Prediction devices, prediction methods, and prediction programs.

Abstract

Invention details;

Description

Prediction device, prediction method, and prediction program

[0001] The present disclosure relates to a prediction device, a prediction method, and a prediction program.

[0002] The applicant has filed a patent application and obtained a patent right for a prediction device that accurately predicts broadcast contact information according to conditions related to broadcast information (Patent Document 1).

[0003] The prediction device of Patent Document 1 includes a storage unit, a calculation unit, and a prediction unit. The storage unit stores broadcast contact information indicating the degree of contact from receivers with each of a plurality of broadcast information items broadcast on a broadcast medium. The calculation unit calculates at least two or more statistical values ​​from a first statistical value, a second statistical value, and a third statistical value. The prediction unit calculates predicted contact information indicating the degree of contact from receivers with broadcast information scheduled to be broadcast based on the at least two or more statistical values. The first statistical value represents a moving average of broadcast contact information for each of a plurality of broadcast information items broadcast on a broadcast medium on a plurality of days immediately preceding a scheduled broadcast date on which the scheduled broadcast information is to be broadcast. The second statistical value represents a moving average of broadcast contact information for each of a plurality of broadcast information items broadcast on a broadcast medium on a plurality of days immediately preceding the scheduled broadcast date, which correspond to the same day of the week as the scheduled broadcast date. The third statistical value represents a moving average of broadcast contact information for each of a plurality of broadcast information items broadcast on a broadcast medium in the same time slot on a plurality of days immediately preceding the scheduled broadcast date, which correspond to the same day of the week as the scheduled broadcast date.

[0004] Patent No. 6902636

[0005] However, the conventional techniques including Patent Document 1 have room for improvement in terms of predicting audience ratings with high accuracy.

[0006] Therefore, one of the objects of the present disclosure is to provide a prediction device, a prediction method, and a prediction program that can predict audience ratings with high accuracy.

[0007] A prediction device according to one aspect of the present disclosure is characterized by having a memory unit that stores reference features of the audience ratings of past programs, a setting unit that sets a request slot for at least one of target programs, an extraction unit that extracts features of the request slot, and an acquisition unit that acquires predicted features related to the prediction of the audience rating of the request slot based on the reference features and the features of the request slot.

[0008] According to one aspect of the present disclosure, it is possible to provide a prediction device, a prediction method, and a prediction program that are capable of predicting audience ratings with high accuracy.

[0009] FIG. 1 is a diagram showing an example of a prediction calculation method in conventional long-term trend prediction. FIG. 2 is a diagram showing an example of the relationship between explanatory variables, past audience ratings, and a target period in conventional long-term trend prediction. FIG. 3 is a diagram showing a first example of a process for storing reference features of audience ratings of past programs. FIG. 4 is a diagram showing a second example of a process for storing reference features of audience ratings of past programs. FIG. 5 is a diagram showing an example of acquisition of predicted features of a request slot based on reference features and features of the request slot. FIG. 6 is a diagram showing an example of a schematic configuration of a system including a prediction device. A functional block diagram showing an example of the internal configuration of a prediction device. A diagram showing an example of the functional configuration of a cost calculation device, a broadcast distribution device, and a prediction device. A diagram showing an example of the hardware configuration of a cost calculation device, a broadcast distribution device, and a prediction device.

[0010] As a mechanism (model) for predicting the audience rating of a target program, for example, there is a recent trend prediction (SHAREST_RT, Recent Trend Prediction Version) and a long-term trend prediction (SHAREST_LT, Long-term Trend Prediction Version).

[0011] Recent trend prediction is a process for predicting the audience ratings of a target program whose scheduled broadcast date is approaching, and is a prediction process that predicts audience ratings with high accuracy based on detailed data such as program lineup, genre, online ratings, etc. The predicted target date in recent trend prediction (the broadcast date of the target program) may be, for example, one to seven days from now (one to seven days before broadcast).

[0012] Long-term trend prediction is a prediction of the audience ratings of a target program whose scheduled broadcast date is far away, and is a prediction process for mid- to long-term prediction that takes into account, for example, trends in audience ratings over time and seasonality (seasonal changes). The predicted target date in long-term trend prediction (the broadcast date of the target program) may be, for example, 8 to 70 days from now (8 to 70 days before broadcast).

[0013] In the long-term trend prediction, the audience rating of a target program is predicted based on the real-time audience rating (average audience rating) every 10 minutes from 8 to 70 days from now (70 days to 8 days before broadcast). In this case, a predetermined period, for example, the past two years, learned weekly (starting at 7:00 PM every Wednesday), may be used as learning data. Furthermore, for example, a daily prediction (starting at 3:00 AM every day) may be used as the prediction target period. In this case, the prediction target period may be up to 70 days before the latest flash data. The prediction target period may be extendable.

[0014] FIG. 1 is a diagram illustrating an example of a conventional prediction calculation method for long-term trend prediction. The model in FIG. 1 utilizes a recurrent neural network (RNN) to perform daily audience rating predictions for all targets (e.g., 31 layers) and all stations. In the example of FIG. 1, when it is desired to predict the audience rating for July 21st using actual measured values ​​up to July 20th of a certain year, the actual measured values ​​are learned in chronological order, and a predicted value is output. Specifically, in the 5:00-5:10 time slot, a predicted value of 0.3 for July 21st is output based on the actual measured values ​​up to July 20th (the actual measured values ​​shown in the figure are from July 18th to July 20th). Similarly, in the 5:10-5:20 time slot, a predicted value of 0.2 for July 21st is output based on the actual measured values ​​up to July 20th (the actual measured values ​​shown in the figure are from July 18th to July 20th). Furthermore, in the time slot of 5:20-5:30, a predicted value of 0.6 for July 21 is output based on the actual measured values ​​up to July 20 (the actual measured values ​​shown are from July 18 to July 20). Furthermore, in the time slot of 5:30-5:40, a predicted value of 0.7 for July 21 is output based on the actual measured values ​​up to July 20 (the actual measured values ​​shown are from July 18 to July 20). Furthermore, in the time slot of 28:50-29:00, a predicted value of 0.1 for July 21 is output based on the actual measured values ​​up to July 20 (the actual measured values ​​shown are from July 18 to July 20).

