JP6968033B2 - Information processing equipment, content distribution equipment, information processing methods, and programs - Google Patents

Description

The present invention relates to an information processing device, a content distribution device, an information processing method, and a program.

Conventionally, an invention of an apparatus for generating a model for predicting a click rate of contents has been disclosed (see Patent Document 1).

Japanese Patent No. 5996747

In content distribution, for example, a predetermined number of contents are distributed in descending order of some index related to the contents. Examples of the content index include the number of clicks and the click probability. These indicators may depend on probabilistic events such as the user clicking on the content, often following some probability distribution with expected value and variance. That is, the content is distributed in the form of predicting and ranking the index (click probability, etc.) of the content when it is distributed from now on, based on the past distribution record of the content. What should be taken into consideration here is that the more the predicted value of an index of a certain content (hereinafter referred to as the predicted index) is accidentally overestimated (the more the prediction is mistaken), the more the content tends to be ranked high in the ranking. This is the point where the bias occurs. This bias is called a selection bias. Due to this selection bias, for example, when delivering an advertisement which is an example of content, it may not be possible to deliver a truly profitable advertisement. That is, it may not be possible to select suitable content.

The present invention has been made in consideration of such circumstances, and an object of the present invention is to provide an information processing device, a content distribution device, an information processing method, and a program capable of selecting more suitable contents. I will do it.

One aspect of the present invention comprises a predictive index derivation unit for deriving a predictive index that predicts an index related to the content and a predictive index derivation unit for each of a plurality of content candidates that are candidates for the content provided to the terminal device. Information processing including a ranking processing unit that ranks the plurality of content candidates based on the derived prediction index, and a correction processing unit that corrects the prediction index based on the ranking result by the ranking processing unit. It is a device.

According to one aspect of the present invention, more suitable content can be selected.

It is a figure which shows an example of the configuration and use environment of the content distribution apparatus (advertisement distribution apparatus) 100 using the information processing apparatus of 1st Embodiment. It is a figure which shows the appearance that a plurality of advertisement content CAs are embedded in the main content MC. It is a figure which shows how the probability distribution is unevenly distributed for each rank (rank) of a prediction index. It is a figure for demonstrating the problem caused by a selection bias. It is a flowchart which shows an example of the flow of the process executed by the information processing apparatus of 1st Embodiment. It is a figure which shows an example of the content of the processing by the correction processing unit 130A. It is a figure for demonstrating the processing of the correction processing unit 130B and the correction coefficient learning unit 140B. It is a figure which shows the improvement tendency of the profit obtained as a result of a simulation.

Hereinafter, the information processing apparatus, the content distribution apparatus, the information processing method, and the embodiment of the program of the present invention will be described with reference to the drawings.

The information processing device is realized by one or more processors. The information processing device derives a predictive index that predicts an index related to the content for each of a plurality of content candidates that are candidates for the content provided to the user's terminal device, and selects the content to be provided to the terminal device. It is a device that outputs as information of. The destination to output the information may be another module of the own device or another device.

The content is content such as advertising content, news articles, and product sales screens on shopping sites. In the following description, the content is assumed to be advertising content. The advertising content is provided to the terminal device, for example, in a form of being embedded in an advertising space set in the main content provided by a certain service. The advertising content is, for example, a combination of a sample image and an introductory text (text) to which a hyperlink is attached. Not limited to this, the advertising content may be content such as text only, image only, or moving image. When the advertisement content is clicked or tapped (selected), the content indicated by the hyperlink attached to the advertisement content (for example, the landing page) is downloaded to the terminal device and played on the terminal device. In the following description, clicks, taps, and other selection modes are described in a unified manner for clicks.

The index related to the advertising content is, for example, an index based on the probability, frequency, number of times, or a combination thereof that the user selects the advertising content. Examples of this include CTR (Click Through Ratio) and the number of clicks. Further, the index may be an index indicating advertising efficiency such as CPM (Cost Per Mille) or eCPM (effective CPM). In the following description, the index is a CTR, and the information processing apparatus derives a predicted CTR that predicts the CTR as a predictive index.

