JP2013161358A - Advertisement evaluation apparatus, advertisement evaluation method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To properly evaluate advertisement.SOLUTION: An advertisement evaluation calculation unit 232 counts, for advertisement, the number of impression logs corresponding to the advertisement from an impression log storage unit 253, to define it as the number of displays; counts the number of click logs corresponding to the advertisement from a click log storage unit 254 to define it as the number of clicks; divides a value obtained by adding 1 to the number of actions by a value obtained by adding 2 to the number of displays to calculate an expectation value xCTR of a probability that the advertisement may be clicked; and defines the xCTR or a value obtained by multiplying a bidding amount to xCTR, as an evaluation value of the advertisement.

Description

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

  Since the click rate (also referred to as CTR; Click Through Rate) of an advertisement displayed on a web page varies depending on the advertisement, CTR may be used as an evaluation value of the advertisement. Further, Patent Document 1 proposes that an advertisement is evaluated by profit efficiency (eCPM; effective cost per mill) obtained by multiplying a CTR by a bid result.

Japanese Patent No. 4431058

  In order to properly evaluate the ad, it is necessary to correctly estimate the click rate. However, since the conventional CTR was calculated by dividing the number of clicks on the advertisement (clicks) by the number of times the advertisement was actually displayed (number of impressions) based on past performance, While the number of times is small, it is difficult to grasp an accurate click rate, and the advertisement is evaluated too large or too small, so that a proper advertisement cannot be evaluated.

  The present invention has been made in view of such a background, and an object of the present invention is to provide an advertisement evaluation device, an advertisement evaluation method, and a program that can appropriately evaluate an advertisement.

  A main invention of the present invention for solving the above-mentioned problems is an apparatus for evaluating an advertisement, a display number acquisition unit for acquiring the number of display times of displaying the advertisement, and a viewer taking action in response to the advertisement. An action number acquisition unit for acquiring the number of actions, and a Bayesian estimation of a probability that the viewer will take the action in response to the advertisement in the case of the action number and the display number with a predetermined beta distribution as a prior distribution. An expected value calculation unit that calculates an expected value of the posterior distribution and an advertisement evaluation unit that determines an evaluation value of the advertisement according to the expected value are provided.

  The advertisement evaluation device of the present invention further includes an upper limit storage unit that stores the upper limit value of the probability, the prior distribution is a beta distribution Be (a, b), and the advertisement evaluation unit A value obtained by giving an incomplete beta function the first parameter obtained by adding a to the number of actions, a first parameter obtained by adding a to the number of actions, and a second parameter obtained by adding b to the value obtained by subtracting the number of actions from the display number, A quotient is calculated by dividing the normalized incomplete beta function with the upper limit, a value obtained by adding 1 to the first parameter, and a value given the second parameter, and the quotient is set as the expected value. The evaluation value may be determined according to the multiplied value.

  In addition, the advertisement evaluation device of the present invention further includes a reward amount acquisition unit that acquires a reward amount for the action of the viewer in response to the advertisement, and the advertisement evaluation unit adds the reward to the expected value. The evaluation value may be calculated by multiplying the amount.

  According to another aspect of the present invention, a display frequency acquisition unit that acquires a display frequency of displaying the advertisement, an action frequency acquisition unit that acquires the number of times the viewer has acted according to the advertisement, An expected value calculation unit for calculating an expected value of a posterior distribution based on a Bayesian estimation of a probability that the viewer will take the action in response to the advertisement in the case of the number of actions and the number of times of display with a beta distribution of An advertisement evaluation unit that determines an evaluation value of the advertisement according to the expected value, and an advertisement distribution unit that distributes a predetermined number of the advertisements in descending order of the evaluation value.

  According to another aspect of the present invention, there is provided a method for evaluating an advertisement, in which a computer obtains the number of times of displaying the advertisement, and the number of actions that a viewer takes action in response to the advertisement. And calculating an expected value of the posterior distribution by Bayesian estimation of the probability that the viewer will take the action in response to the advertisement in the case of the number of actions and the number of times of display with a predetermined beta distribution as a prior distribution. And a step of determining an evaluation value of the advertisement according to the expected value.

  According to another aspect of the present invention, there is provided a program for evaluating an advertisement, the step of obtaining a display number of times the advertisement is displayed on a computer, and an action that a viewer takes action in response to the advertisement. The expected value of the posterior distribution by the Bayesian estimation of the probability that the viewer will take the action in response to the advertisement in the case of the number of actions and the number of times of display with a predetermined beta distribution as a prior distribution. And a step of determining an evaluation value of the advertisement according to the expected value.

  Other problems and solutions to be disclosed by the present application will be made clear by the embodiments of the invention and the drawings.

  According to the present invention, an appropriate advertisement can be evaluated.

