bid) obtained by the above equation (1) is substituted into the following equation (27). However, in the following formula (27), it is assumed that “eCPM (= CPC · CTR · 1000)> 1”. That is, in the following formula (27), it is assumed that “Bid> F ⁻¹ {1 / (CTR · 1000)}”.

When the constant “A” included in the equation (27) is “0 <A <1”, the prediction model represented by the equation (27) is the same as the curve graph g21 illustrated in FIG. It becomes a shape. In FIG. 11, “Bid> 0”, but in the case of the prediction model represented by the above equation (27), “Bid> F ⁻¹ {1 / (CTR · 1000)}”. However, when the constant “A” included in the equation (27) is “A> 1,” the prediction model represented by the equation (27) is the curve graph g61 illustrated in FIG. That is, the prediction result represented by the above formula (19) corresponds to the prediction result graph g62 indicated by a thick line in FIG.

[3-5. (Restriction of delivery)
In the second embodiment, the example in which the distribution is stopped due to the limitation of the advertisement budget has been described as the distribution limitation of the advertisement content. However, the distribution unit 132 may restrict the distribution of the advertising content regardless of whether or not the advertising budget is consumed. Specifically, when a plurality of advertising spaces are included in a web page, the distribution unit 132 may restrict the display of two or more advertisement contents of the same advertiser on the same web page. For example, it is assumed that the distribution unit 132 specifies the advertisement contents C11 and C12 as distribution targets based on the unit price of CPC, CTR, and targeting conditions. At this time, when the advertisers of the advertising contents C11 and C12 are the same and the advertising contents C11 and C12 are displayed in different advertising spaces on the same web page, the distribution unit 132 displays the advertising contents C11 and C12. And the advertising content corresponding to the other advertisers is distributed. In this example, when the advertising content C11 is distributed, the advertising content C12 is not distributed due to distribution restrictions although it should have been originally distributed. That is, the number of impressions of the advertisement content C12 is suppressed by the distribution restriction. Hereinafter, such distribution restriction may be referred to as “duplication elimination processing”.

The estimation unit 237 and the prediction unit 235 described above may estimate the number of impressions in consideration of such deduplication processing. That is, the estimation unit 237 and the prediction unit 235 may estimate the number of impressions (or the number of clicks) of the distributed advertisement content when the deduplication process is not performed. For example, the estimation unit 237 refers to the distribution history storage unit 122, and counts the number of advertisement contents whose distribution date is the same day and whose advertiser is the same. Then, the prediction unit 235 may add the reciprocal of the counting result (here, “ADID number”) by the estimation unit 237 to the feature of the equation (16). Specifically, the prediction unit 235 may set the right side of the equation (16) as “... + W _n x _n + w _{n + 1} (1 / ADID number)”.

Further, in the above example, the prediction unit 235 may add a feature considering the bid price (Bid) to the above equation (16). This is because the advertising content has different probabilities of being distributed by the distribution unit 132 according to the bid price (Bid). For example, the prediction unit 235 may set the right side of the equation (16) as “... + W _n x _n + w _{n + 1} (1 / ADID number) · (Bid / ΣBid)”. Note that “ΣBid” indicates the total bid price (Bid) corresponding to the advertising content having the same delivery date and the same advertiser. Further, the bid price (Bid) here may be a cost per click (CPC).

[3-6. (Unit of prediction model)
Moreover, in the said 2nd Embodiment, the example which produces | generates a prediction model based on the delivery history memorize | stored in the delivery history memory | storage part 122 was shown. Here, the prediction unit 235 may generate a prediction model only from the distributed advertisement content satisfying a predetermined condition among the distributed advertisement contents stored in the distribution history storage unit 122. Then, the prediction unit 235 may generate a prediction model for each predetermined condition. The predetermined condition here is, for example, (Y1) that the targeting condition includes gender, (Y2) that the targeting condition includes age, and (Y3) the targeting condition includes a region (for example, the user's location area). Is included. And the prediction part 235 produces | generates a prediction model based on the delivery history of the delivered advertisement content which satisfy | fills conditions (Y1), for example. Similarly, the prediction unit 235 generates a prediction model corresponding to the condition (Y2) and a prediction model corresponding to the condition (Y3). Furthermore, the prediction unit 235 generates a prediction model based on the distribution history of the distributed advertisement content that satisfies the condition (Y1) and the condition (Y2). Thus, the prediction unit 235 generates a prediction model corresponding to a plurality of conditions. That is, when six predetermined conditions exist, the prediction unit 235 generates “2 ⁶ (= 64)” prediction models. In this case, the prediction unit 235 predicts the advertising effect using a prediction model corresponding to the condition that the advertising content scheduled to be submitted satisfies. Thereby, since the prediction unit 235 can use a prediction model that matches the characteristics of the advertisement content scheduled to be submitted, the advertisement effect can be predicted with high accuracy.