Then, the information processing apparatus corrects the prediction index based on the rank of the prediction index, and selects the advertisement content provided to the terminal device from the plurality of advertisement content candidates based on the corrected prediction index. This allows the information processing device to select more suitable advertising content.

<First Embodiment>
FIG. 1 is a diagram showing an example of a configuration and a usage environment of a content distribution device (advertisement distribution device) 100 using the information processing device of the first embodiment.

The content distribution device 100 communicates with the user's terminal device 10 and the service server 20 via the network NW. The network NW includes, for example, the Internet, a WAN (Wide Area Network), a LAN (Local Area Network), a provider terminal, a wireless communication network, a wireless base station, a dedicated line, and the like.

The terminal device 10 is a terminal device including an input device, a display device, a communication device, a storage device, and an arithmetic unit, such as a mobile phone such as a smartphone, a tablet terminal, and various personal computers. The communication device includes a network card such as a NIC (Network Interface Card), a wireless communication module, and the like. In the terminal device 10, a UA (User Agent) such as a browser or an application program is activated, and a request according to the content input by the user is transmitted to the service server 20. Further, the UA displays various images based on the information acquired from the service server 20.

The service server 20 functions as a web server that provides a web page to the terminal device 10 in response to a request from a browser, or an application server that provides images and sounds to the terminal device 10 in response to a request from an application program. The service server 20 embeds the advertisement content provided by the content distribution device 100 in the main content and provides it to the terminal device 10. FIG. 2 is a diagram showing a state in which a plurality of advertising content CAs are embedded in the main content MC. The main content MC shown in FIG. 2 is a portal site that provides a search service and distributes various news. A plurality of advertising spaces are set in the main content MC, and the advertising content CA (CA (1) to CA (3) in the figure) is displayed in each advertising space.

The advertising content CA is, for example, a combination of a sample image and text, and clicking on either the sample image or the text causes the screen to transition to the introduction content of the product or service advertised by the advertising content. The service server 20 holds, for example, the browsing history of the main content or the advertisement content associated with the user's identification information (user ID) as log information 22 in the storage device, and the content distribution device 100 receives a request. The log information 22 is provided to the user.

The content distribution device 100 includes, for example, a prediction index derivation unit 110, a ranking processing unit 120, a correction processing unit 130, a correction coefficient learning unit 140, and a content providing unit 150. An example of the information processing device is a combination of the prediction index derivation unit 110, the ranking processing unit 120, the correction processing unit 130, and the correction coefficient learning unit 140. These components are realized by, for example, a hardware processor such as a CPU (Central Processing Unit) executing a program (software). Some or all of these components are hardware (circuit parts) such as LSI (Large Scale Integration), ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array), and GPU (Graphics Processing Unit). It may be realized by (including circuits), or it may be realized by the cooperation of software and hardware. The program may be stored in advance in a storage device such as an HDD (Hard Disk Drive) or a flash memory, or may be stored in a removable storage medium such as a DVD or a CD-ROM, and the storage medium is stored in the drive device. It may be installed by being attached.

The prediction index derivation unit 110 derives, for example, a predicted CTR that predicts the CTR of the advertising content for each of the plurality of advertising content candidates that are candidates for the advertising content provided (hereinafter, delivered) to the terminal device 10. .. The prediction index derivation unit 110 derives a prediction CTR for each main content and for each advertising space in the main content, for example. In the following description, the expression "candidate" indicating that it is unclear whether or not the terminal device 10 is provided at that time will be omitted as appropriate.

For example, the prediction index derivation unit 110 determines the degree of similarity between the user's attributes (hobbies, tastes, etc.) obtained from the actual CTR when the advertisement content was delivered in the past, the log information 22, and the content of the advertisement content. And derive a predictive CTR using a linear or non-linear logistic regression model. Further, when considering the time change of the CTR, the predicted CTR may be derived using a time series model.