[0015] As such, conventional long-term trend predictions are models that learn and predict the average viewership ratings for every 10 minutes in the past, which means that there is a tendency (technical issue) for the accuracy of viewership rating predictions to be poor (there is a discrepancy between the predicted viewership ratings and the actual viewership ratings).

[0016] FIG. 2 is a diagram illustrating an example of the relationship between explanatory variables, past audience ratings, and the target period in conventional long-term trend prediction. In the example of FIG. 2, February 18th of a certain year is set as the prediction reference date, and the past average audience rating up to n days prior to that date is calculated and defined as the explanatory variable. Here, the n-day-prior average is calculated from the day before the prediction reference date, where n=28, that is, the average for the previous four weeks (average for each day regardless of the day of the week). Specifically, the past average audience rating from January 21st to February 17th of a certain year is calculated and defined as the explanatory variable, and used to predict the audience rating on February 18th. In this way, data for the day before the prediction reference date based on daily linkage exists, and by utilizing the most recent data possible, the past average audience rating is obtained from the latest linked data as of the prediction reference date. Furthermore, to take recent trends into account, the past audience rating is calculated starting from the day before the prediction reference date. The past audience rating is used as a variable for performing highly accurate audience rating predictions on the prediction reference date. Furthermore, if the prediction target is February 19th to April 29th of a certain year, the predicted value for April 29th, which is 70 days from the prediction reference date of February 18th, may use the average viewership rating for the past n days, calculated from February 17th, as an explanatory variable.

[0017] In conventional long-term trend forecasting, for example, even if some of the 60-day-ahead forecast values ​​are incorrect and corrected to the correct values, the forecast accuracy may still be inferior to the average for the previous eight weeks. Furthermore, when comparing the accuracy of each program broadcast length, there are cases where the forecast accuracy is worse than the average for the previous eight weeks depending on the broadcast time (e.g., for short programs of approximately six minutes). In particular, there is a tendency to predict higher audience ratings than actual for short programs included in the A-time slots, such as those from 7:00 PM to 11:00 PM on weekdays and from 6:00 PM to 11:00 PM on weekends, which are considered to have the highest audience ratings. This is presumably due to the fact that conventional long-term trend forecasting is a model that learns and predicts average audience ratings every 10 minutes in the past. For the same reason, the forecast accuracy of audience ratings for 56-minute programs adjacent to (affected by) short programs inserted in the A-time slot tends to deteriorate. Thus, even if there is an overall upward trend, the accuracy of audience rating forecasts using long-term trend forecasting may not be able to keep up with actual growth. In other words, even if a trend is captured, it may not be possible to predict sudden growth that has rarely occurred in the past.

[0018] The inventors have recognized the above-mentioned problems as important technical challenges and have come up with the idea of ​​a prediction device, a prediction method, and a prediction program having the following components, in order to perform highly accurate predictions of audience ratings, particularly in long-term trend predictions.

[0019] To improve the recurrent neural network (RNN), the model structure is changed to one that makes it easier to learn time-series changes by day of the week, and the objective variable is changed to one that makes it easier to learn time-series trends. Furthermore, a request slot, which is a program slot that is likely to affect audience rating predictions, is set, and predicted values ​​are corrected according to the request slot. Specifically, a mini-program slot that is likely to worsen audience rating predictions (easily overestimated) is set as the request slot, and a model is constructed (newly added) that converts and outputs predicted values ​​in long-term trend predictions based on the features of the request slot.

[0020] <Storage Process (Storage Step)> The reference feature amount of the audience ratings of past programs is stored. The reference feature amount may be the audience ratings of past programs themselves, or may be some parameter related to the audience ratings of past programs.

[0021] 3 is a diagram showing a first example of a process (storage step) for storing reference features of audience ratings of past programs. In the model shown in FIG. 3, a recurrent neural network (RNN) is used to predict audience ratings daily for all targets (e.g., 31 layers) x all stations.

[0022] As shown in FIG. 3 , actual measured values ​​(measured audience ratings, measured feature quantities) for a certain broadcasting station (● broadcasting) for 10-minute time slots such as 5:00-5:10, 5:10-5:20, and 5:20-5:30 on Mondays are stored for 20 weeks to 1 week before the prediction reference date (prediction target date). Also, actual measured values ​​(measured audience ratings, measured feature quantities) for a certain broadcasting station (● broadcasting) for 10-minute time slots such as 5:00-5:10, 5:10-5:20, and 5:20-5:30 on Tuesdays are stored for 20 weeks to 1 week before the prediction reference date. Similarly, actual measured values ​​(measured audience ratings, measured feature quantities) for a certain broadcasting station (● broadcasting) for 10-minute time slots such as Wednesday-Sunday for 20 weeks to 1 week before the prediction reference date are stored. Since the prediction reference date falls on any day of the week between Monday and Sunday, the feature values ​​of the audience ratings for a predetermined period spanning multiple past programs are stored, and the time-series audience ratings for the day of the week that corresponds to the prediction reference date are stored as the reference feature values. In other words, when storing the reference feature values, a structure is modified to make it easier to learn time-series changes for each day of the week. Furthermore, as will be described in detail later, by applying correction to the reference feature values ​​based on the feature values ​​of the request slots set for at least one of the target programs, a predicted feature value related to the predicted audience rating for the request slot can be obtained. This predicted feature value may be the predicted audience rating for the request slot itself, or may be some parameter related to the predicted audience rating for the request slot. Note that, in FIG. 3 , from the second week onward, the predicted feature value may be added to the feature value for parameter correction to perform recursive prediction (updating the predicted audience rating).