It is a figure which shows an example of the web page 1 displayed in the advertising system of this embodiment. It is a graph which shows posterior probability distribution Be (3, 9). It is a graph which shows posterior probability distribution Be (5, 16). It is a figure which shows the whole structure of the advertisement delivery system of this embodiment. 2 is a diagram illustrating a hardware configuration example of a Web server 20. FIG. 3 is a diagram illustrating an example of a software configuration of a Web server 20. FIG. It is a figure which shows the structural example of the advertisement information memorize | stored in the advertisement database. It is a figure which shows the structural example of the impression log memorize | stored in the impression log memory | storage part 253. 6 is a diagram illustrating a configuration example of a click log stored in a click log storage unit 254. FIG. FIG. 4 is a diagram showing a flow of processing performed by a Web server 20. It is a figure which shows the flow of the selection process of the advertisement in step S408. It is a figure explaining the delivery condition of the advertisement from which only the advertisement with a large bid amount is selected. It is a figure explaining the delivery condition of the advertisement at the time of using xCTR instead of CTR. It is a graph which shows a posteriori probability distribution in case the number of clicks is 1, the number of display times is 16, and the click rate upper limit value θ = 0.1.

== (1) Summary of Invention ==
Hereinafter, an advertisement distribution system according to an embodiment of the present invention will be described. In the present embodiment, an advertisement is assumed to be an advertisement (referred to as a listing advertisement) that is displayed on a Web page according to the content or the like. FIG. 1 is a diagram showing an example of a web page 1 displayed in the advertising system of the present embodiment. A plurality of advertisements 2 are displayed on the Web page 1 together with various contents (displayed as topics in the example of FIG. 1). In the present embodiment, a predetermined number (three in the example of FIG. 1) of advertisements 2 are inserted into information for displaying the Web page 1 (hereinafter referred to as screen information) in descending order of evaluation value and transmitted. Assumed.

The advertisement distribution system according to the present embodiment is intended to enhance the advertising effect by preferentially distributing advertisements with high evaluation. As an index indicating advertisement evaluation (hereinafter referred to as an advertisement evaluation value), click rate (CTR), advertisement revenue (eCPM = CTR × bid amount), and the like are often used. Then, the average (expected value) of the posterior distribution based on Bayesian estimation regarding the probability that the advertisement is clicked (click rate) is set as the advertisement evaluation value.
Specifically, the prior distribution is assumed to be a uniform beta distribution Be (1, 1) from the principle of insufficient reason. When an advertisement is distributed n times and clicked r times, the posterior distribution becomes Be (1 + r, 1 + n−r). In the following description, the number n of times an advertisement has been distributed is referred to as a display number, the number of clicks r is referred to as the number of clicks, and the parameters of the posterior distribution are α and β (α = 1 + clicks, β = 1 + displays− Clicks). The expected value of the beta distribution at this time is calculated as the expected value of the click rate of the advertisement (hereinafter referred to as xCTR (eXpected Click Through Rate)), and xCTR is used as the evaluation value of the advertisement. Note that xCPM (eXpected Cost Per Mill) obtained by multiplying the expected value by the bid amount may be used as the advertisement evaluation value.

Since the expected value (xCTR) of the beta distribution Be (α, β) is α ÷ (α + β), the xCTR in the posterior probability distribution is obtained by the following equation.
xCTR = (1 + click count) ÷ (2 + display count) (E1)

== (2) Specific example of xCTR ==
FIG. 2 shows the posterior probability distribution Be () when the number of times an advertisement A is displayed (also called the number of display times and the number of impressions) is 10, and the number of clicks on the advertisement A (the number of clicks) is 2. 1 + 2, 1 + 10−2) = Be (3, 9). In the conventional formula for calculating the click rate, CTR = number of clicks / number of displays = 2 ÷ 10 = 0.2, which is a mode of the probability distribution Be (3, 9). On the other hand, in this embodiment, the expected value of Be (3, 9), that is, xCTR = (1 + 2) ÷ (10 + 2) = 0.25.

  On the other hand, FIG. 3 is a graph showing the posterior probability distribution Be (1 + 4, 1 + 19−4) = Be (5, 16) when the number of times that an advertisement B is displayed is 19 and the number of clicks is 4. is there. In this case, CTR = number of clicks / number of display times = 4 / 19≈0.21, and xCTR = (1 + 4) ÷ (19 + 2) ≈5 ÷ 21≈0.238.

  Assuming that only one advertisement is distributed, an advertisement with a high click rate is distributed, and thus an advertisement B (0.21> 0.2) having a high CTR has been distributed. On the other hand, in the advertisement distribution system of the present embodiment, the advertisement A (0.25> 0.238) having a large xCTR is distributed.

