JP7084887B2 - Advertising performance evaluation methods, devices and programs - Google Patents

Description

The present invention relates to an advertisement performance evaluation method, a device and a program, and more particularly to an advertisement performance evaluation method, a device and a program for evaluating an advertisement performance based on the cognitive capacity of an advertisement viewing user.

There is known a system that evaluates the results of advertisements delivered through television or radio according to the visual condition and the degree of reaction of the user viewing the advertisement, and determines the advertisement reward. Patent Document 1 is a technique for collecting user vital data and predicting future sales of the product or service targeted for the advertisement based on whether or not the user responds to the delivered advertisement. Is disclosed.

Patent No. 6434177

Kenki Ueda, et al. "Home life behavior recognition system based on user location information and home appliance power consumption." Journal of Information Processing Society 57.2 (2016): 416-425. Yacoob, Yaser, and Michael J. Black. "Parameterized modeling and recognition of activities." Computer Vision and Image Understanding 73.2 (1999): 232-247. Jianming Wu, Toshiyuki Hagiya, Yujin Tang, and Keiichiro Hoashi. 2017. Effects of objective feedback of facial expression recognition during video support chat. In Proc. MUM '17. 293-297.

In order to acquire the user's reaction based on the vital data, each user must be equipped with a dedicated sensor for acquiring the vital data. Therefore, it imposes a burden on the user, and it is difficult to easily acquire a large amount of information. In addition, it may not be possible to accurately evaluate advertising performance based solely on the user's perception and reaction.

FIG. 10 is a diagram schematically showing the relationship between the degree of perception and reaction of a user to an advertisement and the advertisement result for the user.

Sufficient advertising results can be expected if the user is recognized as having a clear reaction to the advertisement, while advertising results may be expected even if the user is not recognized as having a clear reaction. Further, if it is recognized that the user cannot perceive the advertisement, the advertisement result cannot be expected, but even if it is recognized that the user can perceive the advertisement, it cannot be judged that sufficient advertisement result can be expected by itself.

In this way, if the advertising results are evaluated only by the user's reaction and perception, it is possible to evaluate the situation where sufficient advertising results can be expected or the situation where the advertising results cannot be expected at all, but in the middle, that is, the situation where the user can perceive the advertisement. Could not quantitatively evaluate the advertising results, such as when there was no clear reaction.

An object of the present invention is to solve the above-mentioned technical problems, and even when it is difficult to evaluate the advertising performance based on the user's perception or reaction, the advertising performance evaluation can be quantitatively evaluated based on the evaluation value of the user's cognitive capacity. To provide methods, equipment and programs.

In order to achieve the above object, the present invention is characterized in that the following means are taken in the advertising performance evaluation method, device and program for evaluating the performance of the advertisement to the user.

(1) The advertising performance is evaluated based on the evaluation value of the user's cognitive capacity.

(2) The cognitive capacity can be evaluated based on the estimation result of the user's context.

(3) Prepare a context prediction model that estimates the context based on the output signal of the sensor that detects the user's situation, and apply the output signal of the sensor that detects the user's situation to the context prediction model to estimate the context. I made it.

(4) A perceptual evaluation means for evaluating the user's perception and obtaining the perceptual evaluation value was further provided, and the advertising result was evaluated based on the user's cognitive capacity evaluation value and the perceptual evaluation value.

(5) A reaction evaluation means for evaluating the user's reaction and obtaining the reaction evaluation value was further provided, and the advertising result was evaluated based on the user's cognitive capacity evaluation value and the reaction evaluation value.

(6) A perceptual evaluation means for evaluating the user's perception to obtain a perceptual evaluation value and a reaction evaluation means for evaluating the user's reaction to obtain a reaction evaluation value are further provided, and the user's cognitive capacity evaluation value, perceptual evaluation value and The advertising performance is evaluated based on the reaction evaluation value.

(7) The user's perception is evaluated based on the signal-to-noise ratio of the sound and the relative position between the advertisement source and the user.

(8) The user's reaction is evaluated based on the audio and camera image.

According to the present invention, the following results are achieved.

(1) Since the advertising performance is evaluated based on the evaluation value of the user's cognitive capacity, it becomes possible to quantitatively evaluate the advertising performance even when it is difficult to evaluate the advertising performance based on the user's perception and reaction.

