WO2024095779A1 - Système de traitement d'informations et procédé de traitement d'informations - Google Patents

Système de traitement d'informations et procédé de traitement d'informations Download PDF

Info

Publication number
WO2024095779A1
WO2024095779A1 PCT/JP2023/037774 JP2023037774W WO2024095779A1 WO 2024095779 A1 WO2024095779 A1 WO 2024095779A1 JP 2023037774 W JP2023037774 W JP 2023037774W WO 2024095779 A1 WO2024095779 A1 WO 2024095779A1
Authority
WO
WIPO (PCT)
Prior art keywords
advertising
advertisement
reservation
amount
performance
Prior art date
Application number
PCT/JP2023/037774
Other languages
English (en)
Japanese (ja)
Inventor
一輝 柴田
雄介 熊谷
龍 道本
Original Assignee
株式会社博報堂Ｄｙホールディングス
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 株式会社博報堂Ｄｙホールディングス filed Critical 株式会社博報堂Ｄｙホールディングス
Publication of WO2024095779A1 publication Critical patent/WO2024095779A1/fr

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0241—Advertisements
    • G06Q30/0242—Determining effectiveness of advertisements

Definitions

  • This disclosure relates to an information processing system and method.
  • Patent Document 1 Technology for combining broadcast and network advertising has been known.
  • technology for achieving high advertising effectiveness by automatically synchronizing advertising timing and optimizing the display mode is known (see, for example, Patent Document 1).
  • a reservation-based advertisement is a type of advertisement that requires a reservation in advance for distribution.
  • a reservation-based advertisement is distributed according to a schedule determined at the time of reservation. Distribution here includes not only digital distribution but also broadcasting.
  • An example of a reservation-based advertisement is a television commercial, which is an advertisement broadcast on television.
  • Programmatic advertising is different from reservation-based advertising.
  • Programmatic advertising is a type of advertising where the delivery conditions can be changed in real time. Examples of programmatic advertising include advertising through communication networks, especially the Internet. Programmatic advertising can be delivered at a time specified by the advertiser without prior reservation.
  • Reservation-based advertising has ad slots that are highly effective. On the other hand, with reservation-based advertising, it is necessary to decide on the placement in advance. With performance-based advertising, real-time delivery is possible. On the other hand, with performance-based advertising, the advertising effectiveness may be lower than with reservation-based advertising. As such, there are advantages and disadvantages to both reservation-based advertising and performance-based advertising.
  • an information processing system includes an instruction unit, a first performance determination unit, a second performance determination unit, and a schedule determination unit.
  • the instruction unit is configured to determine a first advertising volume and a second advertising volume, and to instruct advertising placement in a first medium based on the first advertising volume, and advertising placement in a second medium based on the second advertising volume.
  • the first advertising volume is the advertising volume for reservation-type advertising exposed through the first medium.
  • the second advertising volume is the advertising volume for non-reservation-type advertising exposed through the second medium.
  • the first performance discrimination unit is configured to discriminate a first performance, which is a performance regarding the exposure of a reservation-type advertisement exposed through a first medium.
  • the second performance discrimination unit is configured to discriminate a second performance, which is a performance regarding the exposure of a non-reservation-type advertisement exposed through a second medium.
  • the schedule discrimination unit is configured to discriminate the exposure schedule of an unexposed advertisement waiting to be exposed among the reservation-type advertisements.
  • the total amount of possible advertising for reservation-type advertising and non-reservation-type advertising is determined in advance.
  • the instruction unit is configured to determine the first advertising amount and the second advertising amount for each of a plurality of time points based on the remaining possible advertising amount, the exposure schedule of unexposed advertising, the first performance, and the second performance.
  • reservation-type ads and non-reservation-type ads can be placed in a highly effective allocation. Specifically, under conditions where the total possible placement volume is set, reservation-type ads and non-reservation-type ads can be placed in a highly effective allocation through determining the placement volumes of reservation-type ads and non-reservation-type ads over multiple times.
  • a first amount of the total possible advertising volume may be predefined as the advertising volume for reservation-type advertising.
  • a second amount of the total volume may be predefined as the advertising volume shared between reservation-type advertising and non-reservation-type advertising.
  • the instruction unit may be given the authority to determine the allocation of the second amount to reservation-type advertising and non-reservation-type advertising.
  • the placement of a reservation-type advertisement corresponding to at least a portion of the first amount is completed at an initial time point that is earlier than the multiple time points.
  • the above-mentioned exposure schedule may include an exposure schedule for a reservation-type advertisement that has been placed at the initial time point and is waiting to be exposed.
  • the instruction unit can determine the first and second advertising volumes for each time point, taking into account the exposure schedule of unexposed advertisements, including reservation-based advertisements that have already been placed at the initial time point.
  • the information processing system can instruct the placement of reservation-type advertisements and non-reservation-type advertisements with high advertising effectiveness by determining the placement volumes of reservation-type advertisements and non-reservation-type advertisements over multiple times while controlling the minimum placement volume for reservation-type advertisements.
  • the first advertising volume may be the advertising amount for reservation-type advertising.
  • the second advertising volume may be the advertising amount for non-reservation-type advertising.
  • the total possible advertising volume may be the advertising budget for advertising including reservation-type advertising and non-reservation-type advertising. Therefore, by using an information processing system according to one aspect of the present disclosure, a user can realize the advertising of reservation-type advertising and non-reservation-type advertising with high advertising effectiveness within a limited budget.
  • the instruction unit may determine the first and second advertising amounts for each time point according to a dynamic optimization algorithm.
  • the dynamic optimization algorithm may include at least one of reinforcement learning, a contextual bandit algorithm, and a Kalman filter.
  • the instruction unit may determine the first and second advertising volumes for each point in time according to at least one of a reinforcement learning and a contextual bandit algorithm.
  • a reinforcement learning and a contextual bandit algorithm By determining the advertising volumes through such an algorithm, the user can realize the placement of reservation-type advertisements and non-reservation-type advertisements with high advertising effectiveness.
  • the instruction unit may determine, for each time point, a first advertising amount and a second advertising amount as an action related to advertising at the corresponding time point, based on a state or context at the corresponding time point.
  • the state or context may be defined using the possible remaining advertising amount, the exposure schedule of unexposed advertisements, the first performance, and the second performance.
  • the instruction unit can determine, for each point in time, the advertising effectiveness of the reservation advertisements and non-reservation advertisements newly exposed through advertising placement up to the corresponding point in time, based on the first performance and the second performance, as a reward for the action.