  Here, assuming that the bid amount of advertisements A and B (advertising revenue paid when clicked) is 10 yen, the expected value of revenue when 10,000 advertisements are distributed is CTR as before. The advertisement B distributed based on the above is 5 ÷ 21 × 10 × 10000 = 23809 yen, and the distribution based on xCTR is 0.25 × 10 × 10000 = 25000 yen, and an increase of about 1191 yen is expected.

  Therefore, in the advertisement distribution system of the present embodiment, advertisements are evaluated by xCTR, and advertisements with high evaluation values are distributed preferentially.

== (3) System configuration ==
FIG. 4 is a diagram showing the overall configuration of the advertisement distribution system of this embodiment. The advertisement distribution system according to the present embodiment includes a Web server 20 that is communicably connected to the user terminal 10 via a communication network 30.

  The communication network 30 is, for example, the Internet or a LAN (Local Area Network) constructed by a public telephone line network, a dedicated communication line network, a mobile phone line network, a wireless communication network, Ethernet (registered trademark), or the like.

  The Web server 20 is a computer such as a workstation or a personal computer that provides various contents such as images, moving images, and texts according to HTTP (HyperText Transfer Protocol). The user terminal 10 is a computer such as a personal computer, a tablet computer, a mobile phone terminal, or a PDA (Personal Digital Assistant) that operates a browser for browsing various Web pages provided by the Web server 20.

== (4) Hardware Configuration of Web Server 20 ==
FIG. 5 is a diagram illustrating a hardware configuration example of the Web server 20. The Web server 20 includes a CPU 201, a memory 202, a storage device 203, a communication interface 204, an input device 205, and an output device 206. The storage device 203 is, for example, a hard disk drive, a solid state drive, or a flash memory that stores various data and programs. The communication interface 204 is an interface for connecting to the communication network 30, for example, an adapter for connecting to Ethernet (registered trademark), a modem for connecting to a public telephone line network, and a wireless communication device for performing wireless communication A USB (Universal Serial Bus) connector for serial communication, an RS232C connector, or the like. The input device 205 is, for example, a keyboard, a mouse, a touch panel, a button, or a microphone that inputs data. The output device 206 is, for example, a display, a printer, or a speaker that outputs data.

== (5) Software Configuration of Web Server 20 ==
FIG. 6 is a diagram illustrating a software configuration example of the Web server 20. The web server 20 includes an HTTP processing unit 211, a web page creation unit 212, an advertisement selection unit 213, a log collection unit 214, a web information database 251, an advertisement database 252, an impression log storage unit 253, and a click log storage unit 254. .

  Note that the functional units of the HTTP processing unit 211, the Web page creation unit 212, the advertisement selection unit 213, and the log collection unit 214 are realized by the CPU 201 included in the Web server 20 executing a program stored in the storage device 202. The storage units of the Web information database 251, the advertisement database 252, the impression log storage unit 253, and the click log storage unit 254 are realized as part of the storage areas provided by the memory 202 and the storage device 203 provided in the Web server 20. The

== (5) Description of Storage Unit ==
(6-1) Web information database 251
The web information database 251 stores information for creating a web page (hereinafter referred to as web information). The Web information is, for example, a file described in HTML, a program file storing a program executed by a mechanism such as CGI (Common Gateway Interface), an image file, or the like. The Web information database 251 can be realized by, for example, a file system directory or folder provided by an operating system operating on the Web server 20.

(6-2) Advertising database 252
The advertisement database 252 stores information for displaying advertisements (hereinafter referred to as advertisement information). FIG. 7 is a diagram illustrating a configuration example of advertisement information stored in the advertisement database 252. The advertisement information includes an advertisement category, a URL (Uniform Resource Locator) related to the advertisement, an advertisement title, and a bid amount in association with the advertisement ID that is the identification information of the advertisement. The bid amount is the maximum unit price paid from the advertiser when the advertisement is clicked, and is determined by the advertiser.

  In the present embodiment, the character string of the title is displayed as the advertisement 2 on the screen 1. The advertisement 2 is set with a link to a predetermined URL designating the advertisement ID (the Web server 20 is designated as the host name; hereinafter referred to as a redirect URL). When the advertisement 2 is clicked, a request including the advertisement ID is transmitted to the Web server 20, a log indicating that the click is collected by the log collection unit 214 described later, and the advertisement information corresponding to the advertisement ID is collected by the HTTP processing unit 211. (For example, a response specifying the URL as a redirect destination is transmitted). Thereby, the user's action (click) according to the advertisement can be detected.

  In addition, you may make it display an image, a moving image, etc. as an advertisement, or you may make it output an audio | voice, and in that case, URL which shows the data of the image displayed as an advertisement, or a moving image may be added, for example Can be.

(6-3) Impression log storage unit 253
The impression log storage unit 253 stores information (hereinafter referred to as an impression log) indicating that an advertisement is displayed on the screen viewed by the user (referred to as an impression). FIG. 8 is a diagram illustrating a configuration example of an impression log stored in the impression log storage unit 253. The impression log includes the date and time (delivery date and time) when the advertisement 2 is displayed on the screen 1, the advertisement ID indicating the advertisement 2 displayed on the screen 1, and the category of the advertisement.