(2) Since the cognitive capacity is calculated based on the context that is highly correlated with the cognitive capacity of the user, the cognitive capacity of the user can be evaluated accurately.

(3) Since the output signal of the sensor that detects the user's situation is applied to the context prediction model to estimate the user's context, the context can be estimated accurately based on the user's situation.

(4) Since the advertising performance is evaluated based on the user's cognitive capacity evaluation value and perceptual evaluation value, the advertising performance when the user can perceive the advertisement is quantitatively evaluated based on the cognitive capacity evaluation value. You will be able to evaluate it.

(5) Since the advertising performance is evaluated based on the user's cognitive capacity evaluation value and reaction evaluation value, the advertising performance when the user does not show a clear reaction is quantified based on the cognitive capacity evaluation value. You will be able to evaluate it.

(6) Since the advertising performance is evaluated based on the evaluation value, perceptual evaluation value, and reaction evaluation value of the user's cognitive capacity, the advertisement when the user can perceive the advertisement but does not show a clear reaction. Results can be evaluated quantitatively based on the evaluation value of cognitive capacity.

(7) Since the user's perception is evaluated based on the SN ratio of the sound and the relative position between the advertisement source and the user, it becomes possible to evaluate various perceptions of the user.

(8) Since the reaction of the user is evaluated based on the voice of the user and the image of the camera, various reactions of the user can be evaluated.

It is a functional block diagram of the advertisement performance evaluation apparatus which concerns on one Embodiment of this invention. It is a figure which showed schematically the evaluation method of the cognitive capacity. It is a functional block diagram of the perception evaluation part. It is a functional block diagram of a reaction evaluation part. It is a functional block diagram of the cognitive capacity evaluation unit. It is a figure which showed an example of the reaction table. It is a figure which showed an example of a context table. It is a flowchart which showed the operation of one Embodiment of this invention. It is a figure which showed an example of how to obtain the image-based perception evaluation value. It is a figure which showed the relationship between the user's perception and reaction to an advertisement, and the advertisement result.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a functional block diagram showing the configuration of a main part of the advertising performance evaluation device 1 according to the embodiment of the present invention together with its application environment.

The advertisement performance evaluation device 1 is based on the user status identification unit 10 that identifies the status of the viewing user U for the advertisement distributed by the advertisement distribution business facility 2 and transmitted through TV TV, radio, or the like, and the user status identification result. The main configuration is the advertisement result determination unit 11 that determines the advertisement result for the user based on the above, and various sensors such as a camera 12, a microphone 13, and a human sensor 14 are provided as a sensor group for detecting the user situation. .. It should be noted that the microphone function of the mobile terminal 15 such as a smartphone or tablet carried by the user can be used together with the microphone 13 or in place of the microphone 13.

The user situation identification unit 10 includes a perception evaluation unit 101, a reaction evaluation unit 102, and a cognitive capacity evaluation unit 103. The perception evaluation unit 101 evaluates the user's perception of the advertisement. The reaction evaluation unit 102 evaluates the reaction of the user to the advertisement. The cognitive capacity evaluation unit 103 evaluates the cognitive capacity of the user's advertisement.

Such an advertisement result evaluation device 1 can be configured by implementing an application (program) that realizes each function on a general-purpose computer or server. Alternatively, it can be configured as a dedicated machine or a single-purpose machine in which a part of the application is made into hardware or software.

FIG. 2 is a diagram schematically showing an evaluation method of cognitive capacity in the cognitive capacity evaluation unit 103. Each user has a predetermined cognitive capacity, but the cognitive capacity is consumed for each context such as "walking", "sleeping", and "reading". Therefore, in the present embodiment, the user's context is analyzed, and the cognitive capacity remaining after being consumed by the context is evaluated as the capacity that the user can devote to the recognition of the advertisement.

When the result of an advertisement broadcast from a television or radio to a user in the house is to be evaluated, the advertisement result evaluation device 1 is, for example, a speaker (smart speaker) having an AI assistant function or an interactive robot. It is implemented and operated in. Further, the sensor groups such as the camera 12, the microphone 13, and the motion sensor 14 are distributed and arranged in each living room, kitchen, corridor, toilet, bath, and the like.

The camera 12 transmits the camera image taken in the installation space as a sensor output to the advertisement result evaluation device 1. The microphone 13 transmits the acoustic signal detected in the installation space as a sensor output to the advertisement result evaluation device 1. The motion sensor 14 transmits the detection result of a person in the installation space as a sensor output to the advertisement result evaluation device 1. The output of each sensor is transmitted to the advertisement performance evaluation device 1 via a wireless or wired LAN or a mobile communication network.