  • the instruction unit may update the state or context for each time point, taking into account the advertisement placement at the corresponding time point and the first and second results at the corresponding time points.
  • the instruction unit may update the policy regarding the selection of an action based on the reward for each time point.
  • the instruction unit can determine the first and second advertising volumes for each point in time by taking into account the exposure schedule of unexposed advertisements, including reservation-based advertisements that have already been placed at the initial point in time, as the above-mentioned state or context.
  • the instruction unit may be configured to determine, for each time point, a first advertising amount and a second advertising amount at a corresponding time point as actions through reinforcement learning in which a state, a reward, and an action are defined.
  • the state may be defined using the remaining possible advertising volume at the corresponding time, the first performance, the second performance, and the schedule of unexposed advertising.
  • the reward may be defined using the advertising effectiveness at the corresponding time determined from the first performance and the second performance.
  • the instruction unit may be configured to determine, for each time point, a first advertising volume and a second advertising volume at a corresponding time point as an action by a contextual bandit algorithm in which a context, a reward, and an action are defined.
  • the context may be defined using the remaining possible advertising volume at a corresponding time, the first performance, the second performance, and the schedule of unexposed advertising.
  • the reward may be defined using the advertising effectiveness at the corresponding time determined from the first performance and the second performance.
  • the instruction unit may be configured to determine, for each time point, a first advertising amount and a second advertising amount at the corresponding time point, for example, by using a state space model such as a Kalman filter.
  • the state space model may be a model that defines the relationship between state quantities, observation quantities, and input quantities.
  • the state space model may be a model that includes information on the exposure schedule of unexposed advertisements.
  • the input quantities may be quantities related to advertisement placements that are defined using a first placement quantity and a second placement quantity.
  • the state quantity may be a state quantity that changes due to advertising placement.
  • the state quantity may be defined using the possible remaining advertising volume, the first performance, and the second performance.
  • the observation quantity may be a quantity that defines the advertising effectiveness of the placed reservation advertisements and non-reservation advertisements, determined based on the first performance and the second performance.
  • the instruction unit may be configured to determine, for each time point, a first advertising volume and a second advertising volume based on the possible remaining advertising volume, the exposure schedule of the unexposed advertisements, the first performance, and the second performance, using a state space model such as a Kalman filter.
  • Information on the exposure schedule of the unexposed advertisements may be incorporated into the state space model by mathematically expressing the relationship between the input volume and the observed volume according to the exposure schedule.
  • a computer program may be provided for causing a computer to realize the functions of the instruction unit, the first performance determination unit, the second performance determination unit, and the schedule determination unit in the above-mentioned information processing system.
  • the computer program may be provided by being recorded on a computer-readable non-transitive physical recording medium.
  • the information processing method may include determining a first advertising volume and a second advertising volume, and instructing advertising placement in a first medium based on the first advertising volume, and advertising placement in a second medium based on the second advertising volume.
  • the first advertising volume may be an advertising volume for reservation-type advertising exposed through the first medium.
  • the second advertising volume may be an advertising volume for non-reservation-type advertising exposed through the second medium.
  • the information processing method may include determining a first performance, which is a performance regarding the exposure of a reservation-type advertisement exposed through a first medium.
  • the information processing method may include determining a second performance, which is a performance regarding the exposure of a non-reservation-type advertisement exposed through a second medium.
  • the information processing method may include determining an exposure schedule for reservation-type advertisements that are waiting to be exposed.
  • the total amount of possible placements for reservation-type advertisements and non-reservation-type advertisements may be determined in advance.
  • Instructing the placement of advertisements may include determining, for each of a plurality of time points, a first placement volume and a second placement volume based on the remaining possible placement volume, the exposure schedule of unexposed advertisements, the first performance, and the second performance.
  • the information processing method may be executed by a computer.
  • the first advertising volume and the second advertising volume may be determined for each point in time according to a dynamic optimization algorithm, such as reinforcement learning, a contextual bandit algorithm, and a Kalman filter.
  • a dynamic optimization algorithm such as reinforcement learning, a contextual bandit algorithm, and a Kalman filter.
  • an information processing method corresponding to the above-mentioned information processing system may be provided.
  • a computer program may be provided that includes instructions that, when executed by a computer, cause the computer to execute the information processing method described above.
  • a computer-readable non-transitive tangible recording medium that stores the computer program may be provided.
  • FIG. 1 is a block diagram showing a configuration of an information processing system.
  • 11 is a flowchart showing a first advertisement-related process executed by a processor in the first embodiment.
  • FIG. 13 is an explanatory diagram regarding the use of the reserve budget.
  • 11 is a flowchart showing a second advertisement-related process executed by a processor in the first embodiment.
  • 13 is a flowchart showing a first advertisement-related process executed by a processor in the second embodiment.
  • 13 is a flowchart showing a second advertisement-related process executed by a processor in the second embodiment.
  • FIG. 13 is an explanatory diagram relating to the determination of an advertising amount using a Kalman filter.
  • the information processing system 1 of the present embodiment is configured by installing a computer program specific to the present embodiment in a general-purpose computer system.
  • the information processing system 1 shown in Fig. 1 includes a processor 11, a memory 12, a storage 13, a display 15, an input device 17, a media reader/writer 18, and a communication device 19.
  • the processor 11 is configured to execute processing in accordance with a computer program stored in the storage 13.
  • the memory 12 includes a RAM.
  • the memory 12 is used as a working area when the processor 11 executes processing in accordance with the computer program.
  • the memory 12 temporarily stores the computer program and data read from the storage 13.
  • Storage 13 stores computer programs and various data.
  • One of the computer programs stored in storage 13 includes a computer program that causes processor 11 to determine in stages the placement volume for each of reservation-type advertisements and non-reservation-type advertisements based on a dynamic optimization algorithm so as to maximize the advertising effect within a specified advertising budget.