The ranking processing unit 120 assigns ranks to the advertisement contents in descending order of the predicted CTR. Hereinafter, this process is referred to as ranking. A high rank means that the predicted CTR was high.

First, the content providing unit 150 will be described. The content providing unit 150 allocates the advertising content to the advertising space in descending order of the rank given by the ranking processing unit 120 in response to the advertisement request from the service server 20. For example, the advertising space is also given a priority, and the content providing unit 150 allocates the advertising content having a high rank to the advertising space having a high priority. The content providing unit 150 provides the service server 20 with the advertising content together with the allocation result. The content providing unit 150 may allocate randomly extracted advertising content to the advertising space at a fixed ratio.

The correction processing unit 130 performs correction processing of the predicted CTR according to the rank. The correction processing unit 130 performs correction processing of the prediction CTR by multiplying the prediction CTR by a correction coefficient (between 0 and 1) according to the rank. The correction coefficient is, for example, a value learned by the correction coefficient learning unit 140, but the correction coefficient 130 is not limited to this, and the correction processing unit 130 predicts using a correction coefficient given from the outside or set by default. CTR correction processing may be performed.

Hereinafter, the reason why the correction processing by the correction processing unit 130 is necessary will be described. Since the predicted CTR is a prediction of a future event, it is derived according to a predetermined probability distribution. For this reason, high-ranked advertising content includes content in the probability distribution whose value is only accidentally high. FIG. 3 is a diagram showing how the probability distribution is unevenly distributed for each rank of the prediction index. This figure assumes that the CTR of all advertisement contents is 0.10, and the number of clicks when the advertisement is impressed n times follows a binomial distribution with an expected value of n × 0.10. That is, when the predicted CTR of the advertisement is the number of clicks ÷ the number of impressions, it is shown that the predicted CTR follows the Bernoulli distribution with the expected value of 0.10. In the figure, Dollar shows the probability distribution of the predicted CTR of each advertising content before ranking, D1 shows the conditional probability distribution of the predicted CTR of the advertising content having the rank 1 (highest), and D2 shows the conditional probability distribution. Shows the conditional probability distribution of the predicted CTR of advertising content with a rank of 2 (next to 1), where D3 is the conditional probability of the predicted CTR of advertising content with a rank of 3 (next to 2). It shows the distribution. As shown in the figure, the conditional probability distribution of the predicted CTR after ranking is positively biased from the probability distribution before ranking as the rank becomes higher. This indicates that when considering the ranking in descending order of the predicted CTR, there is a selection bias that the predicted CTR that is accidentally overestimated is more likely to be selected as the higher rank.

FIG. 4 is a diagram for explaining the problem caused by the selection bias. As the number of ads targeted for ranking increases, as shown in the figure, the probability that one or more ad predictive indicators will be overestimated by chance increases, and the accuracy (for example, the rank of the most profitable ad content is 1). It is known by simulation that the probability that it can be set to is reduced.

In response to such a problem, the correction processing unit 130 corrects the predicted CTR for the advertisement content to which the rank is given, so that the content providing unit 150 can select an appropriate advertisement content.

The correction coefficient learning unit 140 includes the predicted CTR derived by the predicted index deriving unit 110, the rank of the predicted CTR, and the probability that the advertising content is selected by the user as a result of being provided to the terminal device 10 (actual value of the CTR). ) And, the correction coefficient α (rank) for each rank of the predicted CTR is learned. More specifically, the correction coefficient learning unit 140 compares the predicted CTR derived by the prediction index derivation unit 110 with the actual value of the CTR for each rank of the predicted CTR, and thereby, the correction coefficient α for each rank ( Learn rank).