  In the example of the screen 1 in FIG. 1, since three advertisements 2 are displayed, when the user views the Web page 1 once, three impression logs are registered.

(6-4) Click log storage unit 254
The click log storage unit 254 (corresponding to the action history storage unit of the present invention) stores information indicating that the user has clicked the advertisement 2 displayed on the screen 1 (hereinafter referred to as a click log). FIG. 9 is a diagram illustrating a configuration example of the click log stored in the click log storage unit 254. The click log includes an advertisement ID indicating the clicked advertisement 2, a category of the advertisement, and a date and time (click date and time).

== (7) Description of Functional Units ==
(7-1) HTTP processing unit 211
The HTTP processing unit 211 performs HTTP processing. The HTTP processing unit 211 receives an HTTP request transmitted from the user terminal 10.
When the redirect URL is specified in the request, the HTTP processing unit 211 calls the log collection unit 214 to register the click log, and then transmits the advertisement information corresponding to the advertisement ID included in the request to the advertisement database 252. And a response specifying the URL included in the read advertisement information as a redirect destination is returned to the user terminal 10.

  When a URL indicating a Web page other than the redirect URL is specified in the request, the HTTP processing unit 211 calls information for displaying the Web page by calling the Web page creation unit 212 (hereinafter referred to as screen information). The information is, for example, an HTML file), the image information created by the web page creation unit 212 is responded to the user terminal 10, and the log collection unit 214 is called to register the impression log.

(7-2) Web page creation unit 212
The web page creation unit 212 (corresponding to the advertisement output unit of the present invention) creates screen information for displaying a web page. The web page creation unit 212 calls the advertisement selection unit 213 to select the advertisement 2 to be displayed on the screen 1, and when the selected advertisement is displayed as the advertisement 2 and the advertisement 2 is clicked, the redirect URL is accessed. For example, a character string in which a link tag is set is inserted into the screen information. In addition, about the process in which the web page preparation part 212 produces various contents other than the advertisement 2, the process by a general web server is used, and description is abbreviate | omitted here.

(7-3) Advertisement selection unit 213
The advertisement selection unit 213 selects an advertisement to be displayed on the web page.
The advertisement selection unit 213 includes a distribution candidate extraction unit 231, an advertisement evaluation calculation unit 232, and an advertisement determination unit 233. The distribution candidate extraction unit 231, the advertisement evaluation calculation unit 232, and the advertisement determination unit 233 are also realized by the CPU 201 reading the program stored in the storage device 203 into the memory 202 and executing it.

(7-3-1) Distribution candidate extraction unit 231
The distribution candidate extraction unit 231 determines advertisement candidates. The distribution candidate extraction unit 231 reads a predetermined number (for example, any number such as 10, 50, 100, etc.) of advertisement information from the advertisement database 252. In the present embodiment, the distribution candidate extraction unit 231 reads out a predetermined number of advertisement information at random. For example, the advertisement information may be read out according to the contents and user attributes. Note that a general advertisement selection process can also be used for the process in which the distribution candidate extraction unit 231 selects a predetermined number of advertisements from among a number of advertisements.

(7-3-2) Advertisement evaluation calculation unit 232
The advertisement evaluation calculation unit 232 (corresponding to the evaluation value calculation unit of the present invention) calculates the evaluation value of the advertisement. As described above, in this embodiment, the expected value xCTR of the posterior distribution is calculated as the evaluation value. The advertisement evaluation calculation unit 232 may calculate xCPM obtained by multiplying xCTR by a bid amount as an evaluation value. The advertisement evaluation calculation unit 232 may calculate the evaluation value by multiplying xCTR by a weight other than the bid amount.
The advertisement evaluation calculation unit 232 counts the number of impression logs corresponding to the advertisement candidate from the impression log storage unit 253 for each of the advertisement candidates determined by the distribution candidate extraction unit 231, and the click log storage unit 254. Then, the number of click logs corresponding to the advertisement candidate is counted as the number of clicks, and the number of display times and the number of clicks are given to Equation E1 to calculate xCTR.

(7-3-3) Advertisement determination unit 233
The advertisement determination unit 233 determines an advertisement to be displayed from the candidates determined by the distribution candidate extraction unit 231. The advertisement determining unit 233 determines a predetermined number (3 in the example of FIG. 1) of advertisements to be displayed in descending order of xCTR calculated by the advertisement evaluation calculating unit 232.