  • Examples of storage 13 include hard disk drives and solid-state drives.
  • the display 15 is configured to display various information to the user.
  • the display 15 is, for example, a liquid crystal display.
  • the input device 17 is configured to input operation signals from the user to the processor 11.
  • the input device 17 includes a keyboard and a pointing device that can be operated by the user.
  • the media reader/writer 18 is configured to be capable of reading information recorded on a recording medium such as a memory card, and to be capable of writing new information to the recording medium.
  • the communication device 19 is controlled by the processor 11 and configured to communicate with external systems within the local area network and/or the wide area network.
  • Processor 11 executes advertisement-related processing based on a computer program stored in storage 13 in accordance with instructions input by the user through input device 17.
  • the advertisement-related processing includes a first advertisement-related processing (see FIG. 2) and a second advertisement-related processing (see FIG. 4). After executing the first advertisement-related processing, processor 11 executes the second advertisement-related processing in accordance with further instructions input by the user through input device 17.
  • the minimum advertising budget for multiple media is determined under specified conditions, and an advertising plan is output.
  • the advertising volume for multiple media is determined based on the minimum advertising budget set in the advertising plan, and advertising placement is instructed.
  • the multiple media include multiple reservation media and multiple non-reservation media.
  • Each reservation media is a medium that delivers reservation advertisements.
  • Each non-reservation media is a medium that delivers non-reservation advertisements.
  • Advertising distribution means broadcasting an advertisement.
  • Advertising distribution is not limited to advertising distribution via the Internet.
  • Advertising distribution includes broadcasting advertisements via television broadcasting, radio broadcasting, etc. Advertisements are exposed to consumers through the medium.
  • An advertisement may be, for example, related to a product sold by the advertiser or a service provided by the advertiser.
  • Non-reserved advertisement is a programmatic advertisement that is distributed almost simultaneously with its placement, and is a digital advertisement distributed via the Internet.
  • An example of a reserved advertisement is an advertisement that is distributed via television broadcasting in the form of a television commercial (CM).
  • CM television commercial
  • Television commercials include spot commercials.
  • Reservation-type advertisements as television commercials are broadcast in a specified broadcast time slot according to a schedule determined at the time of advertisement placement.
  • reservation-type advertisements as television commercials are broadcast on a specified broadcast channel at a specified broadcast date and time.
  • a television commercial as a reservation-type advertisement and a digital advertisement as a non-reservation-type advertisement.
  • the following information is given by the user as execution conditions. ⁇ Advertising budget ⁇ Campaign period ⁇ Estimation model of advertising effectiveness ⁇ Target information ⁇ Penalty
  • the processor 11 When the processor 11 starts the first advertisement-related process (see FIG. 2), in S110, the processor 11 acquires this information through the storage 13 or through the input device 17.
  • the advertising budget is the total advertising budget, including reservation-based and non-reservation-based advertising, that can be used in the campaign.
  • the campaign here corresponds to the advertising campaign.
  • the campaign period corresponds to the period during which advertising activities are carried out.
  • the advertising effectiveness estimation model is a mathematical model for estimating advertising effectiveness from indicators related to advertising exposure performance.
  • Advertising effectiveness can be quantified as reach, which is the number of target consumers reached by an advertisement.
  • advertising effectiveness can be quantified as conversion rate.
  • advertising effectiveness can be expressed by quantifying the magnitude of brand lift. Advertising effectiveness may also be expressed numerically as a score that combines multiple indicators such as reach, conversion rate, and brand lift.
  • Targeting information is information that specifies the consumer segment targeted by advertising.
  • the effectiveness of advertising is estimated as the effect on the consumer segment specified by the targeting information.
  • the consumer segment is specified, for example, by gender and age group.
  • the penalty is a parameter that prevents advertising-related processes from instructing advertising that exceeds a specified advertising budget, which is the total amount of advertising that can be performed.
  • the total amount of advertising that can be performed means the total amount of advertising that can be performed, at least up to the corresponding amount. It should not be interpreted as prohibiting advertising that exceeds the "allowable" advertising amount.
  • the advertisement budget here is a purchasing budget for advertising space.
  • b m R (0) is the advertising budget for the mth reservation-type medium among multiple reservation-type media.
  • M is the number of reservation-type media when reservation-type media exist from the first reservation-type medium to the Mth reservation-type medium.
  • M is an integer value of 1 or more.
  • m is an integer value from 1 to M.
  • b n P (0) is the advertising budget for the nth non-reservation medium among a plurality of non-reservation media.
  • N is the number of non-reservation media when the non-reservation media range from the first non-reservation medium to the Nth non-reservation medium.
  • N is an integer value of 1 or more.
  • n is an integer value from 1 to N.
  • the initial state s(0) is defined by an advertising budget.
  • minimum advertising budgets b 1 R (0), b 2 R (0), ..., b M R (0), b 1 P (0), b 2 P (0), ..., b N P (0) for multiple reservation media and multiple non-reservation media are determined based on the specified advertising budget information and also based on a previously learned policy regarding action selection.
  • the processor 11 outputs, as an advertising plan, the minimum advertising budgets b 1 R (0), b 2 R (0), ..., b M R (0), b 1 P (0), b 2 P (0), ..., b N P (0) for each medium determined in S120, together with the advertising budget and reserve budget.
  • the advertising budget is as described above.
  • the reserve budget corresponds to the value obtained by subtracting the sum of the minimum spending budgets b1R(0), b2R ( 0 ) , ..., bMR(0), b1P (0), b2P (0), ..., bNP ( 0) from the advertising budget.
  • the reserve budget corresponds to the part of the advertising budget that is not assigned to either reservation-type media or non-reservation-type media.
  • the processor 11 can output the advertising plan by displaying it on the display 15.
  • the processor 11 can output the advertising plan to the storage 13 as a data file that can be viewed by the user.
  • the processor 11 can store information about the advertising plan that is necessary for the second advertising-related process in the storage 13. Thereafter, the processor 11 ends the first advertising-related process.
  • the advertising plan is presented to the advertiser before the start of the campaign period. If the advertising plan is adopted by the advertiser, advertising space for the reservation-type media is purchased before the start of the campaign period for the minimum advertising budget b 1 R (0), b 2 R (0), ..., b M R (0) according to the advertising plan, and advertising work for each reservation-type media is completed. This advertising work is performed manually by a person in charge of the advertising agency or automatically through a relay system 31 with the media.
  • the distribution schedule for the placed reservation-based advertisement is recorded in the storage 13.
  • the processor 11 can automatically obtain the distribution schedule through the relay system 31.
  • the advertising agency staff can obtain the distribution schedule from the media company and input it into the information processing system 1 through the input device 17 or the media reader/writer 18.