The correction coefficient learning unit 140 learns the correction coefficient α (rank) for each rank based on, for example, the equation (1). In the formula, i is the identification information of the advertising content given the rank during the observation period. In the formula (1), the correction coefficient α (1) is obtained by dividing the total value of the CTRs for each of the advertising contents given the rank during the observation period by the total of the predicted CTRs for the same advertisement contents. Rank) is sought. Since i is the same in the numerator and the denominator, the equation (1) can be considered to be divided by the average value of the actual value of the CTR and the average value of the predicted CTR. The observation period is, for example, one day and one week.

Further, the correction coefficient learning unit 140 may obtain the correction coefficient α for each advertising space of the main content. In this case, the correction coefficient learning unit 140 learns the correction coefficient α (src, domain, rank) for each ad space and rank of the main content, for example, based on the equation (2). In the formula, src is the identification information of the main content, and domain is the identification information of the advertising space. Equation (2) focuses on the ad space of a certain main content, sums the actual CTR values for each ad content ranked during the observation period, and sums the predicted CTs for the same ad content. The correction coefficient α (src, domine, rank) is obtained by dividing by the value obtained.

When the correction processing unit 130 corrects the predicted CTR of each advertising content to which the rank is given by using the correction coefficient α (here, α (rank)) learned in this way, the ranking processing unit 120 causes the ranking processing unit 120 to correct. Ranking is performed again using the corrected predicted CTR. The correction processing unit 130 further multiplies the prediction CTR by the correction coefficient β for correcting the bias of the prediction CTR depending on the distribution position of the advertisement (position bias: see JP2013-182594) for prediction. The CTR may be corrected.

A series of processes in which the correction processing unit 130 performs the correction processing and the ranking processing unit 120 performs the ranking again is performed N times (N = 1 or more natural numbers). Here, when N> 1, the correction coefficient α may be learned as a different value depending on how many times the process is used. For example, assuming that N = 3, the correction coefficient learning unit 140 learns the correction coefficients α (rank (j)) (j = 1, 2, 3), respectively. j is an argument indicating how many times the correction is used. Instead of this, the correction coefficient α (rank (j)) corresponding to j may be required to be attenuated according to a predetermined rule as j becomes larger, for example. For example, it may be obtained as α (rank (j)) = γ ^(j-1) · α (rank (1)). γ is a positive value less than 1. Then, the prediction CTR (i, rank) and the correction processing unit 130 make corrections by properly using the correction coefficient α (rank (j)) depending on the number of corrections. For example, in the case of the second correction, the correction coefficient α (rank (2)) is used. As a result, it is possible to prevent a situation in which the predicted CTR does not converge to an appropriate value due to excessive correction.

Then, the final ranking result as a result of performing the series of processes N times is provided to the content providing unit 150, and the content providing unit 150 allocates the advertising content to the advertising space based on the rank. Through such a series of processes, the overestimated predicted CTR can be corrected to an appropriate value, and more suitable advertising content can be selected.

FIG. 5 is a flowchart showing an example of the flow of processing executed by the information processing apparatus of the first embodiment. The processing of this flowchart is repeatedly executed, for example, once an hour. The processing by the correction coefficient learning unit 140 may be executed asynchronously with the processing of the flowchart of FIG.

First, the information processing device extracts the advertisement content of the distribution candidate (S200). Next, the prediction index derivation unit 110 derives a prediction CTR for each advertising content candidate (S202). Next, the ranking processing unit 120 ranks the advertisement content based on the predicted CTR (S204).

Next, the correction processing unit 130 corrects the predicted CTR according to the rank, which is the processing result in S204, using at least the correction coefficient α learned for each rank (S206). Next, the ranking processing unit 120 adds the correction result in S206 (replaces the predicted CTR before correction with the corrected predicted CTR) and performs ranking again (S208).

Then, the information processing apparatus determines whether or not the processes of S206 and S208 have been executed N times (S210). If it is determined that the process has not been executed N times, the process is returned to S206, and if it is determined that the process has been executed N times, the process proceeds to S212. As described above, N = 1 may be used, in which case the processing itself of S210 is omitted.