[0023] 4 is a diagram showing a second example of the storage process (storage step) of the reference feature amount of the audience ratings of past programs. In the model of FIG. 4, a recurrent neural network (RNN) is used to perform daily audience rating predictions for all targets (e.g., 31 layers) x all stations.

[0024] As shown in FIG. 4 , actual measurement values ​​(measured audience ratings, measured feature quantities) for a certain broadcasting station (● broadcasting) for 10-minute time slots such as 5:00-5:10, 5:10-5:20, and 5:20-5:30 on Mondays are stored for the period from 20 weeks prior to the prediction reference date (prediction target date) to 1 week prior to the prediction reference date (prediction target date). Similarly, actual measurement values ​​(measured audience ratings, measured feature quantities) for a certain broadcasting station (● broadcasting) for Tuesdays through Sundays are stored for the period from 20 weeks prior to the prediction reference date (prediction target date) in 10-minute time slots. These actual measurement values ​​can be the difference between the audience rating for the current week including the prediction reference date and the 24-week exponential moving average audience rating for the prediction reference date. Note that the standard value for calculating the difference between the actual measurement values ​​is not limited to the 24-week exponential moving average audience rating, but may also be a 4-week exponential moving average audience rating, an 8-week exponential moving average audience rating, a 12-week exponential moving average audience rating, or a 48-week exponential moving average audience rating. In this way, the feature values ​​of the audience rating for a predetermined period spanning multiple past programs are stored, and the difference between the audience rating for the current week including the prediction reference date and the exponential moving average audience rating for a predetermined period prior to the prediction reference date is stored as the reference feature value. In other words, the objective variable is changed to one that is easy to learn time-series trends. Furthermore, as will be described in detail later, by applying a correction to the reference feature value based on the feature values ​​of the request slot set for at least one target program, a predicted feature value related to the predicted audience rating for the request slot can be obtained. This predicted feature value may be the predicted audience rating for the request slot itself, or may be some parameter related to the predicted audience rating for the request slot. Note that in FIG. 4 , from the second week onward, the predicted feature value may be added to the feature value for parameter correction to perform recursive prediction (updating the predicted audience rating).

[0025] Here, the past audience rating, which is an example of a reference feature of the audience rating of a past program, includes, for example, the average audience rating for a program a predetermined period ago, the audience rating at the end of a predetermined period ago, the average audience rating for the same day last year, and the average audience rating for the same time slot last year. The past audience rating further includes the most recent moving average audience rating, the moving average audience rating for the same day of the week, missing flag 1, the moving average audience rating for the same day of the week and the same time slot, and missing flag 2. The most recent moving average audience rating, the moving average audience rating for the same day of the week, and the moving average audience rating for the same day of the week and the same time slot are examples of statistical values ​​of broadcast contact information.

[0026] The average audience rating for a program a specified period before represents the average audience rating for each program in the same time slot, the backslot, the previous slot, and the previous backslot that were broadcast a specified period before the prediction target date. The audience rating at the end of the specified period before represents the audience rating at the end of a program in the same time slot that was broadcast a specified period before the prediction target date. The average audience rating for the same day last year represents the average audience rating on the same day last year. The average audience rating for the same time slot last year represents the average audience rating for the same time slot last year.

[0027] The most recent moving average audience rating represents the moving average of the average audience rating of each of the most recent N broadcasts (N is an integer of 2 or greater) of a program on the target prediction date. The N broadcasts are broadcasts of programs in the program's own time slot, and are extracted according to the broadcast interval for each program, such as daily programs, weekly programs, yearly programs, or irregular programs. The most recent N broadcasts on the target prediction date are sometimes called the previous N broadcasts. The average audience rating of each program on the previous N broadcasts may be calculated, and the most recent moving average audience rating may be calculated using the average audience ratings of the program for the N broadcasts.

[0028] The moving average audience rating for the same day of the week represents the moving average audience rating for the previous N times corresponding to the same day of the week as the prediction target date. Missing flag 1 is a flag indicating whether or not a program was broadcast on the same day of the week as the prediction target date. The moving average audience rating for the same day of the week and in the same time slot represents the moving average audience rating for the previous N times corresponding to the same day of the week and in the same time slot as the prediction target date. Missing flag 2 is a flag indicating whether or not a program was broadcast on the same day of the week and in the same time slot as the prediction target date.

[0029] The moving average may be, for example, a simple moving average or a weighted moving average. The weighted moving average may be, for example, a linear weighted moving average or an exponential weighted moving average. The weight of the weighted moving average is set to be larger the later the broadcast date of the broadcast. The moving average audience rating calculated using the weighted moving average is an example of a statistical value in which the contribution of broadcast contact information of the second broadcast information, which was broadcast after the first broadcast information, is greater than the contribution of broadcast contact information of the first broadcast information.