  • the processor 11 starts a second advertising-related process (see FIG. 4 ) when an instruction to execute a process based on the advertising plan is input by the user via the input device 17.
  • advertising placement in non-reservation-type media and additional advertising placement in reservation-type media are determined based on the advertising budget B according to the advertising plan, the minimum advertising budgets b 1 R (0), b 2 R (0), ..., b M R (0), b 1 P (0 ) , b 2 P (0), ..., b N P (0), and the reserve budget Br.
  • advertising to reservation-type media based on the minimum advertising budget b1R(0), b2R ( 0 ) , ..., bMR (0) is completed. Meanwhile, advertising to non-reservation-type media based on the minimum advertising budget b1P (0), b2P (0 ) , ..., bNP (0) has not been carried out. Advertising to non-reservation-type media based on the minimum advertising budget b1P (0), b2P (0), ..., bNP ( 0 ) is carried out after the start of the campaign period.
  • the reserve budget Br is used in stages for placing reservation-type advertisements or non-reservation-type advertisements during the campaign period.
  • the advertising budget for reservation-type media before the start of the campaign period is "40”
  • the advertising budget for non-reservation-type media is "20”
  • the reserve budget is "10”.
  • the numbers are examples and the units are arbitrary.
  • the processor 11 reads from the storage 13 the delivery schedule of the posted reservation-type advertisement that has been posted in advance according to the advertising plan (S210).
  • the state variable consists of six terms.
  • the first term, "1/T,” indicates the percentage of the campaign period that has elapsed on the first day of the campaign period.
  • t/T indicates the percentage of the campaign period that has elapsed at time t.
  • time t means the tth day of the campaign period.
  • t is a discrete time that represents a time within the campaign period in days.
  • T represents the length of the campaign period, specifically, the number of days in the campaign period.
  • B(t)/B is the ratio of the remaining budget B(t) to the advertising budget B on the tth day of the campaign period.
  • the remaining budget B(1) is calculated using the following formula.
  • the budget balance B(1) corresponds to the reserve budget Br.
  • ⁇ 0 ⁇ 1 ⁇ m ⁇ M is a one-dimensional array with M elements and the value of each element being zero.
  • GRP is the gross rating point.
  • G m, ⁇ (1) is the GRP estimated on the first day of the campaign period, and is the sum of the estimated viewership ratings of the placed but not yet delivered reservation-type advertisements to be delivered on the corresponding reservation-type medium (m) on the corresponding day ⁇ according to the delivery schedule. That is, the fifth term is the array ⁇ G m, ⁇ (1) ⁇ 1 ⁇ m ⁇ M, 1 ⁇ T of G m, ⁇ ( 1) for each combination of m, ⁇ in the ranges of 1 ⁇ m ⁇ M and 1 ⁇ T .
  • the processor 11 can generate an array ⁇ G m, ⁇ (1) ⁇ 1 ⁇ m ⁇ M, 1 ⁇ T according to the distribution schedule acquired in S210.
  • the estimated GRP can be calculated from the past record of the viewership rating for the same time slot. Record data of the viewership rating required for the calculation can be stored in the storage 13 in advance.
  • the processor 11 determines the advertising volume ( b1R (t), b2R ( t), ..., bMR (t ) , b1P (t), b2P (t), ..., bNP (t)) based on the state s(t) and the policy regarding behavior selection.
  • bmR (t) is the advertising volume for the mth reservation-type medium at time t.
  • bnP (t) is the advertising volume for the nth non-reservation-type medium at time t.
  • the advertising volume for each medium is specifically the advertising amount for each medium, in other words, the purchase amount of advertising space for each medium.
  • the policy is updated by the reward r(t) described below.
  • multiple submission amounts with different quantities are defined as options for behavior.
  • the submission amount is determined by selecting one quantity from the options as an action selection according to the policy.
  • the processor 11 instructs advertising placement to each medium based on the placement volume ( b1R (t), b2R (t), ..., bMR ( t), b1P(t), b2P ( t), ..., bNP ( t)).
  • the processor 11 can instruct the relay system 31, for example, to place an advertisement.
  • the relay system 31 can automatically place an advertisement in each medium system according to the instructed content.
  • the processor 11 can also instruct the user to place an advertisement in each medium by displaying the amount of advertisement through the display 15. The user can at least partially manually place an advertisement in each medium according to the displayed content.
  • the processor 11 determines the average viewer rating PmR (t) of the reservation-type advertisements distributed through each reservation-type medium from the first reservation-type medium to the Mth reservation-type medium as the exposure performance.
  • I m R (t) represents the exposure amount of a reservation-type advertisement distributed through the mth reservation-type medium at time t.
  • P m R (t) represents the average viewership rate of the mth reservation-type medium at time t
  • c m R (t) represents the advertising cost, i.e., the advertising amount, of an advertisement distributed through the mth reservation-type medium at time t.
  • the processor 11 can obtain information on the average viewership rate P m R (t) of each reservation-type medium, for example, from the measurement system 35 that measures viewing behavior via the communication device 19, and determine the average viewership rate P m R (t).
  • the processor 11 determines the CPM (Cost per Mille): PnP(t) of the non-reservation advertisement distributed through the non-reservation medium at time t for each of the first to Nth non-reservation media as the exposure performance.
  • InP (t) represents the impressions of the nth non-reserved medium at time t
  • PnP (t) represents the CPM of a non-reserved advertisement in the nth non-reserved medium at time t
  • cnP (t) represents the advertising placement cost in the nth non-reserved medium at time t
  • cnP (t) represents the amount of advertising actually required for placement based on bnP (t).
  • the processor 11 calculates the reward r(t) at time t. Specifically, the processor 11 inputs the amount of exposure of the reservation-type advertisement through the reservation-type medium up to time t ⁇ I m R (s) ⁇ 1 ⁇ m ⁇ M, 1 ⁇ s ⁇ t , and the impressions ⁇ I n P (s) ⁇ 1 ⁇ n ⁇ N, 1 ⁇ s ⁇ t , which are the amount of exposure of the non-reservation-type advertisement through the non-reservation-type medium up to time t, into an estimation model of the advertising effectiveness. As a result, the processor 11 obtains the advertising effectiveness Z(t) at time t from the estimation model. The processor 11 calculates the difference Z(t)-Z(t-1) between the advertising effectiveness Z(t) at time t and the advertising effectiveness Z(t-1) at the previous time as the reward r(t).
  • the reward r(t) is defined as the negative reward ⁇ C, where C is a positive value and is the penalty described above.
  • the processor 11 updates the state s(t) to the state at the next time point t when the next time point t arrives. That is, the processor 11 sets the state s(t) as follows. At this time, the latest distribution schedule at the time point t is obtained to set the state s(t).
  • the third term represents the cost performance of each reservation-type medium at time t with respect to advertisement delivery.
  • the cost performance of the mth reservation-type medium is quantified as the average value of the average audience rating P m R ( ⁇ ) for the mth reservation-type medium at each time ⁇ from the first day of the campaign period to the time t-1 immediately preceding the present.
  • the third term is an array whose elements are the average values (1/(t-1)) ⁇ P m R ( ⁇ ) for each medium from the first reservation-type medium to the Mth reservation-type medium.
  • the fourth term represents the cost performance of each non-reservation medium at time t with respect to advertisement delivery.
  • the cost performance of the nth non-reservation medium is quantified as the average value of CRM: P n P ( ⁇ ) for the nth non-reservation medium at each time ⁇ from the first day of the campaign period to the time t-1 immediately preceding the current time.