When it is determined in S210 that the execution has been performed N times, or when N = 1 and the processing of S208 is completed, the information processing apparatus outputs the result of the last ranking to the content providing unit 150. This ends the processing of one routine in this flowchart.

In the process of S212, instead of "outputting the result of the last ranking performed to the content providing unit 150", "outputting the predicted CTR reflecting the last corrected result to the content providing unit 150" may be performed. .. By doing so, the content providing unit 150 can calculate an index such as eCPM, CPM, and CPC by reflecting the bid of the advertisement, and can select the advertisement content to be provided to the terminal device 10 based on the calculation result.

According to the information processing apparatus of the first embodiment described above, an index (for example, CTR) related to the content is predicted for each of the plurality of content candidates that are candidates for the content (for example, advertising content) provided to the terminal device 10. A prediction index derivation unit 110 that derives a prediction index (for example, a prediction CTR), a ranking processing unit 120 that ranks a plurality of content candidates based on the prediction index derived by the prediction index derivation unit 110, and a ranking processing unit 120. By providing the correction processing unit 130 for correcting the prediction index based on the result of the ranking according to the above, the content providing unit 150 can be made to select more suitable content.

<Second Embodiment>
Hereinafter, the second embodiment will be described. In the first embodiment, the correction processing unit 130 corrects the predicted CTR of the advertisement content to which the rank is given. The correction processing unit (hereinafter referred to as correction processing unit 130A) according to the second embodiment further has an extended correction coefficient based on the distribution of the correction coefficient α for each rank for at least a part of the candidates for the advertisement content outside the rank. The predicted CTR is corrected by setting α + and multiplying the predicted CTR by the correction coefficient α or the extended correction coefficient α + set for each rank.

FIG. 6 is a diagram showing an example of the content of processing by the correction processing unit 130A. The rank is given to the advertisement content of a predetermined number X in descending order of the predicted CTR. In the example of FIG. 6, the predetermined number X = 5. Here, the predicted CTR of the advertising content with the lowest rank (rank 5 in the figure) is not significantly different from the predicted CTR of the advertising content (outside 1 in the figure) having the highest predicted CTR next to the advertising content having the rank 5. there is a possibility. Therefore, if the predicted CTR of the advertising content of rank 5 is lowered by the correction, the predicted CTR of the advertising content of rank 5 and the non-rank 1 may be reversed, resulting in an unfavorable result (of course, when the reversal is preferable). There is also). Therefore, for example, the correction processing unit 130A fits the distribution of the correction coefficient α for each rank to a model ML that approximates a straight line or a curve, and based on the coordinates on the straight line or the curve that extends the model ML to the non-rank region. , Set the extended correction coefficient α +. Further, the correction processing unit 130A may, for example, perform processing such as spline interpolation on the distribution of the correction coefficient α for each rank to obtain the extended correction coefficient α +.

Then, the correction processing unit 130A multiplies the predicted CTR of the advertisement content to which the rank is given by the correction coefficient α (rank) for each rank, and a part of the advertisement contents not given the rank (the predicted CTR is high). The predicted CTR is corrected by multiplying the predetermined number Y) in order from the one by the extended correction coefficient α + (rank #). Rank # is an argument according to the order when the advertisement contents out of rank are arranged in the order of the predicted CTR. As a result, it is possible to suppress the occurrence of an inappropriate predictive CTR reversal phenomenon. In addition, in the case where N ≧ 2 which is the number of repetitions of correction / ranking, the correction processing unit 130A performs the correction processing by regarding the reverse order as a new rank when the reversal phenomenon occurs due to the above correction. ..

According to the second embodiment described above, in addition to achieving the same effect as that of the first embodiment, it is possible to suppress the occurrence of an inappropriate predictive CTR reversal phenomenon.

The function of setting the extended correction coefficient α + by the correction processing unit 130A described above may be a function of the correction coefficient learning unit 140. In this case, in the information processing apparatus, the correction coefficient learning unit 140 performs a process of setting the extended correction coefficient α + after learning the correction coefficient α.

