JP2019215767A - Information processor, content distribution device, information processing method, and program - Google Patents

Abstract

To provide an information processor, content distribution device, information processing method, and program, capable of selecting more suitable contents.SOLUTION: The information processor includes: a prediction index derivation section for deriving a prediction index obtained by predicting an index concerning contents provided to a terminal device for each of a plurality of content candidates of the contents; a ranking processing section for executing ranking of the plurality of content candidates on the basis of the prediction index derived by the prediction index derivation section; and a correction processing section for correcting the prediction index on the basis of a result of the ranking by the ranking processing section.SELECTED DRAWING: Figure 1

Description

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

  BACKGROUND ART Conventionally, an invention of an apparatus for generating a model for estimating a click rate of content has been disclosed (see Patent Document 1).

Japanese Patent No. 5999647

  In content distribution, for example, a predetermined number of contents are distributed in the descending order of some index related to the content. Examples of the index of the content include the number of clicks and the click probability. These indicators may depend on stochastic events, such as a user clicking on content, and often follow some probability distribution with expected value and variance. That is, content distribution is performed in such a manner that, based on past distribution results of the content, an index (such as a click probability) of the content at the time of distribution is predicted and ranked. It should be considered here that the more the predicted value of the index of a certain content (hereinafter, the predicted index) is accidentally overestimated (the more wrong the prediction), the more likely that the content tends to be higher in the ranking. This is the point at which the bias of This bias is called selection bias. Due to this selection bias, for example, when an advertisement as an example of content is distributed, an advertisement with a high profit may not be able to be distributed. That is, there may be a case where a suitable content cannot be selected.

  SUMMARY An advantage of some aspects of the invention is to provide an information processing apparatus, a content distribution apparatus, an information processing method, and a program that can select more appropriate content. One.

  One embodiment of the present invention provides a prediction index deriving unit that derives a prediction index that predicts an index relating to the content, for each of a plurality of content candidates that are candidates for content provided to the terminal device, An information processing apparatus comprising: 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 a ranking result by the ranking processing unit. Device.

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

FIG. 2 is a diagram illustrating an example of a configuration and a use environment of a content distribution device (advertisement distribution device) 100 using the information processing device of the first embodiment. It is a figure showing signs that a plurality of advertising contents CA were embedded in main contents MC. It is a figure showing signs that a probability distribution is unevenly distributed for every rank (rank) of a prediction index. FIG. 9 is a diagram for describing a problem caused by a selection bias. 5 is a flowchart illustrating an example of a flow of a process performed by the information processing device according to the first embodiment. It is a figure showing an example of the contents of processing by correction processing part 130A. FIG. 9 is a diagram for describing processing of a correction processing unit 130B and a correction coefficient learning unit 140B. It is a figure which shows the improvement tendency of the profit obtained as a result of simulation.

  Hereinafter, embodiments of an information processing device, a content distribution device, an information processing method, and a program according to 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 apparatus derives a prediction index that predicts an index related to content for each of a plurality of content candidates that are content candidates to be provided to the user terminal device, and selects content to be provided to the terminal device. This is a device that outputs as information. The output destination of the information may be another module of the own device or another device.

  The content is content such as an advertisement content, a news article, and a product sale screen on a shopping site. In the following description, the content is assumed to be advertisement content. The advertising content is provided to the terminal device, for example, in a form embedded in an advertising space set in main content provided by a certain service. The advertisement content is, for example, a combination of a sample image and an introduction text (text) to which a hyperlink is added. However, the present invention is not limited thereto, and the advertisement content may be content such as only text, only image, or moving image. When the advertisement content is clicked or tapped (selected), the content (for example, a landing page) indicated by the hyperlink given to the advertisement content is downloaded to the terminal device and reproduced on the terminal device. In the following description, clicks and taps and other selection modes will be described by unifying expressions for clicks.

  The index related to the advertising content is, for example, an index based on the probability, frequency, number of times the user selects the advertising content, or a combination thereof. An example of this is 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 CTR, and the information processing device derives a predicted CTR that is a prediction of CTR as a prediction index.