(7-4) Log collection unit 214
The log collection unit 214 registers an impression log and a click log. When the redirect URL is designated in HTTP, the log collection unit 214 reads out the category corresponding to the advertisement ID designated in the redirect URL from the advertisement database 252 and reads the advertisement ID, the read category, and the current A click log with the date and time set is created and registered in the click log storage unit 254. When a URL other than the redirect URL is specified in HTTP, the log collection unit 214 includes the current date and time, the advertisement ID indicating the advertisement, and the category of the advertisement for each advertisement selected by the advertisement selection unit 213. An impression log is created and registered in the impression log storage unit 253.

== (7) Processing ==
FIG. 10 is a diagram illustrating a flow of processing performed by the Web server 20.
When receiving an HTTP request from the user terminal 10 (S401), the HTTP processing unit 211 determines whether or not an advertisement has been clicked based on whether or not a redirect URL is specified in the request (S402).

  When it is determined that the advertisement is clicked (S402: YES), the log collection unit 214 reads advertisement information corresponding to the advertisement ID set in the redirect URL from the advertisement database 252, and the advertisement ID set in the request, A click log including the category of the advertisement information and the current date and time is created and registered in the click log storage unit 254 (S403). The HTTP processing unit 211 transmits an HTTP response to the user terminal 10 so as to redirect to the URL of the advertisement information (S404).

  On the other hand, if it is determined that the advertisement has not been clicked (S402: NO), an advertisement list including the advertisement information of the advertisement selected by the advertisement selection unit 213 is created by the process of FIG. 11 (S405). The advertisement information selection process shown in FIG. 11 will be described later.

  The web page creation unit 212 creates screen information such as an HTML file for displaying various contents in response to the request (S406). Note that the screen information creation processing in step S406 can employ processing by a general Web server.

  The Web page creation unit 212 creates a redirect URL in which the advertisement ID included in the advertisement information is set as an argument of a predetermined URL for each advertisement information included in the advertisement list (S407). The redirect URL can be in the form of “www.foobar.com/ad?id=xxxx”, for example. For each advertisement information included in the advertisement list, the web page creation unit 212 includes a link tag to a redirect URL including the title included in the advertisement information (for example, “<a href =" http: //www.foobar. com / ad? id = xxxx "> title </a>") is inserted into the screen information (S408).

  The HTTP processing unit 211 transmits the created screen information to the user terminal 10 as a response to the request (S409). For each advertisement information included in the advertisement list, the log collection unit 214 creates an impression log including the advertisement ID and category included in the advertisement information and the current date and time and registers the impression log in the impression log storage unit 253 (S410).

== (8) Advertisement selection process ==
FIG. 11 is a diagram showing the flow of advertisement selection processing in step S408.
The distribution candidate extraction unit 231 extracts a predetermined number m pieces of advertisement information from the advertisement database 252 (S421). Note that what kind of advertisement information is extracted from the advertisement database 252 can be arbitrarily set. For example, m pieces of advertisement information may be extracted at random, m pieces of advertisement information may be extracted in descending order of the degree that the title of the advertisement information matches the content, and the user is interested in May be stored as an attribute of user information, and m pieces of advertisement information matching the category may be extracted.

The advertisement evaluation calculation unit 232 performs the following processing for each of the extracted advertisement information.
The advertisement evaluation calculation unit 232 counts the number of impression logs corresponding to the advertisement ID included in the advertisement information from the impression log storage unit 253 to obtain the display count (S422), and the click log storage unit 254 adds the advertisement ID to the advertisement ID. The number of corresponding click logs is counted as the number of clicks (S423). The advertisement evaluation calculation unit 232 expects the posterior distribution Be of the click rate based on Bayesian estimation when the number of clicks and the number of display times are observed in the uniform distribution Be (1, 1) (number of clicks + 1, the number of display times−the number of clicks + 1). The value is calculated as xCTR corresponding to the advertisement ID (S424). That is, the advertisement evaluation calculation unit 232 sets a quotient obtained by dividing the value obtained by adding 1 to the number of clicks by the value obtained by adding 2 to the number of display times as xCTR (advertisement ID).

After performing the above processing for each advertisement information, the advertisement evaluation calculation unit 232 selects n pieces of advertisement information in descending order of xCTR to make an advertisement list (S425).
In this way, n pieces of advertisement information are extracted as an advertisement list in descending order of xCTR. Note that n pieces of advertisement information may be selected in descending order of a value obtained by multiplying xCTR by a bid amount of advertisement information.

== (9) Effect ==
As described above, in the advertisement distribution system according to the present embodiment, the expected value of the posterior probability distribution based on Bayesian estimation is used as the estimated value of the click rate of the advertisement, and the evaluation value of the advertisement can be determined based on the estimated value. . It is known that the CTR calculated by the conventional formula for calculating the click rate, that is, the mode value (number of clicks / number of display times) is not an appropriate click rate particularly when the number of display times is small. The expected value xCTR of the posterior probability distribution based on Bayesian estimation is considered to be a reasonable estimate of the probability that an advertisement is clicked after observing the number of display times and the number of clicks. Even in such a case, it is possible to obtain an appropriate advertisement evaluation. Therefore, as in this embodiment, by using xCTR (or xCPM obtained by multiplying xCTR by a bid amount) as an evaluation value of an advertisement, it is possible to obtain an appropriate advertisement evaluation.