<Third Embodiment>
Hereinafter, the third embodiment will be described. The correction processing unit (hereinafter referred to as the correction processing unit 130B) according to the third embodiment further corrects the predicted CTR based on the actual number of impressions of the main content in which the advertisement content is embedded.

The correction coefficient learning unit (hereinafter referred to as the correction coefficient learning unit 140B) of the third embodiment is, for example, any of the regions (BIN) of the actual impressions divided by a predetermined step size, that is, the actual impressions are converted into BIN. The correction coefficient α (or the correction coefficient α and the extended correction coefficient α +) is learned for each BIN of the actual impressions. The actual number of impressions is a value obtained by counting the number of impressions in a predetermined period (1 day, 1 week, January, etc.).

FIG. 7 is a diagram for explaining the processing of the correction processing unit 130B and the correction coefficient learning unit 140B. As shown in the figure, for example, the correction coefficient learning unit 140B converts the number of actual impressions into BINs in increments of 10,000, and the main content corresponding to each BIN is used as a population, and the correction coefficient α is the same as in the first or second embodiment. To learn. Then, the correction processing unit 130B acquires the actual impression number of the main content to be processed, and corrects the predicted CTR by using the correction coefficient α corresponding to the BIN to which the acquired actual impression number belongs. As a result, if the number of impressions affects the magnitude of the selection bias, the predicted CTR can be corrected more appropriately.

According to the third embodiment described above, in addition to achieving the same effect as that of the first or second embodiment, the predicted CTR can be corrected more appropriately.

<Simulation>
The inventor of the present application sets a predetermined value for the number of advertisement contents, the number of users, the number of advertisement contents to be ranked, and the number of advertisement requests per day, and ε in order to search for a highly profitable advertisement. -On the condition that the advertisement content is randomly delivered to the request of ε [%] using the Greedy algorithm, and the advertisement with high advertising revenue (bid amount x predicted CTR) is delivered to other requests. A simulation was performed.

As a result of the above simulation, as a result of obtaining the actual value of CTR every day, an increase of several [%] to several tens [%] was observed. This means that as a result of the more accurate prediction CTR being sought, users have more opportunities to click on the advertising content.

In addition, as a result of determining the bid of the advertisement with a uniform random number for each advertisement content and reflecting it in the above simulation, it was observed that the expected advertising revenue was also improved. FIG. 8 is a diagram showing an improvement tendency of the profit obtained as a result of the simulation. As shown in the figure, as a result of suppressing the upside of the predicted CTR due to selection bias compared to the comparative example, the cumulative forecasted return is suppressed, while the profit (expected return) that will actually be obtained is in the comparative example. It was observed to be improved in comparison. The comparative example is one in which the correction for the selection bias is not performed.

Although the embodiments for carrying out the present invention have been described above using the embodiments, the present invention is not limited to these embodiments, and various modifications and substitutions are made without departing from the gist of the present invention. Can be added.

10 Terminal device 20 Service server 22 Log information 100 Content distribution device (advertisement distribution device, information processing device)
110 Prediction index derivation unit 120 Ranking processing unit 130 Correction processing unit 140 Correction coefficient learning unit 150 Content provision unit

Claims (10)