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

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

  The content distribution device 100 communicates with the user terminal device 10 and the service server 20 via the network NW. The network NW includes, for example, the Internet, a wide area network (WAN), a local area network (LAN), 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 device, 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 an 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 transmits a request corresponding to the content input by the user to the service server 20. 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 an image or sound to the terminal device 10 in response to a request from an application program. The service server 20 provides the terminal device 10 with the advertisement content provided from the content distribution device 100 embedded in the main content. FIG. 2 is a diagram illustrating a state in which a plurality of advertisement contents CA 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 advertisement frames are set in the main content MC, and the advertisement content CA (CA (1) to CA (3) in the figure) is displayed in each advertisement frame.

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

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

  For example, the prediction index deriving unit 110 derives a predicted CTR that predicts the CTR of the advertising content for each of a plurality of advertising content candidates that are candidates for the advertising content provided (hereinafter, distributed) to the terminal device 10. . For example, the prediction index deriving unit 110 derives a predicted CTR for each main content and each advertising space in the main content. In the following description, the expression “candidate” indicating that it is unknown at this time whether or not to be provided to the terminal device 10 is omitted.

  The prediction index deriving unit 110 determines, for example, the actual CTR obtained when the advertisement content was distributed in the past, and the similarity between the user attribute (hobby preference, etc.) obtained from the log information 22 and the content of the advertisement content. And a predicted CTR is derived 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 a rank to the advertisement content in descending order of the predicted CTR. Hereinafter, this processing is referred to as ranking. A higher rank means that the predicted CTR was higher.

  First, the content providing unit 150 will be described. In response to the advertisement request from the service server 20, the content providing unit 150 allocates the advertising content to the advertising space in the descending order of the rank given by the ranking processing unit 120. For example, priorities are also assigned to the advertising spaces, and the content providing unit 150 assigns high-ranked advertising content to the high-priority advertising spaces. The content providing unit 150 provides the service server 20 with the advertisement content along with the assignment result. Note that the content providing unit 150 may allocate the advertisement content randomly extracted to the advertisement space at a fixed rate.

  The correction processing unit 130 performs a process of correcting the predicted CTR according to the rank. The correction processing unit 130 performs a correction process on the predicted CTR by multiplying the predicted 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 is not limited thereto, and the correction processing unit 130 may perform prediction using a correction coefficient given from 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, the advertising content with a high rank includes a probability distribution in which only a high value appears by chance. FIG. 3 is a diagram illustrating a state in which the probability distribution is unevenly distributed for each rank (rank) of the prediction index. This drawing assumes that the CTR of all the advertising contents is 0.10, and that the number of clicks when the advertisement is impression n times follows a binomial distribution with an expected value of n × 0.10. That is, when the number of clicks of the predicted CTR of the advertisement is equal to the number of impressions, the predicted CTR follows the Bernoulli distribution with an expected value of 0.10. In the figure, Dall indicates the probability distribution of the predicted CTR of each advertising content before being ranked, D1 indicates the conditional probability distribution of the predicted CTR of the advertising content whose rank is 1 (highest), and D2 indicates The conditional probability distribution of the predicted CTR of the advertisement content having the rank of 2 (highest next to 1) is shown, and D3 is the conditional probability of the predicted CTR of the advertisement content of rank 3 (highest next to 2). The distribution is shown. As shown, the conditional probability distribution of the predicted CTR after ranking is more positively biased from the probability distribution before ranking as the rank becomes higher. This indicates that when the ranking is considered in the descending order of the predicted CTR, there is a selection bias that the predicted CTR that is overestimated by chance is more likely to be selected in the higher rank.

  FIG. 4 is a diagram for explaining a problem caused by the selection bias. As shown in the figure, as the number of advertisements to be ranked increases, the probability that one or more advertisement prediction indicators are accidentally overestimated increases, and the accuracy (for example, the rank of the most profitable advertisement content is 1 rank). It has been found by simulation that the probability of being able to be set to) decreases.

  For such a problem, the correction processing unit 130 corrects the predicted CTR for the ranked advertisement content, so that the content providing unit 150 can select an appropriate advertisement content.