  • the fourth term is an array having as its elements the average CRM value (1/(t-1)) ⁇ P n P ( ⁇ ) for each medium from the first non-reservation medium to the Nth non-reservation medium.
  • the fifth term is an array of GRPs for each day ⁇ of the campaign period (1 ⁇ T) for each reservation type medium from the first reservation type medium to the Mth reservation type medium estimated at time t.
  • the fifth term is an array of G m, ⁇ (t) for each combination of m and ⁇ in the ranges of 1 ⁇ m ⁇ M and 1 ⁇ T ⁇ G m, ⁇ (t) ⁇ 1 ⁇ m ⁇ M, 1 ⁇ T .
  • G m, ⁇ (t) is the estimated GRP on the tth day of the campaign period, which is the sum of the estimated viewership ratings of the placed but not yet delivered reservation-type advertisements to be delivered on the corresponding day ⁇ through the corresponding reservation-type medium (m) according to the delivery schedule.
  • G m, ⁇ (t) also takes into account the viewer ratings according to the delivery schedule of reservation-based advertisements that are additionally placed after the start of the campaign period.
  • viewer ratings for advertisements that have already been delivered are not used in the calculation of GRP. Therefore, all G m, ⁇ (t) corresponding to periods ⁇ t prior to the current time t are zero.
  • the reward r(t) is reflected in the reward r(t) as advertising effectiveness based on the exposure amount I m R (t) as described above.
  • the sixth term is an array ⁇ W n (t) ⁇ of the remaining advertising budget for each non-reserved medium at time t.
  • the balance W n (t) is the remaining advertising budget for the nth non-reserved medium at time t.
  • the balance W n (t) is updated according to the following formula.
  • c n P (t-1) is the placement cost for the advertisement distributed through the nth non-reservation medium at time t-1.
  • the budget balance B(t) at time t is calculated by the following formula.
  • the function E(c n P (t-1)) is a function that outputs the value 1 when the advertising cost c n P (t-1) of the nth non-reserved medium at time t-1 is greater than the budget balance W n (t-1) of the nth non-reserved medium at time t-1, and outputs the value 0 otherwise.
  • the budget balance B(t) at time t corresponds to the budget balance B(t-1) at time t-1 minus the total of the reserve budget consumed for placing reservation-type ads and non-reservation-type ads at time t-1.
  • the processor 11 further instructs, in S250, to place advertisements in each medium based on the placement volume ( b1R (t), b2R (t), ..., bMR ( t), b1P (t), b2P (t), ..., bNP (t)).
  • the processor 11 calculates the reward r(t).
  • the state s(t ) is updated.
  • processor 11 repeatedly selects an action based on state s(t) (i.e., determines the amount of advertising), calculates reward r(t) based on the action, and updates state s(t). In this way, processor 11 determines the amount of advertising for reservation media and non-reservation media at each point in time using the reserve budget Br, in a direction that maximizes the advertising effect, which is reward r(t), in accordance with a reinforcement learning algorithm. The policy regarding action selection is adjusted based on reward r(t) in accordance with a reinforcement learning algorithm.
  • the processor 11 ends the second advertising-related process.
  • a reinforcement learning algorithm is used to determine advertising in multiple media, including reservation-based media and non-reservation-based media, so as to maximize advertising effectiveness within a predetermined advertising budget.
  • this problem is solved by adding information about the GRP of undelivered advertisements to the state s(t).
  • information on the time lag between posting and delivery of reservation-based advertising is stored as a state s(t) in the form of the GRP of undelivered advertising. This makes it possible to determine the amount of advertising to multiple media, including reservation-based media and non-reservation-based media, so as to minimize the effects of delays and maximize advertising effectiveness.
  • an information processing system 1 that is useful for advertising across different media, including reservation-based media and non-reservation-based media.
  • the reservation-based advertisement is a television commercial and the non-reservation-based advertisement is a programmatic advertisement that is digitally distributed.
  • the reservation-based advertisement may also be a radio commercial or a digitally distributed advertisement.
  • a negative reward r1(t) may be added to the reward r(t).
  • the amount of advertising is determined to be more aggressive before the start of the campaign period or in the early part of the campaign period.
  • the processor 11 can obtain information on the tuning parameter ⁇ of the reward r1(t) through the storage 13 or through the input device 17 in the first step (S110) of the advertisement submission-related process.
  • ⁇ is a positive real number.
  • the information processing system 1 of the second embodiment differs from the first embodiment in that a contextual bandit algorithm is used instead of a reinforcement learning algorithm for stepwise determination of the advertising volume.
  • the information processing system 1 of the second embodiment is basically configured in the same manner as the first embodiment.
  • the hardware configuration of the information processing system 1 in the second embodiment is the same as that in the first embodiment. Therefore, in the following, the contents of the processing related to the contextual bandit algorithm will be selectively explained, and other explanations will be omitted.
  • a reinforcement learning algorithm selects an action based on a defined state, action, and reward to maximize the reward.
  • the action is a decision action for the amount of advertising to be placed in each medium.
  • the contextual bandit algorithm is an algorithm that selects an action to maximize reward based on a defined context, action, and reward.
  • the action and reward are defined in the same way as in the first embodiment. That is, the action is the action of determining the amount of advertising to be placed in each medium. The reward is the advertising effect obtained by delivering advertisements through each medium.
  • the processor 11 executes a first advertising-related process (see FIG. 5) and a second advertising-related process (see FIG. 6) using a contextual bandit algorithm as advertising-related processes based on instructions from a user.
  • the variables are defined as follows:
  • the number of days in the campaign period is T.
  • the number of media for reservation-type advertising is M
  • the M reservation-type media include the mth reservation-type medium.
  • m is an integer value in the range of 1, 2, ..., M.
  • the number of media for non-reserved advertising is N, and the N non-reserved media include the nth non-reserved medium.
  • n is an integer value ranging from 1, 2, ..., N.
  • the behavior space A is an (M + N)-dimensional discrete behavior space, with each dimension having multiple elements corresponding to options for advertising volume.
  • an average viewership rating PmR (t) is observed for the mth reservation medium.
  • a CPM: PnP ( t) is observed for the nth non-reservation medium.
  • the "GRP of reservation-type advertisements delivered on the ⁇ th day of the campaign period through the mth reservation-type medium" estimated at time t is Gm , ⁇ (t).
  • UMB Upper Confidence Bound
  • the processor 11 determines the minimum advertising budgets b 1 R (0), b 2 R (0), ..., b m R (0), ..., b M R (0), b 1 P (0), b 2 P (0), ..., b n P (0), ..., b N P (0) for a plurality of reservation-type media and a plurality of non-reservation-type media.
  • the processor 11 outputs the minimum advertising budgets b1R (0), b2R (0), ... , bMR (0), b1P ( 0), b2P ( 0), ..., bNP ( 0) for each medium determined in S320 as an advertising plan. Thereafter, the processor 11 ends the first advertising - related process.
  • the processor 11 After the first advertising-related process is completed, when an instruction to execute a process based on the advertising plan is input by the user via the input device 17, the processor 11 starts the second advertising-related process shown in FIG. 6. In the second advertising-related process, the processor 11 reads from the storage 13 the delivery schedule of the reservation-type advertisement that was previously placed according to the advertising plan (S410).