For each of the plurality of content candidates that are candidates for the content provided to the terminal device, a predictive index derivation unit that derives a predictive index that predicts the index related to the content, and a predictive index derivation unit.
A ranking processing unit that ranks a plurality of content candidates based on the prediction index derived by the prediction index derivation unit, and a ranking processing unit.
A correction processing unit that corrects the prediction index based on the ranking result by the ranking processing unit, and a correction processing unit.
Equipped with
The correction processing unit
Of the correction coefficients set for each rank of the prediction index, the prediction index is corrected by multiplying the prediction index by the correction coefficient corresponding to the rank of the content candidate.
For at least a part of the content candidates outside the rank, the extended correction coefficient is set based on the distribution of the correction coefficient for each rank, and the correction coefficient set for each rank of the prediction index or the extended correction coefficient is used as the prediction index. Corrects the prediction index by multiplying by
Information processing device. For each of the plurality of content candidates that are candidates for the content provided to the terminal device, a predictive index derivation unit that derives a predictive index that predicts the index related to the content, and a predictive index derivation unit.
A ranking processing unit that ranks a plurality of content candidates based on the prediction index derived by the prediction index derivation unit, and a ranking processing unit.
A correction processing unit that corrects the prediction index by multiplying the prediction index by a correction coefficient according to the actual number of impressions of the main content in which the content is embedded.
Information processing device equipped with. Correction coefficient for each rank based on the prediction index derived by the prediction index derivation unit, the rank of the prediction index, and the event that the content is selected by the user as a result of being provided to the terminal device. Further equipped with a correction coefficient learning unit for learning
The information processing apparatus according to claim 1. The correction coefficient learning unit compares the prediction index derived by the prediction index derivation unit with the event in which the content is selected by the user as a result of being provided to the terminal device for each rank. , Learn the correction coefficient for each rank of the prediction index,
The information processing apparatus according to claim 3. The correction processing unit corrects the prediction index by multiplying the prediction index by the correction coefficient corresponding to the rank of the content candidate among the correction coefficients set for each rank of the prediction index.
The correction coefficient learning unit learns the extended correction coefficient based on the distribution of the correction coefficient for each rank for at least a part of the content candidates outside the rank.
The correction processing unit corrects the prediction index by multiplying the prediction index of the content candidate by the correction coefficient set for each rank of the prediction index or the extended correction coefficient.
The information processing apparatus according to claim 3 or 4. The information processing apparatus according to any one of claims 1 to 5.
A providing unit that provides content to the terminal device based on a prediction index output by the information processing device.
Content distribution device equipped with. The computer
For each of the plurality of content candidates that are candidates for the content provided to the terminal device, a prediction index that predicts the index related to the content is derived.
Based on the derived prediction index, the plurality of content candidates are ranked, and the content candidates are ranked.
Based on the result of the ranking, the prediction index is corrected .
At the time of the above correction
Of the correction coefficients set for each rank of the prediction index, the prediction index is corrected by multiplying the prediction index by the correction coefficient corresponding to the rank of the content candidate.
For at least a part of the content candidates outside the rank, the extended correction coefficient is set based on the distribution of the correction coefficient for each rank, and the correction coefficient set for each rank of the prediction index or the extended correction coefficient is used as the prediction index. Corrects the prediction index by multiplying by
Information processing method. The computer
For each of the plurality of content candidates that are candidates for the content provided to the terminal device, a prediction index that predicts the index related to the content is derived.
Based on the derived prediction index, the plurality of content candidates are ranked, and the content candidates are ranked.
The prediction index is corrected by multiplying the prediction index by a correction coefficient according to the actual number of impressions of the main content in which the content is embedded.
Information processing method. On the computer
For each of the plurality of content candidates that are candidates for the content provided to the terminal device, a prediction index that predicts the index related to the content is derived.
Based on the derived prediction index, the plurality of content candidates are ranked.
Based on the result of the ranking, the prediction index is corrected.
When making the correction, the computer
Of the correction coefficients set for each rank of the prediction index, the prediction index is corrected by multiplying the prediction index by the correction coefficient corresponding to the rank of the content candidate.
For at least a part of the content candidates outside the rank, the extended correction coefficient is set based on the distribution of the correction coefficient for each rank, and the correction coefficient set for each rank of the prediction index or the extended correction coefficient is used as the prediction index. By multiplying by, the prediction index is corrected.
program. On the computer
For each of the plurality of content candidates that are candidates for the content provided to the terminal device, a prediction index that predicts the index related to the content is derived.
Based on the derived prediction index, the plurality of content candidates are ranked.
The prediction index is corrected by multiplying the prediction index by a correction coefficient according to the actual number of impressions of the main content in which the content is embedded.
program.

Images