(Other embodiments)
Each embodiment mentioned above may be implemented with various different forms other than the above-mentioned embodiment. Accordingly, other embodiments will be described below. In the following, other embodiments will be described mainly using the advertising device 100 as an example, but other embodiments described below can also be applied to the advertising device 200.

[1. (Prediction process according to distribution history)
In the first embodiment, it is assumed that a combination of advertisement contents included in a campaign advertisement group to be predicted has been distributed as a campaign advertisement group in the past. However, the advertisement content included in the campaign advertisement group to be predicted is not always distributed as the campaign advertisement group. Therefore, the advertising device 100 predicts the maximum number of impressions using the prediction model described in the second embodiment when the campaign advertisement group to be predicted has not been distributed as a campaign advertisement group in the past. May be. Hereinafter, this point will be specifically described. In the following, it is assumed that the advertising device 100 includes the estimation unit 237.

  First, the advertising device 100 stores the prediction target campaign advertisement group in the distribution history storage unit 122 as described in the second embodiment when the campaign advertisement group has not been distributed as a campaign advertisement group in the past. Based on the distribution history, a prediction model represented by the above equation (18) or the like is generated.

  Specifically, when the acquisition unit 134 according to the modification is instructed by the submission receiving unit 133 to predict the advertising effect, when the campaign advertisement group to be predicted has not been delivered in the past, A distribution history for each distribution date and each advertising content is acquired from the distribution history storage unit 122. For example, the acquisition unit 134 determines whether or not the campaign advertisement group to be predicted satisfies both (Condition 1) and (Condition 2) described above. And the acquisition part 134 acquires the delivery log | history for every delivery date and every advertising content, when any one of conditions is not satisfy | filled.

Then, the prediction unit 135 predicts the number of impressions of each advertisement content included in the campaign advertisement group to be predicted using the above prediction model. That is, the prediction unit 135 predicts the number of impressions of each advertisement content when distribution is not stopped. Then, the prediction unit 135 predicts the maximum advertising budget _ni that each advertising content can consume by substituting the predicted value of the number of impressions of each advertising content into the following formula (28).

  Note that when the advertising content to be predicted has been distributed in the past, the prediction unit 135 substitutes the past click price (CPC), the average value of CTR, and the like into the above equation (28). Moreover, the prediction part 135 substitutes the click price (CPC) and CTR predicted by a predetermined method in the above formula (28) when the advertising content to be predicted has not been delivered in the past. Moreover, the prediction part 135 may substitute the received bid price into said Formula (28), when the bid price of each advertisement content is received.

Here, the equation (28) will be described. As described above, the prediction unit 135 predicts the number of impressions using the prediction model. The number of impressions predicted here is based on the premise that “there is no restriction by the advertising budget”, “the distribution is not stopped”, and “the number of impressions per day”. That is, the prediction unit 135 can predict the number of impressions when the distribution conditions are the same by using the above-described prediction model. Since the condition is that “there is no restriction by the advertising budget”, the advertising budget _ni predicted by the above equation (28) can be said to be the maximum advertising budget that each advertising content can consume.

Then, the prediction unit 135 predicts the number of impressions for each advertising content by the above equation (28), and sets the ratio of the prediction results as the digestion ratio. That is, the prediction unit 135 obtains the maximum number of impressions _i (max (Imps _i )) of the advertising content _ni included in the prediction target campaign advertisement group by the following equation (29).

  Note that the prediction unit 135 can convert the equation (29) into the following equations (30) and (31).

  In this way, the advertising device 100 can predict the maximum number of impressions of each advertising content with high accuracy even in a campaign advertising group including advertising content that has never been distributed as a campaign advertising group in the past.

[2. Estimating the maximum number of impressions)
Moreover, in the said 2nd Embodiment, the example which makes the theoretical value calculated | required by the said Formula (8) the maximum number of impressions was shown. However, the prediction unit 235 may also predict the maximum number of impressions by regression analysis or the like based on the distribution history stored in the distribution history storage unit 122. In this case, the prediction unit 235 uses the number of impressions estimated by the estimation unit 237 for the number of impressions of the distributed advertisement content that has been stopped in the past.

[3. Bid and CPC]
In each of the above embodiments, the bid price (Bid) and the click unit price (CPC) are different from each other as in the above formula (1). However, each of the embodiments described above can also be applied to a system in which the bid price (Bid) and the click unit price (CPC) are the same. In this case, the advertisement content storage unit 121 stores only one of the bid price (Bid) and the click unit price (CPC).

[4. Web page〕
Further, in each of the above embodiments, the advertisement content displayed on the web page has been described as an example. However, the advertising device 100 may distribute the advertising content displayed in the application installed in the terminal device 10 as the advertising content displayed in the electronic book or video stored in the terminal device 10. Then, the advertising device 100 may predict the advertising effect of the advertising content displayed on the electronic book or the like by performing the above-described prediction processing.