[0030] By using a weighted moving average, the contribution of broadcasts that air later to the moving average audience rating is greater than the contribution of broadcasts that air earlier to the moving average audience rating, so it is possible to calculate a moving average audience rating that largely reflects recent audience trends. This makes it easier for recent audience trends to be reflected in the predicted audience rating, improving prediction accuracy. Multiple moving average audience ratings with different values ​​of N, such as the moving average audience rating for the previous four times, the moving average audience rating for the previous eight times, and the moving average audience rating for the previous ten times, may be used in combination.

[0031] The previous N broadcasts that are the subject of the moving average do not necessarily have to be the same program as the program to be predicted, but may be various programs that belong to the same program category as the program to be predicted. Furthermore, the program description information of each program included in the previous N broadcasts may include a predetermined keyword or its synonym that is included in the program description information of the program to be predicted. Synonyms of the predetermined keyword can be identified using a thesaurus.

[0032] <Setting Process (Setting Step)> A request slot is set for at least one of the target programs for which audience ratings are predicted. While there is a degree of flexibility in how the request slot is set, and various design modifications are possible, as an example, it is preferable to set a relatively short sub-program (mini-program) adjacent to a relatively long main program among the target programs as the request slot. This is because, as mentioned above, when focusing on accuracy comparisons for each program broadcast length, depending on the broadcast time, prediction accuracy for a mini-program of, for example, approximately 6 minutes may be poor. In particular, there is a tendency for the audience ratings of mini-programs included in A time slots, such as 7:00 PM to 11:00 PM on weekdays and 6:00 PM to 11:00 PM on weekends, which are considered to have the highest audience ratings, to be predicted higher than they actually are. Alternatively, a 56-minute program adjacent to (affected by) a mini-program, which also tends to have poor audience rating prediction accuracy, may be set as the request slot.

[0033] <Extraction Process (Extraction Step)> The feature quantities of the set request slot, for example, a mini program slot, are extracted. Examples of the feature quantities of the request slot include the following (1) to (9). (1) Broadcast minutes of the request slot (2) Day of the week of the request slot (starting day, ending day) (3) Time period of the request slot (starting hour, ending hour, starting minute, ending minute) (4) Broadcast station of the request slot (5) Saturday / Sunday flag of the request slot (starting time, ending time) (6) Holiday flag of the request slot (starting time, ending time) (7) Weekday holiday flag of the request slot (starting time, ending time) (8) Time spanning flag of the request slot (whether the starting hour and ending hour are the same) (9) Day spanning flag of the request slot (whether it spans 29 hours)

[0034] The purpose of adopting (1) above as a feature (variable) of the request slot is that in the case of mini-slot programs with short broadcast minutes, the actual observed audience rating is likely to be lower than the base 10-minute average predicted audience rating, and this tendency can be learned (reflected) in the audience rating prediction.

[0035] The purpose of adopting (2) to (4) above as feature quantities (variables) of the request slot is to learn (reflect) the difference in the tendency of the actual audience rating to rise, fall, or remain unchanged compared to the base 10-minute average predicted audience rating depending on the specific broadcasting station, day of the week, and time slot, in the audience rating prediction.

[0036] The purpose of adopting (5) to (7) above as feature quantities (variables) for the request slot is that viewers’ behavior is more likely to change on weekends, national holidays, and weekday national holidays than on weekdays, and to learn (reflect) such trends in the viewer rating prediction.

[0037] The purpose of adopting (8) above as a feature (variable) of the request slot is that viewers tend to switch channels at time intervals (● o'clock), and the difference between the base 10-minute average predicted audience rating and the actual audience rating often differs between a program that starts exactly at 0 minutes and a program that starts a few minutes before that, and this tendency can be learned (reflected) in the audience rating prediction.

[0038] The purpose of adopting (9) above as a feature (variable) of the request slot is to learn (reflect) the difference in viewership ratings across time periods between early morning programs (especially those spanning from 4:00 to 5:00) and regular programs, and to reflect this difference in viewership rating predictions.

[0039] In light of the above, when extracting feature quantities of a set request slot, such as a mini program slot, it is preferable to extract the broadcast duration of the request slot, a weekday holiday flag, a time-spanning flag, and a day-spanning flag. These feature parameters are extremely useful for eliminating the drawbacks caused by conventional long-term trend predictions being models that learn and predict average audience ratings every 10 minutes in the past, and for performing highly accurate audience rating predictions. Alternatively, when extracting feature quantities of a set request slot, such as a mini program slot, it is also possible to extract the broadcast duration of the request slot, the day of the week, the time slot, the broadcast station, the weekend flag, the holiday flag, the weekday holiday flag, the time-spanning flag, and the day-spanning flag. Even in this case, it is possible to eliminate the drawbacks caused by conventional long-term trend predictions being models that learn and predict average audience ratings every 10 minutes in the past, and perform highly accurate audience rating predictions.

[0040] <Acquisition process (acquisition step)> A predicted feature quantity related to the predicted audience rating of the request slot is acquired based on the stored reference feature quantity and the extracted feature quantity of the request slot. In other words, a predicted feature quantity related to the predicted audience rating of the request slot is acquired by correcting the stored reference feature quantity based on the extracted feature quantity of the request slot. Here, the predicted feature quantity may be the predicted audience rating of the request slot itself, or may be some parameter related to the predicted audience rating of the request slot.