  The correction coefficient learning unit 140 calculates the prediction CTR derived by the prediction index deriving unit 110, the rank of the prediction CTR, and the probability that the advertisement content is selected by the user as a result provided to the terminal device 10 (the actual value of the CTR). ), 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 deriving unit 110 with the actual value of the CTR for each rank of the predicted CTR, thereby obtaining a correction coefficient α ( rank).

  The correction coefficient learning unit 140 learns a correction coefficient α (rank) for each rank based on, for example, Equation (1). In the formula, i is identification information of the advertising content that has been ranked during the observation period. Equation (1) is obtained by dividing the value obtained by summing the actual values of CTRs for each ranked advertising content during the observation period by the value obtained by summing the predicted CTRs for the same advertising content, and calculating a correction coefficient α ( rank). Since i is the same in the numerator and the denominator, equation (1) can be considered to be divided by the average value of the actual CTR value and the average value of the predicted CTR. The observation period is, for example, a period of one day or one week.

  Further, the correction coefficient learning unit 140 may calculate 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 advertising space and rank for the main content, for example, based on Expression (2). In the formula, src is identification information of the main content, and domain is identification information of the advertising space. Equation (2) focuses on the advertising space of a certain main content, sums up the actual value of the CTR for each ranked advertising content during the observation period, and sums the predicted CT for the same advertising content. The correction coefficient α (src, domain, rank) is obtained by dividing by the calculated value.

  When the correction processing unit 130 corrects the predicted CTR of each ranked advertisement content using the correction coefficient α (here, α (rank)) learned in this way, the ranking processing unit 120 Ranking is performed again using the corrected predicted CTR. The correction processing unit 130 further multiplies the predicted CTR by a correction coefficient β for correcting a bias of the predicted CTR (position bias: see Japanese Patent Application Laid-Open No. 2013-182594) which depends on the distribution position of the advertisement. The CTR may be corrected.

A series of processes in which the correction processing unit 130 performs the correction process and the ranking processing unit 120 performs the ranking again is performed N times (N = 1 or a natural number). Here, when N> 1, the above-described correction coefficient α may be learned as a different value depending on the number of processes used. For example, assuming that N = 3, the correction coefficient learning unit 140 learns the correction coefficient α (rank (j)) (j = 1, 2, 3). j is an argument indicating the number of times of correction to be used. Alternatively, the correction coefficient α (rank (j)) corresponding to j may be determined, for example, such that the larger the j is, the more attenuated according to a predetermined rule. For example, α (rank (j)) = γ ^(j−1) · α (rank (1)) may be obtained. γ is a positive value less than 1. Then, the predicted CTR (i, rank) and the correction processing unit 130 perform the correction 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 where the predicted CTR does not converge to an appropriate value due to excessive correction.

  Then, the final ranking result of the result of performing the series of processing 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. By such a series of processing, the predicted CTR for the overestimated value can be corrected to an appropriate value, and more suitable advertisement content can be selected.

  FIG. 5 is a flowchart illustrating an example of a flow of a process performed by the information processing device according to the first embodiment. The process of this flowchart is repeatedly executed at a frequency of, for example, once an hour. Note that the processing by the correction coefficient learning unit 140 may be executed asynchronously with the processing in the flowchart of FIG.

  First, the information processing device extracts advertisement contents as distribution candidates (S200). Next, the prediction index deriving unit 110 derives a prediction CTR for each advertisement 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 performs the ranking again taking into account the correction result in S206 (replace the predicted CTR before correction with the corrected predicted CTR) (S208).

  Then, the information processing apparatus determines whether or not the processing of S206 and S208 has been performed N times (S210). If it is determined that the execution has not been performed N times, the process returns to S206. If it is determined that the execution has been performed N times, the process proceeds to S212. Note that N = 1 may be set as described above, and in that case, the processing itself of S210 is omitted.