  Further, since the conventional click rate is CTR = number of clicks / number of display times and xCTR of this embodiment = (number of clicks + 1) ÷ (number of display times + 2), it is an expected value of the beta distribution, so by adopting xCTR or xCPM , Maximize the expected clickthrough rate or advertising revenue. In other words, by displaying an advertisement with a high xCTR, it is possible to increase the possibility that the advertisement is clicked on the Web page or the entire website, and to increase advertising revenue by displaying an advertisement with a high xCPM. Can do. Therefore, it is possible for the advertiser to maximize the advertising effect, and it is possible to improve the advertising profit for the publisher of the advertisement (operator of the Web server 20).

  On the other hand, the expected value of the posterior distribution can be calculated by a relatively simple calculation formula like the formula (E1). Even if compared with the conventional formula for calculating the click rate, the calculation load is not so high. Therefore, the advertising effect can be improved without increasing the processing load.

  Further, when determining an advertisement to be distributed, an advertisement is selected in descending order of CTR or eCPM, and when the CTR or eCPM is the same, the one with the larger bid is selected. In such a situation, only an advertisement with a large bid amount is selected for an advertisement with a click result value of 0, such as when there is no submission.

  The advertisement distribution status in this case is shown in FIG. As shown in FIG. 12, the advertisement C and the advertisement D are in a state where there is no submission, and both the number of clicks and the number of display times are zero. If the bid amount of the advertisement C is 50 yen and the bid amount of the advertisement D is 45 yen, both the CTR and eCPM are 0, so the advertisement C having a large bid amount is selected. In this situation, the advertisement D is not displayed as long as there is an advertisement C with a large bid amount. Even if an advertisement with a small number of impressions is displayed at random, generally the click rate is often smaller than several percent, and the CTR of the advertisement C exceeds 0 (clicked once) compared to the time until , The time until the CTR of the advertisement D exceeds 0 becomes extremely long.

  On the other hand, by using xCTR instead of CTR as in the present embodiment, as shown in FIG. 13, xCTR starts from 50% of a uniform distribution even for an advertisement with 0 clicks. Since the expected value of the posterior distribution is adjusted according to the number of times of display, the opportunity for displaying the advertisement D increases. This is because even when advertisements are selected in descending order of xCPM calculated by multiplying the bid price by xCTR, since the distribution of eCPM is proportional to the distribution of CTR, the opportunity for displaying advertisement D is also increased. become. Therefore, according to the advertisement distribution system of the present embodiment, the expected value of the posterior distribution can be calculated as the estimated click rate (xCTR), and the advertisement can be distributed based on this, so the number of clicks is small (or Advertise can also be provided with a reasonable distribution of advertisements (often with fewer impressions).

== (10) Modification: Upper limit value θ ==
In the above description, the value of xCTR may be any value between 0 and 1, and the expected value of the posterior probability distribution Be (1 + click number, 1 + display number−click number), that is, Σ (click rate (x) XProbability P (x)) ÷ Σ (Probability P (x)) = (1 + Number of clicks) ÷ (2 + Number of display times) is set as xCTR, but an upper limit value θ may be set for the click rate x. FIG. 14 is a graph when the upper limit value θ of the click rate is set to 0.1 in the posterior probability distribution when a certain advertisement C has 1 click and 16 display times. As shown in the graph of FIG. 14, the expected value of the x <θ portion of the beta distribution is calculated as xCTR. In this case, xCTR is calculated by the following equation (E2).

That is, the number obtained by dividing the incomplete beta function B (θ; α + 1, β) by the incomplete beta function B (θ; α, β) is xCTR. As described above, α = 1 + the number of clicks, β = 1 + the number of display times−the number of clicks.

これは、正規化不完全ベータ関数Ｉ_θ（α＋１，β）にベータ関数Ｂ（α＋１，β）を乗じた値を、正規化不完全ベータ関数Ｉ_θ（α，β）にベータ関数Ｂ（α，β）を乗じた値で割った商がｘＣＴＲであることを示している。 Further, the above formula (E2) is transformed into the following formula (E3).

This is obtained by multiplying the normalized incomplete beta function I _θ (α + 1, β) by the beta function B (α + 1, β) and the normalized incomplete beta function I _θ (α, β). , Β) indicates that the quotient divided by the value multiplied by xCTR is xCTR.