[0041] FIG. 5 illustrates an example of acquiring predicted features for a request slot based on reference features and features for a request slot. As shown in FIG. 5 , reference features for the audience ratings of past programs based on the 10-minute average predicted audience rating and features for the request slot set for the target program are input. Since program slot information is not available at the time of prediction, the reference features are calculated based on the 10-minute average predicted audience rating without using program slot information. Furthermore, a predicted value that takes into account trends is calculated using the time-series audience rating for the day of the week that is the same as the prediction reference date, or by adding or subtracting a 24-week moving average difference (recent trend). As described above, the features for the request slot may include at least one of the broadcast minutes of the request slot, the day of the week, the time slot, the broadcast station, a Saturday / Sunday flag, a holiday flag, a weekday holiday flag, a time-spanning flag, and a day-spanning flag. Then, predicted features related to the predicted audience rating for the request slot are acquired based on the input reference features and features for the request slot (the predicted features are acquired by correcting the reference features based on the features for the request slot). In this way, a model (request slot predicted value conversion model) is added that corrects predicted features based on the features of the request slot (broadcast time, etc.). For example, by reading features such as time slots and time spans from the program information and applying corrections to programs with short durations such as mini program slots, it is possible to improve the accuracy of audience rating predictions. More specifically, taking into account that the predicted audience ratings for mini program slots tend to be overestimated, the predicted audience ratings for mini program slots can be corrected downward to bring them closer to the actual measured value.

[0042] 6 is a diagram showing an example of a schematic configuration of a system including a prediction device (viewership prediction device). The system 1 includes a cost calculation device 10, a broadcast distribution device 20, and a prediction device (viewership prediction device) 30. The cost calculation device 10, the broadcast distribution device 20, and the prediction device 30 are connected to each other via a network 40 so as to be able to communicate with each other.

[0043] The cost calculation device 10 calculates the cost of, for example, a television commercial based on a calculation formula that includes the audience rating (predicted audience rating) of a target program into which the television commercial is inserted.

[0044] The cost calculation device 10 acquires, for example, the cost for each advertising pattern of spot commercials for a certain broadcasting station (which may be replaced with the channel) over a predetermined period (e.g., one week). Here, the cost in this disclosure may refer to cost per rating, which indicates the fee per GRP (1%). The advertising pattern may include existing (arbitrary) advertising patterns (e.g., inverted L-shaped, all-day, square-shaped, square-shaped, square-shaped, late-night, all-day daytime, etc.). Furthermore, the cost calculation device 10 may acquire the cost for each advertising pattern, the time periods that make up the advertising pattern, etc., based on external input (e.g., settings by the broadcasting station).

[0045] The cost calculation device 10 may acquire (e.g., receive) information for cost calculation. For example, the cost calculation device 10 may acquire (receive) information on viewership ratings for each advertising slot (e.g., a household viewership rating, an individual viewership rating, and an overall individual viewership rating that is a combined viewership rating of real-time viewing and time-shifted viewing up to seven days after broadcast (which may be called "All & P + C7 (Program + Commercial 7)")) from the prediction device 30. The cost calculation device 10 may derive the cost of one A-time slot, for example, based on the unit price set by the broadcast station for one A-time slot and the viewership rating information acquired from the prediction device 30.

[0046] The cost calculation device 10 may be a mobile terminal (mobile communication terminal) such as a mobile phone, smartphone, or tablet terminal, or may be a fixed communication terminal such as a personal computer (PC) or server. In other words, the cost calculation device 10 in the present disclosure can be interpreted as a communication device. The cost calculation device 10 may also communicate with a network (such as the Internet) via wired and / or wireless communication (e.g., Long Term Evolution (LTE), New Radio (NR), Wi-Fi (registered trademark), etc.).

[0047] The broadcast distribution device 20 is a device that distributes broadcasts to televisions (TVs). Here, the television (television receiver) may be a device that has the function of receiving at least one of terrestrial broadcasts, broadcasts via broadcasting satellites (BS) / communications satellites (CS), internet broadcasts (internet television), etc. For example, the television may be a multi-function television, a smart TV, an IP (Internet Protocol) TV, a set-top box, etc. Note that broadcasts refer to the transmission of telecommunications intended for direct reception by the public, and may include wireless broadcasts, internet broadcasts, etc.

[0048] The prediction device 30 is a device that tallyes actual audience ratings of programs broadcast on television. The prediction device 30 is also a device that predicts audience ratings of programs (target programs) that will be broadcast on television in the future. The audience ratings may be stored in predetermined time units (e.g., seconds, minutes, hours). The prediction device 30 may transmit information related to audience ratings (e.g., the audience rating (predicted audience rating) of a target program in which a television commercial is inserted) to another device (e.g., the cost calculation device 10) via a network.

[0049] The prediction device 30 may be, for example, a device managed by a television audience rating researcher. The prediction device 30 may acquire and manage viewing history data from a measuring device connected to the television of a predetermined household (e.g., a household cooperating in the survey). Viewership ratings may be calculated or predicted from the data. The prediction device 30 may record audience ratings (individual audience ratings of the target) for each predetermined target (users corresponding to predetermined attributes). In the present disclosure, the targets are assumed to be gender and age categories (C (Child) demographic, T (Teen) demographic, M (Male)1-M3 demographic, F (Female)1-F3 demographic) used in the field of marketing, but are not limited to these categories.