  • the processor 11 can generate the array ⁇ G m, ⁇ (1) ⁇ 1 ⁇ m ⁇ M, 1 ⁇ T in the same manner as in the first embodiment, according to the distribution schedule acquired in S410.
  • the processor 11 determines the advertising volume (b 1 R (t), b 2 R (t), ..., b M R ( t ) , b 1 P (t), b 2 P (t), ..., b N P (t)) for each reservation-type medium and non - reservation-type medium at time t as behavior related to advertising placement at time t using a contextual bandit algorithm.
  • b m R (t) is the advertising volume of reservation-type advertising for the mth reservation-type medium at time t.
  • b n P (t) is the advertising volume of non-reservation-type advertising for the nth non-reservation-type medium at time t.
  • the processor 11 instructs advertising placement to each medium based on the placement volume ( b1R (t), b2R (t), ..., bMR ( t), b1P(t), b2P ( t), ..., bNP ( t)).
  • the processor 11 determines the average viewership PmR(t) of the reservation-type advertisements distributed through each reservation-type medium from the first reservation-type medium to the Mth reservation-type medium by acquiring the corresponding information.
  • the processor 11 determines the CPM PnP(t) of the non-reservation advertisement delivered through the non-reservation medium at time t for each of the first to Nth non-reservation media by acquiring corresponding information.
  • the processor 11 waits for the next time point t to arrive, and then updates the context xt . That is, the processor 11 sets the context xt as follows.
  • G m, ⁇ (t) The calculation of G m, ⁇ (t) is the same as the process in S290 in the first embodiment.
  • the budget balance B(t) at time t is calculated as follows.
  • the processor 11 repeatedly selects an action based on the context xt (i.e., determines the amount of advertising), calculates the reward r(t) based on the action, and updates the context xt . In this manner, the processor 11 determines the amount of advertising for reservation-type media and non-reservation-type media at each time point in a direction that maximizes the advertising effect, which is the reward r(t), according to the contextual bandit algorithm.
  • the processor 11 ends the second advertising-related process.
  • the information processing system 1 of this embodiment described above it is possible to appropriately determine advertising in multiple media, including reservation-based media and non-reservation-based media, in accordance with the contextual bandit algorithm so as to maximize advertising effectiveness.
  • this information processing system 1 is useful for advertising across different media, including reservation-based media and non-reservation-based media.
  • first embodiment uses a reinforcement learning algorithm (first embodiment) and a contextual bandit algorithm (second embodiment) as dynamic optimization algorithms to achieve optimal advertising across different media, including reservation-based media and non-reservation-based media.
  • second embodiment uses a reinforcement learning algorithm (first embodiment) and a contextual bandit algorithm (second embodiment) as dynamic optimization algorithms to achieve optimal advertising across different media, including reservation-based media and non-reservation-based media.
  • the information processing system 1 may be configured to determine the amount of advertising in reservation-based media and non-reservation-based media at each point in time during the campaign period using a state space model, in particular a Kalman filter.
  • the Kalman filter is a type of dynamic control technique that is composed of a state equation and an observation equation.
  • the state equation and the observation equation in the linear model are expressed by the following equations, and define the relationship between the state quantity xt , the observation quantity yt , and the control input ut .
  • the advertising amount in each medium at time t can be allocated to the control input u t as shown in Fig. 7.
  • the control input u t can be associated with the action in the reinforcement learning algorithm.
  • the state quantity xt can be assigned with a budget balance, CPM, viewership rate, etc. at time t. That is, the state quantity xt can be associated with a state s(t) in a reinforcement learning algorithm.
  • the observed quantity yt can be assigned with an advertising effect at time t. That is, the observed quantity yt can be associated with a reward r(t) in a reinforcement learning algorithm.
  • Reward indicators may include advertising contact indicators such as reach and frequency, as well as indicators related to consumer awareness.
  • Indicators related to consumer awareness may include indicators of awareness, interest, concern, understanding, purchase intent, first recall, and continued purchase intent.
  • Reward indicators may include indicators obtained by the advertiser, specifically, the sales amount of the advertised product, the number of sales, the number of accesses to the product introduction site, the number of uses of the advertised service, the number of installations of related application software, MAU (Monthly Active Users), and the number of inquiries related to the advertisement.
  • Indicators of status that should be considered when choosing an action to take regarding advertising may include indicators related to reach and the cost required for acquisition, specifically, CPM, CPC (Cost Per Click), CPA (Cost Per Acquisition), and ROI (Return on Investment), etc.
  • CTR Click Through Rate
  • CVR Conversion Rate
  • VTR View Through Rate
  • TARP Target Audience Rating Point
  • the preferred indicator for the state varies depending on the indicator used as the reward for reinforcement learning.
  • examples of indicators that are preferred for the state indicator are viewership rating (or GRP) and attention rate.
  • examples of indicators that are preferred for the state indicator are CPM and vCPM (viewable cost per mille). The same applies to non-reservation-type advertisements.
  • examples of preferred status indicators are viewer ratings (or GRP) and attention rates.
  • examples of preferred status indicators are CPM, number of conversions, CPA, and CPC.
  • examples of preferred status indicators are the number of conversions and CPA. If the non-reserved ads are distributed digitally, examples of preferred status indicators are the number of conversions, CPA, and CPC.
  • an example was given of advertising in multiple reservation media and multiple non-reservation media, but at least one of the reservation media and non-reservation media to which advertising is targeted may be limited.
  • a common advertising volume may be determined for multiple reservation media.
  • a common advertising volume may be determined for multiple non-reservation media.
  • the above embodiment includes the concept of additionally placing a reservation-type advertisement, but may further include the concept of canceling a reservation for a reservation-type advertisement (in other words, canceling the placement).
  • the action options in the dynamic optimization algorithm may include the action of canceling a reservation for a placed reservation-type advertisement.
  • the action of canceling a reservation corresponds to receiving a refund of the purchase amount for the advertising space due to the reservation cancellation. Therefore, the concept of reservation cancellation can be introduced into the dynamic optimization algorithm by introducing negative advertising volume. In this way, the present disclosure can determine the advertising volume taking reservation cancellation into consideration. According to the reinforcement learning algorithm, when determining the negative advertising volume, the delivery schedule of the advertisement held as the state can be reduced and the budget balance B(t) can be changed in an increasing direction.
  • the function of one component in the above embodiments may be distributed among multiple components.
  • the functions of multiple components may be integrated into one component.
  • a part of the configuration of the above embodiments may be omitted.
  • At least a part of the configuration of the above embodiments may be added to or substituted for the configuration of another of the above embodiments.