これは上述した式（Ｅ１）により算出されるｘＣＴＲに、正規化不完全ベータ関数Ｉ_θ（α＋１，β）を正規化不完全ベータ関数Ｉ_θ（α，β）で割った値を乗じたものである。すなわち、クリックされる確率の上限値をθとした場合の期待値をｘＣＴＲ（θ；α，β）、上限値を設けない場合（θ＝１と同値である）の期待値をｘＣＴＲ（α，β）とすると、ｘＣＴＲ（θ；α，β）は次式（Ｅ５）により算出することができる。

When this equation (E3) is modified, the following equation (E4) is obtained.

This is obtained by multiplying xCTR calculated by the above formula (E1) by a value obtained by dividing the normalized incomplete beta function I _θ (α + 1, β) by the normalized incomplete beta function I _θ (α, β). It is. That is, the expected value when the upper limit value of the click probability is θ is xCTR (θ; α, β), and the expected value when the upper limit value is not provided (same value as θ = 1) is xCTR (α, If β), xCTR (θ; α, β) can be calculated by the following equation (E5).

  As described above, the xCTR can be calculated by providing the upper limit value θ of the probability that the advertisement is clicked. The upper limit value θ may be, for example, input of the knowledge of the distributor of the advertisement, or may be about a predetermined number of times of display (for example, an arbitrary value such as 1,000 times or 10,000 times). It is good also as an upper limit of the track record value of click rate (CTR). By setting the upper limit value θ in this way, it is possible to prevent xCTR from being calculated as a problem, and thus the validity of the probability that an advertisement is clicked can be increased. This also maximizes the click rate (or advertising revenue) of the advertisement.

  Further, since the normalized incomplete beta function is prepared in a general statistical processing library or the like, it is easy to calculate the equation (E5). Therefore, even when the upper limit value θ is provided, it is expected that the mounting cost for calculating xCTR is not so high. Therefore, it is possible to easily estimate a reasonable click rate, and based on this (by multiplying xCTR itself or xCTR by the bid amount), it is possible to evaluate the advertisement appropriately. it can.

In the present embodiment, the prior distribution is premised on the uniform distribution Be (1, 1). However, the present invention is not limited to this, and an arbitrary prior distribution may be set. For example, the click number A and the display frequency B are counted from the impression log storage unit 253 and the click log storage unit 254 for all advertisements or for the same category for each advertisement category, and the uniform distribution Be is based on this count. A posterior distribution Be (1 + A, 1 + BA) with (1, 1) as a prior distribution is calculated, and xCTR is calculated using the posterior distribution Be (1 + A, 1 + BA) as a prior distribution. Good. That is, the formula (E1) is as follows.
xCTR = (1 + A + click rate) ÷ (2 + B + number of display times) (E1 ′)

  In this embodiment, xCTR is calculated at the time of advertisement distribution. However, the present invention is not limited to this, and xCTR is calculated in advance for each piece of advertisement information, and display advertisements are displayed in descending order of xCTR calculated in advance. You may make it select. In this case, for example, xCTR is included in the advertisement information, and the advertisement evaluation calculation unit 232 calculates xCTR at the timing when the log collection unit 214 registers the impression log or click log in the impression log storage unit 253 or click log storage unit 254. Then, the advertisement information can be updated with the calculated xCTR. Further, the advertisement evaluation calculation unit 232 may periodically calculate xCTR based on the impression log storage unit 253 and the click log storage unit 254.

  Further, the Web server 20 may include a probability table that stores xCTR in association with the number of clicks and the number of display times. Thereby, the processing load concerning calculation of xCTR can be reduced. Whether or not to use the upper limit value θ can be set, and the advertisement evaluation calculation unit 232 searches for the xCTR from the probability table when the upper limit value θ is used, and when the upper limit value θ is not used, XCTR may be calculated by (E1).

  In addition, the advertisement evaluation calculation unit 232 calculates xCTR by the equation (E5) using the upper limit value θ only until the number of times of display reaches a predetermined threshold, and when the number of times of display becomes equal to or greater than the above threshold. XCTR may be calculated by the equation (E1). In this case, the upper limit value θ is set and xCTR calculated in advance is registered in the probability table, and the advertisement evaluation calculation unit 232 acquires xCTR from the probability table until the number of display times reaches the threshold value, and displays it. When the number of times is equal to or greater than the above threshold value, xCTR can be calculated by equation (E1).

  In the present embodiment, the advertisement 2 is displayed on a part of the Web page 1. However, the present invention is not limited thereto, and a screen on which only the advertisement 2 is displayed may be displayed on the user terminal 10. Alternatively, the advertisement 2 may be displayed on a different screen from the content.

  Although the present embodiment has been described above, the above embodiment is intended to facilitate understanding of the present invention and is not intended to limit the present invention. The present invention can be changed and improved without departing from the gist thereof, and the present invention includes equivalents thereof.

  For example, in this embodiment, the advertisement distribution system is configured by one Web server 20, but is not limited thereto, and may be configured by a plurality of computers. For example, a web server that provides a web page, an application server that creates a web page, and a database server that manages various data can be used.