As shown in FIG. 3, the perception evaluation unit 101 includes an SN ratio measuring unit 101a that measures the SN ratio of sound as an index value of perception, and a relative position measuring unit 101b that measures the relative position between the television and the user. Further, the evaluation value calculation unit 101c for calculating the perceptual evaluation value p (x) based on the SN ratio and the relative position is included.

The SN ratio measuring unit 101a calculates the ratio (S / N) of the audio signal (S) output by the television to the environmental noise (N). In the present embodiment, the S / N in the vicinity of the user can be measured by using the microphone function of the smartphone 15 carried by the user. Alternatively, the user's location may be estimated based on the camera image and the output of the motion sensor 14, and the S / N may be measured based on the audio signal of the microphone 13 installed at the estimated location. Alternatively, the S / N may be estimated based on the distance between the microphone and the user and the output from the speaker.

The relative position measuring unit 101b identifies the television and the user based on the video output output by the camera 12 and the output of the motion sensor 14, estimates the location of the television, and then calculates the distance between the television and the user.

As shown in FIG. 4, the reaction prediction model 102a and the reaction table 102b are registered in advance in the reaction evaluation unit 102. In this embodiment, "reply", "nod", "look at", "change facial expression", "interrupt work", etc. are registered in advance as candidates for the user's reaction form. By machine learning each sensor output for each reaction candidate or a combination thereof, a reaction prediction model 102a that estimates the user's reaction from each sensor output is constructed. As shown in FIG. 6, in the reaction table 102b, the correspondence between the user's reaction and the weight value is registered in advance.

The evaluation value calculation unit 102c estimates the user's reaction by applying the audio signal detected by the microphone 13 or the like and the user's behavior analyzed from the camera image to the reaction prediction model 102a, and further estimates the reaction estimation result in the reaction table. By applying to 102b, the weight value is obtained for each reaction. Then, the reaction evaluation value r (x) is calculated by multiplying all the weight values corresponding to each reaction. A specific method for estimating the reaction is disclosed in Non-Patent Document 1.

As shown in FIG. 5, the cognitive capacity prediction model 103a and the context table 103b are registered in advance in the cognitive capacity evaluation unit 103. In this embodiment, "walking", "bathing", "meal", "watching TV", "cooking", etc. are defined as context candidates, and each sensor output is machine-learned for each context candidate in advance. A cognitive capacity prediction model 103a that estimates the context from each sensor output is constructed. As shown in FIG. 7, the correspondence between the context and the cognitive capacity is registered in advance in the context table 103b.

The evaluation value calculation unit 103c estimates the user's context by applying each sensor output to the context prediction model 102a, and obtains the cognitive capacity by applying the estimation result of the context to the context table 103b. Specific methods for estimating the context are disclosed in Non-Patent Documents 2 and 3.

FIG. 8 is a flowchart showing the operation of the advertisement result evaluation device 1 according to the embodiment of the present invention, and in step S1, prior to the timing of advertisement distribution scheduled in advance, the sensors 12, 13, 14 are shown. Output signal is acquired. The acquisition of each output signal is repeated in a predetermined sampling cycle thereafter.

In step S2, the perception evaluation unit 101 measures the perception index value based on the output signal of each sensor. In the present embodiment, the SN ratio measuring unit 101a measures the SN ratio of the sound, and the relative position measuring unit 101b measures the relative distance between the television and the user.

In step S3, the cognitive capacity evaluation unit 103 estimates the user's context. In this embodiment, the context is estimated by applying the output signal of each sensor to the context prediction model 103a. In this embodiment, the prediction model 103a is constructed so that a plurality of contexts can be estimated at the same time, and when the user is reading while walking, "walking" and "reading" are estimated as the contexts.

In step S4, it is determined whether or not the advertisement to be evaluated has been transmitted. Until the call is issued, the process returns to step S1 and each of the above processes is repeated, and the perceptual index value and the user's context are updated with the latest information.