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

  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 finally corrected result to the content providing unit 150” may be performed. . In this way, the content providing unit 150 can calculate an index such as eCPM, CPM, or CPC while 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) regarding a content is predicted for each of a plurality of content candidates that are candidates for content (for example, advertising content) provided to the terminal device 10. A prediction index deriving 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 deriving unit 110; And a correction processing unit 130 that corrects the prediction index based on the result of the ranking by the content providing unit 150, so that the content providing unit 150 can select more suitable content.

Second Embodiment
Hereinafter, a second embodiment will be described. In the first embodiment, the correction processing unit 130 corrects the predicted CTR of the ranked advertisement content. The correction processing unit according to the second embodiment (hereinafter, referred to as a correction processing unit 130A) further includes an extended correction coefficient for at least a part of the advertising content candidates not ranked based on the distribution of the correction coefficient α for each rank. α + is set, and the predicted CTR is corrected by multiplying the predicted CTR by the correction coefficient α or the extended correction coefficient α + set for each rank.

  FIG. 6 is a diagram illustrating an example of the content of the processing by the correction processing unit 130A. A rank is given to a predetermined number X of advertising content in descending order of predicted CTR. In the example of FIG. 6, the predetermined number X = 5. Here, the predicted CTR of the advertising content having the lowest rank (rank 5 in the figure) is not significantly different from the predicted CTR of the advertising content having the highest predicted CTR next to the advertising content of rank 5 (1 in the figure). 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 advertising content of rank 1 are reversed, which may cause undesired results. There is also). Thus, the correction processing unit 130A fits the distribution of the correction coefficient α for each rank to a model ML approximating a straight line or a curve, for example, based on coordinates on a straight line or a curve obtained by extending the model ML to an area outside the rank. , The extended correction coefficient α + is set. In addition, the correction processing unit 130A may determine the extended correction coefficient α + by performing processing such as spline interpolation on the distribution of the correction coefficient α for each rank.

  Then, the correction processing unit 130A multiplies the predicted CTR of the ranked advertisement content by the correction coefficient α (rank) for each rank, and partially (the predicted CTR is high) of the unranked advertisement content. The predetermined number Y) is multiplied by the extension correction coefficient α + (rank #) in this order to correct the predicted CTR. The rank # is an argument corresponding to the order when the advertising content out of the rank is arranged in the order of the predicted CTR. As a result, it is possible to suppress the occurrence of an inappropriate reverse phenomenon of the predicted CTR. In addition, when the number of repetitions of the correction / ranking is N ≧ 2, when the reversal phenomenon occurs due to the above-described correction, the correction processing unit 130A performs the correction processing by regarding the reversed order as a new rank. .

  According to the second embodiment described above, in addition to achieving the same effects as the first embodiment, it is possible to suppress the occurrence of an inappropriate inversion of the predicted CTR.

  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 device, after the correction coefficient learning unit 140 learns the correction coefficient α, the correction coefficient learning unit 140 performs a process of setting the extended correction coefficient α +.

<Third Embodiment>
Hereinafter, the third embodiment will be described. The correction processing unit according to the third embodiment (hereinafter, referred to as a correction processing unit 130B) 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 a correction coefficient learning unit 140B) of the third embodiment converts the number of actual impressions into BIN, that is, any one of the areas (BIN) of the actual number of impressions divided by a predetermined interval. And the correction coefficient α (or the correction coefficient α and the expansion correction coefficient α +) is learned for each BIN of the number of actual impressions. The actual number of impressions is a value obtained by counting the number of impressions during a predetermined period (one day, one week, one month, etc.).

  FIG. 7 is a diagram for describing processing of the correction processing unit 130B and the correction coefficient learning unit 140B. As shown in the figure, the correction coefficient learning unit 140B converts the number of actual impressions into BINs in increments of 10000, and sets the main content corresponding to each BIN as a population, as in the first or second embodiment. To learn. Then, the correction processing unit 130B acquires the actual number of impressions of the main content to be processed, and corrects the predicted CTR using the correction coefficient α corresponding to the BIN to which the acquired actual number of impressions belongs. As a result, if the number of impressions affects the magnitude of the selection bias, the predicted CTR can be more appropriately corrected.