[0050] The prediction device 30 is configured to predict television "viewer ad ratings" with high accuracy using a new analytical method that handles big data. The prediction device 30 may include a data input unit that inputs data source information, an explanatory variable conversion unit that applies specific processing to the data source information input from the data input unit to convert the data source information into explanatory variables, a program implementation / execution unit that implements a program of a predetermined algorithm, and a data output unit that outputs the results of the program implementation / execution unit. The program implementation / execution unit executes the algorithm of the program using the data source information input from the data input unit and the explanatory variables obtained by converting the data source information into explanatory variables by the explanatory variable conversion unit, and outputs predicted audience rating data obtained by executing the algorithm from the output unit. Advertising slots can be optimized based on the output predicted audience rating data. Here, audience ratings of related programs related to the target program (e.g., previous broadcasts of a drama series, previous broadcasts of a special program, etc.) can be used as data source information for calculating the audience rating (predicted audience rating) of a target program in which a television commercial is inserted.

[0051] Note that this system configuration is an example and is not limited to this. For example, while FIG. 6 shows a configuration including one of each device, the number of each device is not limited to this and multiple devices may be present. The system 1 may be configured not to include some devices, or may be configured such that the functions of one device are realized by multiple devices. Conversely, the system 1 may be configured such that the functions of multiple devices are realized by one device. For example, at least two of the cost calculation device 10, the broadcast distribution device 20, and the prediction device 30 may be implemented on a single server.

[0052] 7 is a functional block diagram showing an example of the internal configuration of the prediction device 30. The prediction device 30 includes a storage unit 31, a setting unit 32, an extraction unit 33, and an acquisition unit 34. The prediction method and prediction program of the present disclosure are realized by causing a computer included in the prediction device 30 (the storage unit 31, the setting unit 32, the extraction unit 33, and the acquisition unit 34) to execute predetermined processing steps.

[0053] The storage unit 31 stores reference feature quantities of audience ratings for past programs. The storage unit 31 stores feature quantities of audience ratings for a predetermined period spanning a plurality of past programs, and stores the time-series audience rating for the same day of the week as the prediction reference date as the reference feature quantity. The storage unit 31 stores feature quantities of audience ratings for a predetermined period spanning a plurality of past programs, and stores the difference between the audience rating for the current week including the prediction reference date and the exponential moving average audience rating for a predetermined period before the prediction reference date as the reference feature quantity.

[0054] The setting unit 32 sets a request slot for at least one of the target programs. The setting unit 32 sets a sub-program (mini-program) having a relatively short broadcast time adjacent to a main program having a relatively long broadcast time among the target programs as the request slot.

[0055] The extraction unit 33 extracts feature quantities of the request slot. The extraction unit 33 extracts, as feature quantities of the request slot, the broadcast minutes of the request slot, a weekday holiday flag, a time-spanning flag, and a day-spanning flag. The extraction unit 33 extracts, as feature quantities of the request slot, the broadcast minutes of the request slot, the day of the week, the time slot, the broadcast station, the Saturday / Sunday flag, the holiday flag, the weekday holiday flag, the time-spanning flag, and the day-spanning flag.

[0056] The acquisition unit 34 acquires a predicted feature quantity related to the prediction of the audience rating of the request slot based on the reference feature quantity and the feature quantity of the request slot. The acquisition unit 34 acquires a predicted feature quantity related to the prediction of the audience rating of the request slot by correcting the reference feature quantity based on the feature quantity of the request slot.

[0057] Note that the present disclosure also covers content in which "viewership rate" is replaced with a specific index (or the degree of improvement of a specific index, offset, etc.). The specific index may be an index for determining advertising effectiveness, an intermediate index for improving the advertiser's key performance indicators (KPIs), etc., and may include, for example, awareness rate, reach, ROAS (Return On Advertising Spend), ROI (Return on Investment), etc. At least a part of the term "viewership rate" may be replaced with any of the above specific indexes.

[0058] FIG. 8 is a diagram showing an example of the functional configuration of the cost calculation device 10, the broadcast distribution device 20, and the prediction device (viewership prediction device) 30. As shown in FIG. 8, the cost calculation device 10, the broadcast distribution device 20, and the prediction device 30 each include a control unit 110, a storage unit 120, a communication unit 130, an input unit 140, and an output unit 150. Note that FIG. 8 shows an example of functional blocks of the cost calculation device 10, the broadcast distribution device 20, and the prediction device 30, and they may also include other functional blocks not shown in FIG. 8. Furthermore, they may be configured to not include some of the functional blocks.

[0059] The control unit 110 controls the cost calculation device 10, the broadcast distribution device 20, and the prediction device 30. The control unit 110 can be configured by a controller, a control circuit, or a control device that is described based on common understanding in the technical field related to the present invention.

[0060] The storage unit 120 stores information used by the cost calculation device 10, the broadcast distribution device 20, and the prediction device 30. The storage unit 120 can be configured, for example, by a memory, storage, storage device, or the like that is described based on common understanding in the technical field related to the present invention.

[0061] The communication unit 130 performs communication (e.g., mutual communication) between the cost calculation device 10, the broadcast distribution device 20, and the prediction device 30. The communication unit 130 can be configured by a transmitter / receiver, a transmission / reception circuit, or a transmission / reception device that is described based on common understanding in the technical field related to the present invention. Note that the communication unit 130 may be configured by a transmission unit and a reception unit.

[0062] The input unit 140 accepts inputs to the cost calculation device 10, the broadcast distribution device 20, and the prediction device 30. The input unit 140 may also be connected to a predetermined device, storage medium, etc., and accept data input. The input unit 140 may output the input results to, for example, the control unit 110. The input unit 140 may be configured with input devices such as a keyboard, a mouse, and buttons, input / output terminals, input / output circuits, etc., which are described based on common understanding in the technical field related to the present invention. The input unit 140 may also be configured as an integrated unit with a display unit (for example, a touch panel).