When the advertisement is transmitted, in step S5, the evaluation value calculation unit 101c of the perception evaluation unit 101 calculates the perception evaluation value p (x) based on the perception index value measured in step S2. In the present embodiment, the perception of voice and the perception of video are evaluated separately according to the advertisement, and in the case of a voice advertisement such as a radio advertisement, only the voice-based perception evaluation value p1 (x) is required. On the other hand, in the case of an advertisement composed of a combination of audio and video such as a television advertisement, an audio-based perceptual evaluation value p1 (x) and a video-based perceptual evaluation value p2 (x) are required.

For the signal-to-noise ratio, the speech-based perception evaluation value p1 (x) can be obtained by normalizing the value in the range of 0 to 1.0. As a method for converting the signal-to-noise ratio into the range of 0 to 1.0, the method (SII) standardized by ANSI in 1997 can be used as an improved version of the articulation index. In SII, the higher the value, the higher the intelligibility.

On the other hand, with respect to the relative position, as shown in FIG. 9, a video-based perception evaluation value p2 (x) is obtained according to the distance to the user with reference to the television as the advertisement generation source. In this embodiment, p2 (x) = 1.0 if the distance from the television is within 2 meters, p2 (x) = 0 if the distance is 10 meters or more, and p2 (x) is linearly set according to the distance between them. Calculated using a changing function.

When the audio-based perceptual evaluation value p1 (x) and the video-based perceptual evaluation value p2 (x) are obtained as described above, the perceptual evaluation values p1 (x) and p2 (x) are applied to the following equation (1). By doing so, the perceptual evaluation value p (x) for the video advertisement can be obtained.
p (x) = (p1 (x) + p2 (x)) / 2… (1)

In step S6, the cognitive capacity evaluation unit 103 obtains the cognitive capacity evaluation value c (x) using the user's context as the cognitive capacity index value. In the present embodiment, the cognitive capacity of the user can be obtained by referring to the context table 103b based on the user's context estimated in step S3.

In the present embodiment, as shown in FIG. 7, the cognitive capacity of 0 to 1.0 is registered for each context in the context table 103b. Then, as described with reference to FIG. 2, the initial value of the cognitive capacity is set to the maximum value of 1.0, and the cognitive capacity corresponding to the estimation result of the context is subtracted from this to calculate the current cognitive capacity of the user. To.

For example, if it is estimated that the user is walking based on the output signal of each sensor, "walking" is estimated as the context. Therefore, based on the following equation (2), the user's current cognitive capacity c. (x) is 0.9.
c (x) = 1.0－0.1 = 0.9… (2)

Further, if it is estimated that the user is reading while walking, "walking" and "reading" are estimated as contexts. Therefore, based on the following equation (3), the user's current cognitive capacity c ( x) is 0.2.
c (x) = 1.0－0.1－0.7 = 0.2… (3)

Furthermore, when the user is reading while cleaning, "cleaning" and "reading" are presumed as contexts, so the user's current cognitive capacity c (x) is based on the following equation (4). It becomes -0.4. However, in this case, the cognitive capacity c (x) is 0 because it is recognized as an overflow state.
c (x) = 1.0－0.7－0.7 = －0.4… (4)

It should be noted that the relationship between the user's context and the cognitive capacity may differ depending on the type of advertisement, that is, whether it is an audio advertisement or a video advertisement. Therefore, a context table for audio advertisement and a context table for video advertisement may be provided to make the relationship between the context and the cognitive capacity different, and each context table may be used properly according to the type of advertisement.

In step S7, the reaction evaluation value r (x) of the user is calculated based on the audio signal analyzed from the output signal of the microphone 13 and the reaction of the user analyzed from the camera image. In the present embodiment, the weight value is obtained by referring to the reaction table 102b based on the estimation result of the reaction. Then, the reaction evaluation value r (x) is obtained by multiplying all the weight values corresponding to the estimated reaction.

For example, if a reaction in which the user changes his / her facial expression while replying to the advertisement is estimated, the reaction evaluation value r (x) is obtained as r (x) = 1.32 based on the following equation (5).
r (x) = 1.2 × 1.1 ＝ 1.32… (5)

Since the relationship between the user's reaction and the weight value may also differ depending on the type of advertisement, if a reaction table for audio advertisement and a reaction table for video advertisement are provided and the relationship between the reaction and the weight value is different. However, each reaction table may be used properly according to the type of advertisement.