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Accounting & Taxation (AREA)
  • Development Economics (AREA)
  • Strategic Management (AREA)
  • Finance (AREA)
  • Game Theory and Decision Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

Selon la présente invention, une première quantité de placement de publicité et une seconde quantité de placement de publicité sont déterminées, et un placement de publicité sur un premier support sur la base de la première quantité de placement de publicité et un placement de publicité sur un second support sur la base du second placement de publicité sont ordonnés. Une première réalisation relative à l'exposition d'une publicité de type réservation est déterminée. Une seconde réalisation relative à l'exposition d'une publicité de type non réservation est déterminée. Un programme d'exposition pour une publicité non exposée en attente d'exposition est déterminé. Concernant une pluralité de points temporels, la première quantité de placement de publicité et la seconde quantité de placement de publicité sont déterminées, pour chaque point temporel, sur la base d'une quantité de publicité restante possible, du calendrier d'exposition pour la publicité non exposée, de la première réalisation et de la seconde réalisation.
PCT/JP2023/037774 2022-10-31 2023-10-18 Système de traitement d'informations et procédé de traitement d'informations WO2024095779A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2022174867A JP7319443B1 (ja) 2022-10-31 2022-10-31 情報処理システム、コンピュータプログラム、及び情報処理方法
JP2022-174867 2022-10-31