[0063] The output unit 150 performs output in the cost calculation device 10, the broadcast distribution device 20, and the prediction device 30. For example, the output unit 150 may be configured to include a display unit that displays images, an audio output unit that outputs audio, and the like. The display unit may be configured, for example, by a display device such as a monitor or the like that is described based on common understanding in the technical field related to the present invention. Furthermore, the audio output unit may be configured by an output device such as a speaker that is described based on common understanding in the technical field related to the present invention.

[0064] The block diagrams used to explain the above embodiments show functional blocks. These functional blocks (components) are realized by any combination of hardware and / or software. Furthermore, the means for realizing each functional block is not particularly limited. That is, each functional block may be realized by a single physically coupled device, or may be realized by two or more physically separate devices connected by wire or wirelessly.

[0065] For example, the devices (cost calculation device 10, broadcast distribution device 20, prediction device 30) according to an embodiment of the present disclosure may function as a computer that performs the prediction process (viewership prediction process) of the present disclosure. FIG. 9 is a diagram illustrating an example of the hardware configuration of the cost calculation device 10, broadcast distribution device 20, and prediction device 30. The above-described cost calculation device 10, broadcast distribution device 20, and prediction device 30 may be physically configured as a computer device including a processor 1001, a memory 1002, a storage 1003, a communication device 1004, an input device 1005, an output device 1006, a bus 1007, etc.

[0066] In this disclosure, the terms apparatus, circuit, device, unit, server, etc. may be interchangeable. The hardware configurations of the cost calculation apparatus 10, the broadcast distribution apparatus 20, and the prediction apparatus 30 may be configured to include one or more of the apparatuses shown in the drawings, or may be configured to exclude some of the apparatuses.

[0067] For example, although only one processor 1001 is shown, there may be multiple processors. Furthermore, processing may be performed by one processor, or processing may be performed by two or more processors simultaneously, serially, or in other ways. Furthermore, processor 1001 may be implemented by one or more chips.

[0068] Each function of the cost calculation device 10, the broadcast distribution device 20, and the prediction device 30 is realized by loading predetermined software (programs) onto hardware such as a processor 1001 and a memory 1002, causing the processor 1001 to perform calculations and control communication via a communication device 1004, and reading and / or writing of data in the memory 1002 and storage 1003, etc.

[0069] The processor 1001 controls the entire computer by running, for example, an operating system. The processor 1001 may be configured as a central processing unit (CPU) including an interface with peripheral devices, a control device, an arithmetic unit, a register, etc. Note that each unit, such as the control unit 110 described above, may be realized by the processor 1001.

[0070] The processor 1001 also reads programs (program codes), software modules, data, etc. from at least one of the storage 1003 and the communication device 1004 into the memory 1002 and executes various processes in accordance with these. The programs used are those that cause a computer to execute at least some of the operations described in the above-described embodiments. For example, the control unit 110 may be implemented by a control program stored in the memory 1002 and executed by the processor 1001, and similar implementations may be made for other functional blocks.

[0071] The memory 1002 is a computer-readable recording medium and may be configured by at least one of, for example, a read-only memory (ROM), an erasable programmable read-only memory (EPROM), an electrically programmable programmable read-only memory (EEPROM), a random access memory (RAM), or other suitable storage medium. The memory 1002 may also be referred to as a register, a cache, a main memory, or the like. The memory 1002 may store executable programs (program codes), software modules, and the like for implementing a method according to an embodiment.

[0072] Storage 1003 is a computer-readable recording medium, and may be composed of at least one of, for example, a flexible disk, a floppy (registered trademark) disk, a magneto-optical disk (e.g., a compact disk (CD-ROM (Compact Disc ROM)), a digital versatile disk, a Blu-ray (registered trademark) disk), a removable disk, a hard disk drive, a smart card, a flash memory device (e.g., a card, a stick, a key drive), a magnetic stripe, a database, a server, or other suitable storage medium. Storage 1003 may also be called an auxiliary storage device. Note that the above-mentioned storage unit 120 may be realized by memory 1002 and / or storage 1003.

[0073] The communication device 1004 is hardware (transmission / reception device) for communicating between computers via at least one of a wired network and a wireless network, and is also referred to as a network device, a network controller, a network card, a communication module, etc. The communication device 1004 may include a SIM card. Note that the above-mentioned communication unit 130 may be realized by the communication device 1004.

[0074] The input device 1005 is an input device (e.g., a keyboard, a mouse, etc.) that accepts input from the outside. The output device 1006 is an output device (e.g., a display, a speaker, etc.) that performs output to the outside. The input device 1005 and the output device 1006 may be integrated into one unit (e.g., a touch panel). The above-mentioned input unit 140 and output unit 150 may be realized by the input device 1005 and the output device 1006, respectively.

[0075] Furthermore, each device such as the processor 1001 and the memory 1002 is connected by a bus 1008 for communicating information. The bus 1008 may be configured as a single bus or may be configured as different buses between the devices.

[0076] Furthermore, the cost calculation device 10, the broadcast distribution device 20, and the prediction device 30 may be configured to include hardware such as a microprocessor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a programmable logic device (PLD), or a field programmable gate array (FPGA), and some or all of the functional blocks may be realized by the hardware. For example, the processor 1001 may be implemented by at least one of these pieces of hardware.

[0077] (Modifications) Note that terms explained in this disclosure and / or terms necessary for understanding this disclosure may be replaced with terms having the same or similar meanings.