In step S8, the advertising result is calculated based on the perceptual evaluation value p (x), the cognitive capacity evaluation value c (x), and the reaction evaluation value r (x). In this embodiment, as shown in the following equation (6), the advertising result v is calculated by multiplying all the evaluation values.
v = p (x) × c (x) × r (x)… (6)

As described above, in the present embodiment, the advertising result is obtained based on the perceptual evaluation value p (x), the cognitive capacity evaluation value c (x), and the reaction evaluation value r (x), but the reaction evaluation value r (x) is Since it always takes a value of 1.0 or more, even if the user's reaction to the advertisement is not recognized, the advertisement result can be quantitatively evaluated based on the perceptual evaluation value p (x) and the cognitive capacity evaluation value c (x). Become.

Further, in the present embodiment, the perceptual evaluation value p (x) becomes 0 when the user cannot perceive the advertisement. Therefore, when it is physically impossible for the user to perceive the advertisement, the cognitive capacity evaluation value c. The advertising result can be set to 0 regardless of (x).

1 ... Advertising performance evaluation device, 2 ... Advertising distribution business equipment, 10 ... User status identification unit, 11 ... Advertising performance determination unit, 12 ... Camera, 13 ... Mike, 14 ... Human sensor, 15 ... Mobile terminal, 101 ... Perception Evaluation unit, 101a ... SN ratio measurement unit, 101b ... Relative position measurement unit, 101c ... Evaluation value calculation unit, 102 ... Reaction evaluation unit, 102a ... Reaction prediction model, 102b ... Reaction table, 102c ... Evaluation value calculation unit, 103 ... The cognitive capacity evaluation unit, 103a ... Cognitive capacity prediction model, 103b ... Context table, 103c ... Evaluation value calculation unit

Claims (8)

In the advertising performance evaluation device that evaluates the performance of advertisements for users
A means of estimating the user's context,
A means of evaluating the user's cognitive capacity based on the estimation result of the context ,
It is provided with a means for evaluating the advertising performance for the user based on the recognition capacity of the user.
The means for estimating the user's context includes a context prediction model that estimates the context based on the output signal of the sensor that detects the user's situation.
An advertising performance evaluation device characterized in that an output signal of a sensor that detects a user's situation is applied to the context prediction model to estimate the user's context . Further equipped with a perceptual evaluation means for evaluating the user's perception and obtaining a perceptual evaluation value,
The advertising performance evaluation device according to claim 1 , wherein the means for evaluating the advertising performance evaluates the advertising performance based on the cognitive capacity and the perceptual evaluation value of the user. Further equipped with a reaction evaluation means for evaluating the reaction of the user and obtaining the reaction evaluation value,
The advertising performance evaluation device according to claim 1 , wherein the means for evaluating the advertising performance evaluates the advertising performance based on the cognitive capacity and the reaction evaluation value of the user. Perceptual evaluation means that evaluates the user's perception and obtains the perceptual evaluation value,
Further provided with a reaction evaluation means for evaluating the reaction of the user and obtaining the reaction evaluation value,
The advertising performance evaluation device according to claim 1 , wherein the means for evaluating the advertising performance evaluates the advertising performance based on the cognitive capacity, the perceptual evaluation value, and the reaction evaluation value of the user. The advertising performance evaluation device according to claim 2 or 4 , wherein the perception evaluation means evaluates the perception of the user based on the SN ratio of sound and the relative position between the advertisement source and the user. The advertising performance evaluation device according to claim 3 or 4 , wherein the reaction evaluation means evaluates the reaction of the user based on the voice of the user and the camera image. In the advertising performance evaluation method that allows the computer to evaluate the performance of the advertisement sent to the user
The computer
Estimate the user's context
The recognition capacity of the user is evaluated based on the estimation result of the context .
Evaluate the advertising performance for the user based on the recognition capacity of the user,
The advertisement is characterized in that the context is estimated by applying the output signal of the sensor that detects the user's situation to a context prediction model that estimates the context based on the output signal of the sensor that detects the user's situation. Achievement evaluation method. In the advertising performance evaluation program that evaluates the performance of advertisements sent to users
The procedure for estimating the user's context and
A procedure for evaluating the user's cognitive capacity based on the estimation result of the context, and
Let the computer perform the procedure of evaluating the advertising performance for the user based on the cognitive capacity of the user .
In the procedure for estimating the context, the context is estimated by applying the output signal of the sensor that detects the user's situation to the context prediction model that estimates the context based on the output signal of the sensor that detects the user's situation. Advertising performance evaluation program featuring .

Images