[5. Device configuration〕
Moreover, in the said embodiment, the example which the advertising device 100 delivers advertisement content was shown. However, the device that predicts the advertising effect and the device that distributes the advertising content may be different devices. For example, the advertising device 100 described above may perform only the process of predicting the advertising effect without distributing the advertising content. That is, the advertising device 100 illustrated in FIG. 3 may not include the receiving unit 131, the distribution unit 132, the submission receiving unit 133, the advertisement content storage unit 121, and the distribution history storage unit 122. In this case, the advertisement distribution apparatus that distributes the advertisement content includes the receiving unit 131, the distribution unit 132, the submission receiving unit 133, the advertisement content storage unit 121, and the distribution history storage unit 122 illustrated in FIG. In this example, the advertising device 100 acquires a distribution history from the advertisement distribution device.

  Further, the advertising device 100 illustrated in FIG. 3 may not include the advertising content storage unit 121 and the distribution history storage unit 122. Specifically, the advertising device 100 may be connected to a database server that holds the advertising content storage unit 121 and the distribution history storage unit 122.

[6. Others]
In addition, among the processes described in the above embodiment, all or part of the processes described as being automatically performed can be performed manually, or the processes described as being performed manually can be performed. All or a part can be automatically performed by a known method. In addition, the processing procedures, specific names, and information including various data and parameters shown in the document and drawings can be arbitrarily changed unless otherwise specified. For example, the various types of information illustrated in each drawing is not limited to the illustrated information.

  Further, each component of each illustrated apparatus is functionally conceptual, and does not necessarily need to be physically configured as illustrated. In other words, the specific form of distribution / integration of each device is not limited to that shown in the figure, and all or a part thereof may be functionally or physically distributed or arbitrarily distributed in arbitrary units according to various loads or usage conditions. Can be integrated and configured. For example, the reception unit 131 and the distribution unit 132 illustrated in FIG. 3 may be integrated, or the acquisition unit 134 and the prediction unit 135 may be integrated.

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

[7. Hardware configuration)
In addition, the advertising devices 100 and 200 according to the above-described embodiments are realized by a computer 1000 configured as shown in FIG. 19, for example. Hereinafter, the advertisement apparatus 100 will be described as an example. FIG. 19 is a hardware configuration diagram illustrating an example of a computer 1000 that implements the functions of the advertising device 100. The computer 1000 includes a CPU 1100, RAM 1200, ROM 1300, HDD 1400, communication interface (I / F) 1500, input / output interface (I / F) 1600, and media interface (I / F) 1700.

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

  The HDD 1400 stores programs executed by the CPU 1100, data used by the programs, and the like. The communication interface 1500 receives data from other devices via the communication network 50 (corresponding to the network N shown in FIG. 2), sends the data to the CPU 1100, and transmits the data generated by the CPU 1100 to the other devices via the communication network 50. Send to.

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

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

  For example, when the computer 1000 functions as the advertising device 100 according to the first embodiment, the CPU 1100 of the computer 1000 implements the function of the control unit 130 by executing a program loaded on the RAM 1200. Further, the HDD 1400 stores data in the advertisement content storage unit 121 and the distribution history storage unit 122. The CPU 1100 of the computer 1000 reads these programs from the recording medium 1800 and executes them. However, as another example, these programs may be acquired from other devices via the communication network 50.

  When the computer 1000 functions as the advertising device 200 according to the second embodiment, the CPU 1100 of the computer 1000 implements the function of the control unit 230 by executing a program loaded on the RAM 1200.

  As described above, some of the embodiments of the present application have been described in detail with reference to the drawings. However, these are merely examples, and various modifications, including the aspects described in the disclosure section of the invention, based on the knowledge of those skilled in the art, It is possible to implement the present invention in other forms with improvements.

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

DESCRIPTION OF SYMBOLS 1 Advertisement delivery system 100 Advertising apparatus 131 Receiving part 132 Delivery part 133 Submission acceptance part 134 Acquisition part 135 Prediction part 136 Provision part

Claims (13)