[0078] The information, parameters, etc. described in this disclosure may be expressed using absolute values, relative values ​​from a predetermined value, or other corresponding information. Furthermore, the names used for parameters, etc. in this disclosure are not limiting in any way.

[0079] The information, signals, etc. described in this disclosure may be represented using any of a variety of different technologies. For example, data, instructions, commands, information, signals, bits, symbols, chips, etc. that may be referred to throughout the above description may be represented by voltages, currents, electromagnetic waves, magnetic fields or magnetic particles, optical fields or photons, or any combination thereof.

[0080] Information, signals, etc. may be input / output via multiple network nodes. The input / output information, signals, etc. may be stored in a specific location (e.g., memory) or may be managed using a table. The input / output information, signals, etc. may be overwritten, updated, or added. Output information, signals, etc. may be deleted. Input information, signals, etc. may be transmitted to another device.

[0081] Furthermore, notification of specified information (e.g., notification that "it is X") is not limited to explicit notification, but may be made implicitly (e.g., by not notifying the specified information or by notifying other information).

[0082] Software shall be construed broadly to mean instructions, instruction sets, code, code segments, program code, programs, subprograms, software modules, applications, software applications, software packages, routines, subroutines, objects, executable files, threads of execution, procedures, functions, etc., whether referred to as software, firmware, middleware, microcode, hardware description language, or otherwise.

[0083] Software, instructions, information, etc. may also be transmitted and received via a transmission medium and / or signal waveform. For example, if software is transmitted from a website, server, or other remote source using wired technologies (such as coaxial cable, fiber optic cable, twisted pair, Digital Subscriber Line (DSL)), and / or wireless technologies (such as infrared, microwave), then these wired and / or wireless technologies are included within the definition of transmission media.

[0084] As used in this disclosure, the terms "system" and "network" may be used interchangeably.

[0085] Each aspect / embodiment described in this disclosure may be used alone, in combination, or switched depending on the implementation. Furthermore, the order of the processing procedures, sequences, flowcharts, etc. of each aspect / embodiment described in this disclosure may be changed unless inconsistent. For example, the methods described in this disclosure present elements of various steps using an example order, and are not limited to the particular order presented.

[0086] As used in this disclosure, the phrase "based on" does not mean "based only on," unless expressly stated otherwise. In other words, the phrase "based on" means both "based only on" and "based at least on."

[0087] As used in this disclosure, any reference to an element using a designation such as "first," "second," etc. does not generally limit the quantity or order of those elements. These designations may be used in this disclosure as a convenient method of distinguishing between two or more elements. Thus, a reference to a first and a second element does not imply that only two elements may be employed or that the first element must in some way precede the second element.

[0088] When the terms "include," "including," and variations thereof are used in this disclosure, these terms are intended to be inclusive, similar to the term "comprising." Furthermore, when the term "or" is used in this disclosure, it is not intended to be an exclusive or.

[0089] In this disclosure, where articles are added by translation, such as a, an, and the in English, the disclosure may include that the nouns following these articles are in the plural form.

[0090] Although the invention according to the present disclosure has been described in detail above, it is clear to those skilled in the art that the invention according to the present disclosure is not limited to the embodiments described in the present disclosure. The invention according to the present disclosure can be implemented in modified and altered forms without departing from the spirit and scope of the invention as defined by the description of the claims. Therefore, the description of the present disclosure is intended to be illustrative and does not impose any limiting meaning on the invention according to the present disclosure.

[0091] This application is based on Japanese Patent Application No. 2022-112232, filed on July 13, 2022, the contents of which are incorporated herein in their entirety.

Claims

DEPCT681. The predictive device comprises: a storage unit that stores the reference feature quantity of the viewing rate of previous programs; a setting unit that sets the request frame to at least one of the target programs; a sorting unit that sorts the feature quantity of the request frame; and a supply unit that supplies the predictive feature quantity related to the prediction of the viewing rate of the request frame based on the reference feature quantity and the request frame feature quantity.

2. The predictive device under claim 1, where the sorting unit sorts the broadcast duration, weekday holiday indicator, time crossing indicator, and day crossing indicator of the request frame as the request frame feature quantity.

3. The predictive device under claim 1 or 2, where the sorting unit sorts the broadcast duration, day of the week, time zone, broadcasting station, Saturday and Sunday indicator, holiday indicator, weekday holiday indicator, time crossing indicator, and day crossing indicator of the request frame as the request frame feature quantity. 4.

5. A predictive device based on claim 1 or 2 where the storage unit stores the viewership feature quantity for a predetermined time period over the duration of more than one previous program and stores the time-series viewership of the same day of the week as the prediction reference date as the reference feature quantity.

6. A predictive device based on claim 1 or 2 where the storage unit stores the viewership feature quantity for a predetermined time period over the duration of more than one previous program and stores the difference between this week's viewership, including the prediction reference date, and the index moving average viewership for a predetermined time period before the prediction reference date as the reference feature quantity. 7.The prediction method comprises: the process of storing a reference feature quantity of the viewership rate of previous programs; the process of setting the request frame to at least one of the target programs; the process of separating the feature quantity of the request frame; and the process of providing the predictive feature quantity related to the prediction of the request frame viewership rate based on the reference feature quantity and the request frame feature quantity.

8. Prediction program for computer execution: the process of storing a reference feature quantity of the viewership rate of previous programs; the process of setting the request frame to at least one of the target programs; the process of separating the feature quantity of the request frame; and the process of providing the predictive feature quantity related to the prediction of the request frame viewership rate based on the reference feature quantity and the request frame feature quantity.

Images