Publications (1)

Publication Number Publication Date
WO2024095779A1 true WO2024095779A1 (fr) 2024-05-10

Family

ID=87469722

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2023/037774 WO2024095779A1 (fr) 2022-10-31 2023-10-18 Système de traitement d'informations et procédé de traitement d'informations

Country Status (2)

Country Link
JP (1) JP7319443B1 (fr)
WO (1) WO2024095779A1 (fr)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2007075544A2 (fr) * 2005-12-19 2007-07-05 Jeff Apple Systemes, appareils, procedes et produits de programmes informatiques pour optimiser l'affectation d'un budget publicitaire qui maximise les ventes et/ou profits et pour permettre a des annonceurs d'acheter des medias en ligne
WO2009050546A1 (fr) * 2007-10-16 2009-04-23 Neosaej Inc. Procédé et appareil pour du commerce électronique basé sur internet
JP2013041335A (ja) * 2011-08-11 2013-02-28 Yahoo Japan Corp 出稿リソース最適化システム、出稿リソース最適化方法、および出稿リソース最適化プログラム
JP2015028732A (ja) * 2013-07-30 2015-02-12 株式会社ビデオリサーチ 出稿先選定装置及び出稿先選定方法
JP2016505996A (ja) * 2013-01-10 2016-02-25 ロヴィ テクノロジーズ コーポレイション データ駆動型メディア配置のためのシステムおよび方法
JP2019169079A (ja) * 2018-03-26 2019-10-03 株式会社ブレインパッド 配分装置、配分方法、プログラム
JP2022088339A (ja) * 2020-12-02 2022-06-14 株式会社オリコミサービス 情報処理装置、方法、プログラム、プラットフォームを提供するコンピュータ実装方法

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2007075544A2 (fr) * 2005-12-19 2007-07-05 Jeff Apple Systemes, appareils, procedes et produits de programmes informatiques pour optimiser l'affectation d'un budget publicitaire qui maximise les ventes et/ou profits et pour permettre a des annonceurs d'acheter des medias en ligne
WO2009050546A1 (fr) * 2007-10-16 2009-04-23 Neosaej Inc. Procédé et appareil pour du commerce électronique basé sur internet
JP2013041335A (ja) * 2011-08-11 2013-02-28 Yahoo Japan Corp 出稿リソース最適化システム、出稿リソース最適化方法、および出稿リソース最適化プログラム
JP2016505996A (ja) * 2013-01-10 2016-02-25 ロヴィ テクノロジーズ コーポレイション データ駆動型メディア配置のためのシステムおよび方法
JP2015028732A (ja) * 2013-07-30 2015-02-12 株式会社ビデオリサーチ 出稿先選定装置及び出稿先選定方法
JP2019169079A (ja) * 2018-03-26 2019-10-03 株式会社ブレインパッド 配分装置、配分方法、プログラム
JP2022088339A (ja) * 2020-12-02 2022-06-14 株式会社オリコミサービス 情報処理装置、方法、プログラム、プラットフォームを提供するコンピュータ実装方法

Also Published As

Publication number Publication date
JP2024065818A (ja) 2024-05-15
JP7319443B1 (ja) 2023-08-01

Similar Documents

Publication Publication Date Title
US10185971B2 (en) Systems and methods for planning and executing an advertising campaign targeting TV viewers and digital media viewers across formats and screen types
US7835939B1 (en) Systems and methods for predicting advertising revenue
Naik et al. Planning media schedules in the presence of dynamic advertising quality
AU2012294601B2 (en) Cross-media attribution model for allocation of marketing resources
US20130268347A1 (en) Yield management system and method for advertising inventory
US10102547B2 (en) Deal scheduling based on user location predictions
US20100223641A1 (en) System and method for predicting the optimum delivery of multimedia content based on human behavior patterns
JP2002007884A (ja) 情報端末、方法、記憶媒体、及びビデオ情報表示ビジネスシステム
US20070214025A1 (en) Business engagement management
JP6621954B1 (ja) 計画策定装置及びコンピュータプログラム
US20180114242A1 (en) Deal recommendation based on triggering event
EP2359328A1 (fr) Système de fourniture d'informations relatives à l'efficacité d'une publicité
US20100223637A1 (en) System and method for optimizing the pricing of multimedia content delivery
US20220076199A1 (en) Flexible assignment of media advertising orders to broadcast inventory
JP2017037432A (ja) 入札調整装置、入札調整方法および入札調整プログラム
WO2024095779A1 (fr) Système de traitement d'informations et procédé de traitement d'informations
US20140188631A1 (en) Bid system for advertisement offerings
Seifert et al. Exclusivity strategies for digital products across digital and physical markets
WO2003027926A1 (fr) Systeme de prediction du nombre de biens vendus
JP2003006518A (ja) 限られた数量の材の配分方法、広告配信方法、および、配分プログラム
US9865004B1 (en) System and method for advertisement impression volume estimation
US20120053999A1 (en) Delivering Highly Targeted Advertisements Based on a Coupon-Exchanging System
US11295337B1 (en) Transaction-based promotion campaign
JP7416837B2 (ja) 情報処理装置、情報処理方法及び情報処理プログラム
JP7414861B2 (ja) 情報処理装置、情報処理方法及び情報処理プログラム

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23885541

Country of ref document: EP

Kind code of ref document: A1