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

<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 given a rank, and the number of advertisement requests per day, and sets ε in order to search for a high-margin advertisement. -Using the Greedy algorithm, the advertisement content is randomly distributed for a request of ε [%], and an advertisement with a high advertisement revenue (bid amount × predicted CTR) is distributed for other requests. A simulation was performed.

  As a result of the simulation, the actual value of CTR was obtained every day, and as a result, an increase of several [%] to several tens [%] was observed. This means that the user has more opportunities to click on the advertisement content as a result of more accurately determining the predicted CTR.

  In addition, as a result of determining the bid of the advertisement by a uniform random number for each advertisement content and reflecting the same in the simulation, it was observed that the expected advertisement profit also improved. FIG. 8 is a diagram illustrating a tendency of improvement in profit obtained as a result of the simulation. As shown in the figure, as a result of suppressing the upward fluctuation of the predicted CTR due to the selection bias as compared with the comparative example, the cumulative predicted revenue is suppressed, while the revenue (expected revenue) that is actually obtained is the same as the comparative example. It was observed that it was improved in comparison with the above. The comparative example does not perform correction for the selection bias.

  As mentioned above, although the form for implementing this invention was demonstrated using embodiment, this invention is not limited to such embodiment at all, In the range which does not deviate from the summary of this invention, various deformation | transformation and substitution 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 deriving unit 120 ranking processing unit 130 correction processing unit 140 correction coefficient learning unit 150 content providing unit

Claims (10)

For each of a plurality of content candidates serving as content candidates provided to the terminal device, a prediction index deriving unit that derives a prediction index that predicts an index related to the content,
A ranking processing unit that ranks the plurality of content candidates based on the prediction index derived by the prediction index deriving unit;
A correction processing unit that corrects the prediction index based on a result of ranking by the ranking processing unit;
An information processing apparatus comprising: The correction processing unit corrects the prediction index by multiplying the prediction index by a correction coefficient corresponding to the rank of the content candidate among correction coefficients set for each rank of the prediction index,
The information processing device according to claim 1. The correction processing unit sets, for at least a part of the content candidates outside the rank, an extended correction coefficient based on a distribution of the correction coefficient for each rank, and sets the extended coefficient or the extended coefficient set for each rank of the prediction index. By multiplying the prediction index by a correction coefficient, the prediction index is corrected,
The information processing apparatus according to claim 2. The correction processing unit further corrects the prediction index based on the number of actual impressions of the main content,
The information processing apparatus according to claim 1. A prediction index derived by the prediction index deriving unit, a rank of the prediction index, and a correction coefficient for each rank based on an event that the content is selected by a user as a result provided to the terminal device. Further comprising a correction coefficient learning unit for learning
The information processing apparatus according to claim 1. The correction coefficient learning unit, for each rank, by comparing the prediction index derived by the prediction index derivation unit and the event that the content is selected by the user as a result provided to the terminal device. Learning a correction coefficient for each rank of the prediction index,
The information processing device according to claim 5. The correction processing unit corrects the prediction index by multiplying the prediction index by a correction coefficient corresponding to the rank of the content candidate, among correction coefficients set for each rank of the prediction index,
The correction coefficient learning unit learns an extended correction coefficient based on a 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 a correction coefficient or the extended correction coefficient set for each rank of the prediction index,
The information processing apparatus according to claim 5. An information processing apparatus according to any one of claims 1 to 7,
A providing unit that provides a content to the terminal device based on the prediction index output by the information processing device,
Content distribution device comprising: Computer
For each of a plurality of content candidates serving as content candidates to be provided to the terminal device, a prediction index that predicts an index related to the content is derived,
Based on the derived prediction index, perform ranking of the plurality of content candidates,
Correcting the prediction index based on the result of the ranking;
Information processing method. On the computer,
For each of a plurality of content candidates that are candidates for content provided to the terminal device, to derive a prediction index that predicts an index for the content,
Based on the derived prediction index, the ranking of the plurality of content candidates is performed,
Correcting the prediction index based on the result of the ranking;
program.

Images