A receiving means for receiving an advertising budget, which is a budget for an advertising fee set for a plurality of advertising contents,
An acquisition means for acquiring a distribution history including an advertising effect of a previously distributed advertisement content in a state where an advertisement budget is set together with an arbitrary advertisement content ,
The ratio of the advertising budget that is consumed by the plurality of advertising contents out of the advertising budget received by the receiving means based on the advertising budget that is consumed to obtain the advertising effect included in the distribution history acquired by the acquiring means And a prediction means for individually predicting the upper limit of the advertising effectiveness in the plurality of advertising contents according to the calculated digestion ratio,
An advertising device comprising: The acquisition means includes
When the plurality of advertising contents have been distributed in a state where an arbitrary advertising budget is set in a batch, a distribution history of the distributed advertising contents corresponding to the plurality of advertising contents is acquired.
The advertising device according to claim 1. The acquisition means includes
Each advertising content included in the plurality of advertising content may have been distributed in the past in a state where an arbitrary advertising budget is set in combination with any of the other advertising content included in the plurality of advertising content, In addition, when each combination of advertisement content distributed in the past includes the same advertisement content as at least one other combination, a distributed advertisement corresponding to each combination of advertisement content distributed in the past Get content distribution history,
The prediction means includes
For each combination of advertisement content distributed in the past, a digest ratio of an advertisement budget that is digested by each advertisement content included in the combination is obtained, and the digest ratio for each combination is calculated based on the same advertisement content. Find the spending ratio of the advertising budget consumed by multiple ad content,
The advertising device according to claim 1 or 2, characterized in that. Based on the distribution history acquired by the acquisition means, further comprising estimation means for estimating the advertising effect of the distributed advertising content,
The acquisition means includes
Obtaining the distribution history including restriction information that is information on which distribution of the distributed advertisement content is restricted;
The estimation means includes
Using the advertising effect and restriction information included in the distribution history acquired by the acquisition means, to estimate the advertising effect of the distributed advertising content when the distribution is not limited,
The prediction means includes
Further predicting the advertising effect of the plurality of advertising contents when the distribution is not limited based on the advertising effect included in the distribution history acquired by the acquiring unit and the advertising effect estimated by the estimating unit;
The advertising device according to any one of claims 1 to 3. The prediction means includes
If any of the plurality of advertisement contents has never been distributed in a state where an arbitrary advertisement budget is set in a batch, the advertising effect of the plurality of advertisement contents when the distribution is not restricted is predicted Predicting the digestion ratio based on the advertising effectiveness
The advertising device according to claim 4, wherein: The acquisition means includes
As the restriction information, obtain the distribution history including a stop period that is a period in which the distribution of the distributed advertisement content is stopped,
The estimation means includes
Using the suspension period to estimate the advertising effectiveness of the delivered advertising content when delivery is not stopped,
The advertising device according to claim 4 or 5, characterized in that. The acquisition means includes
Obtaining the distribution history further including a distribution period in which the distributed advertisement content is distributed;
The estimation means includes
Estimating the advertising effect of the delivered advertising content according to the ratio of the suspension period to the delivery period;
The advertising device according to claim 6. The acquisition means includes
For each predetermined distribution mode, obtain the time variation of the advertising effect in the distributed advertising content distributed in the distribution mode,
The estimation means includes
Estimating the advertisement effect of the distributed advertisement content based on the advertisement effect corresponding to the stop time of the distributed advertisement content among the time variation of the advertisement effect corresponding to the distribution mode;
The advertising device according to claim 6. The acquisition means includes
As the distribution mode, a distribution time when the distributed advertisement content is distributed, or a user attribute that is a distribution target of the distributed advertisement content is used.
The advertising device according to claim 8. Providing means for providing an advertising effect in the plurality of advertisement contents predicted by the prediction means to an advertiser of the plurality of advertisement contents;
The advertising device according to claim 1, further comprising: The prediction means includes
As the advertising effect in the plurality of advertising contents, predict the number of times that the plurality of advertising contents are displayed on the user's terminal device, or the number of times that the user is selected.
The advertising device according to any one of claims 1 to 10, wherein A prediction method performed by a computer,
A process of accepting an advertising budget, which is an advertising fee budget set for a plurality of advertising contents at once,
An acquisition process for acquiring a distribution history including an advertising effect of a previously distributed advertisement content in a state where an advertisement budget is set together with an arbitrary advertisement content ,
Based on the advertising budget digested to obtain the advertising effect included in the distribution history acquired by the acquisition step, the ratio of the advertising budget digested by the plurality of advertising contents out of the advertising budget accepted by the acceptance step And a prediction step for individually predicting the upper limit of the advertising effectiveness in the plurality of advertising contents according to the calculated digestion ratio,
The prediction method characterized by including. Acceptance procedure to accept the advertising budget, which is the budget of the advertising fee set for multiple advertising contents at once,
An acquisition procedure for acquiring a delivery history including the advertising effectiveness of previously delivered ad content with an ad budget set together with any ad content ,
Ratio of advertising budget consumed by the plurality of advertising contents out of the advertising budget received by the receiving procedure based on the advertising budget digested to obtain the advertising effect included in the distribution history acquired by the acquiring procedure And a forecasting procedure for individually predicting the upper limit of the advertising effectiveness in the plurality of advertising content according to the calculated digestion ratio,
A prediction program that is executed by a computer.

Images