  For example, the present invention can be applied to an advertisement evaluation apparatus that evaluates an advertisement itself, for example, when evaluating a program (matching engine) for selecting an advertisement to be distributed. In this case, the advertisement evaluation device calculates a display number acquisition unit that acquires the number of advertisements displayed, a click number acquisition unit that acquires the number of clicks of the advertisement, and xCTR (or xCTR × bid amount) as the evaluation value of the advertisement. An advertisement evaluation calculation unit 232 and an evaluation value output unit that outputs an evaluation value of the advertisement are provided. The advertisement evaluation value device includes an impression log storage unit 253 and a click log storage unit 254, and the display number acquisition unit and the click number acquisition unit correspond to the advertisement ID from the impression log storage unit 253 and the click log storage unit 254. The number of impression logs and click logs may be counted, or when another computer includes the impression log storage unit 253 and the click log storage unit 254, the computer is accessed to correspond to the advertisement ID. The number of impression logs and click logs may be inquired.

DESCRIPTION OF SYMBOLS 1 Web page 2 Advertisement 10 User terminal 20 Web server 30 Communication network 211 HTTP processing part 212 Web page creation part 213 Advertisement selection part 214 Log collection part 231 Distribution candidate extraction part 232 Advertisement evaluation calculation part 233 Advertisement determination part 251 Web information database 252 Advertising database 253 Impression log storage unit 254 Click log storage unit

Claims (7)

A device for evaluating advertisements,
A number-of-impressions acquisition unit that acquires the number of times the advertisement has been displayed;
An action number acquisition unit for acquiring the number of actions that the viewer took action in response to the advertisement;
An expected value calculation unit that calculates an expected value of the posterior distribution by Bayesian estimation of the probability that the viewer will take the action in response to the advertisement in the case of the number of actions and the number of times of display with a predetermined beta distribution as a prior distribution When,
An advertisement evaluation unit that determines an evaluation value of the advertisement according to the expected value;
An advertisement evaluation apparatus comprising: The advertisement evaluation device according to claim 1,
The prior distribution is a beta distribution Be (a, b);
The expected value calculation unit calculates the expected value by dividing a value obtained by adding a to the number of actions and a value obtained by adding a and b to the number of times displayed.
An advertisement evaluation apparatus characterized by. The advertisement evaluation device according to claim 1,
An upper limit storage unit for storing the upper limit of the probability;
The advertisement evaluation unit includes a first parameter obtained by adding a to the upper limit value, the action count, and a second parameter obtained by adding b to the value obtained by subtracting the action count from the display count. Is used as a divisor, and a quotient is calculated by dividing the normalized incomplete beta function by using the upper limit, a value obtained by adding 1 to the first parameter, and a value obtained by providing the second parameter as a dividend. , Determining the evaluation value according to a value obtained by multiplying the expected value by the quotient,
An advertisement evaluation apparatus characterized by. The advertisement evaluation device according to claim 1,
A remuneration amount acquisition unit for acquiring a remuneration amount for the viewer taking action in response to the advertisement;
The advertisement evaluation unit calculates the evaluation value by multiplying the expected value by the reward amount;
An advertisement evaluation apparatus characterized by. A device for delivering advertisements,
A number-of-impressions acquisition unit that acquires the number of times the advertisement has been displayed;
An action number acquisition unit for acquiring the number of actions that the viewer took action in response to the advertisement;
An expected value calculation unit that calculates an expected value of the posterior distribution by Bayesian estimation of the probability that the viewer will take the action in response to the advertisement in the case of the number of actions and the number of times of display with a predetermined beta distribution as a prior distribution When,
An advertisement evaluation unit that determines an evaluation value of the advertisement according to the expected value;
An advertisement distribution unit that distributes a predetermined number of the advertisements in descending order of the evaluation value;
An advertisement distribution apparatus comprising: A method for evaluating ads,
Computer
Obtaining the number of times the advertisement has been displayed;
Obtaining the number of times the viewer has acted in response to the advertisement;
Calculating the expected value of the posterior distribution by Bayesian estimation of the probability that the viewer takes the action in response to the advertisement in the case of the number of actions and the number of times of display with a predetermined beta distribution as a prior distribution;
Determining an evaluation value of the advertisement according to the expected value;
The advertisement evaluation method characterized by performing. A program for evaluating ads,
On the computer,
Obtaining the number of times the advertisement has been displayed;
Obtaining the number of times the viewer has acted in response to the advertisement;
Calculating the expected value of the posterior distribution by Bayesian estimation of the probability that the viewer takes the action in response to the advertisement in the case of the number of actions and the number of times of display with a predetermined beta distribution as a prior distribution;
Determining an evaluation value of the advertisement according to the expected value;
A program for running

Images