JP7743462B2 - Viewer attribute estimation device and viewer attribute estimation method - Google Patents

Description

The present invention relates to an apparatus and method for estimating viewer attribute composition from viewing log information obtained for each television device.

In recent years, television devices equipped with the ability to transmit viewing data, including viewing log information indicating which channels were viewed and when, to television manufacturers, etc. via external networks such as the Internet have become widespread. In this specification, a device that can receive and view television broadcasts and has the ability to transmit viewing log information as described above is referred to as a television device.

As seen in Patent Document 1, for example, the present application proposes technology that uses a pre-trained model (mathematical model) to estimate the attribute composition of viewers of each television device (more specifically, one or more viewers who can watch television broadcasts received via that television device) from viewing log information obtained from that television device. Here, the attribute composition of viewers of a television device indicates which of multiple attribute categories, such as gender and age, the viewers are made up of.

Patent No. 6433615

Patent Document 1, cited above, describes a technology for estimating attribute configuration, which estimates the probability of each attribute being present from viewing log information, and estimates attributes whose presence probability is equal to or greater than a predetermined threshold as the attributes of viewers of television devices.

However, with this technology, it is difficult to set an optimal threshold for the probability of presence, and if the threshold is set inappropriately, it is easy for the number of viewers to appear too high or too low compared to the actual number.

The present invention was made in light of this background, and aims to provide an apparatus and method that enable more reliable estimation of viewer attribute composition.

In order to achieve the above object, the viewer attribute estimation device of the present invention is a viewer attribute estimation device that includes a television broadcast receiver and a viewing data output device that is capable of outputting viewing log information indicating when and which channel or broadcast station affiliated television broadcast was viewed via the receiver, and that estimates the attribute composition of viewers of the television device based on viewing log information output from the television device,
Based on the viewing log information, the system is configured to execute the following processes: for each type of attribute of the viewers of the television device, identify an attribute-specific presence probability, which is the probability that a viewer with that attribute exists; calculate the sum of the values of the attribute-specific presence probability; and select attributes of a predetermined number of people set based on the sum in order of the highest value of the attribute-specific presence probability; and to estimate the selected attributes as the attributes of the constituent members of the viewers of the television device.

With the viewer attribute estimation device of the present invention, the predetermined number of viewers is set based on the total value of the attribute-specific presence probability values, thereby making it possible to increase the reliability of the predetermined number of viewers. In this case, the predetermined number of viewers can be set to be larger the larger the total value of the attribute-specific presence probability values.

The attribute composition of the television device's viewers is estimated by selecting the attributes of the specified number of people set as described above in order of the highest attribute-specific presence probability, thereby increasing the reliability of the estimated attribute composition.

In order to achieve the above object, the viewer attribute estimation method of the present invention is an attribute composition estimation method for estimating the attribute composition of viewers of a television device based on viewing log information output from the television device, the television device including a television broadcast receiver and a viewing data output device capable of outputting viewing log information indicating when and which channel or broadcast station television broadcast was viewed via the receiver, the method comprising:
Based on the viewing log information, the method identifies, for each type of attribute of the viewers of the television device, an attribute-specific presence probability, which is the probability that a viewer with that attribute exists, calculates the sum of the identified attribute-specific presence probabilities, selects attributes of a predetermined number of people set based on the calculated sum in descending order of the attribute-specific presence probability, and estimates the selected attributes as the attributes of the members of the viewers of the television device.

Similar to the viewer attribute estimation device of the present invention, the viewer attribute estimation method of the present invention makes it possible to increase the reliability of the attribute composition estimated for viewers of television devices.

1 is a diagram showing an overall system according to an embodiment of the present invention. FIG. 2 is a block diagram illustrating the processing of the viewer attribute estimation device shown in FIG. 1 .

One embodiment of the present invention will be described below with reference to Figures 1 and 2. Referring to Figure 1, the system described in this embodiment includes a viewer attribute estimation device 1 that estimates the attribute composition of viewers of multiple television devices 23. The viewer attribute estimation device 1 is composed of, for example, one or more computers. The computer includes a processor such as a CPU, memory (storage device), an interface circuit, a communication device, etc. (not shown). The viewer attribute estimation device 1 can acquire viewing data related to television broadcast viewing in each device-specific household 20 from multiple device-specific households 20 belonging to a survey area (e.g., a prefecture-based region, or a region combining multiple prefectures such as the Kanto region or the Kinki region) for viewing conditions, etc. The viewing data includes viewing log information indicating which channel (or which broadcasting station affiliate) the television broadcast was viewed on and when.

Here, in this embodiment, each "device-specific household 20" refers to a household that includes a television device 23 that includes a television broadcast receiver 21 and a viewing data output device 22 that can output viewing data related to the viewing of the television broadcast, and that includes one or more viewers of the television device 23 (more specifically, one or more viewers who can watch the television broadcast received by the receiver 21 of the television device 23) as its members. In other words, a device-specific household 20 refers to a combination of a television device 23 and a viewer of the television device 23.

In this case, one device-specific household 20 is associated with each individual television device 23. Therefore, even if a dwelling unit is equipped with multiple television devices 23, one device-specific household 20 is associated with each of the multiple television devices 23. In this case, each device-specific household 20 corresponding to each of the multiple television devices 23 may include the same members as viewers. In other words, one or more members of one device-specific household 20 may also be members of another device-specific household 20.

The television device 23 of each device-specific household 20 may be composed of a television, or may be composed of a television and a recording device connected to it. The viewing data output device 22 of the television device 23 is composed of, for example, a processor such as a microcomputer (not shown), memory, an interface circuit, a communication device, etc. The viewing data output device 22 is capable of detecting the channel of the television broadcast viewed via the television device 23 that includes it, as well as the date and time when the television broadcast on that channel was viewed (more specifically, the date and time when viewing of the television broadcast on that channel began and ended). From this detected information, viewing log information for the device-specific household 20 (hereinafter referred to as device-specific viewing log information) can be generated. In other words, the device-specific viewing log information is the overall viewing log information of the members included in the device-specific household 20.

The viewing data output device 22 can then periodically (or upon request from) transmit viewing data including the generated device-specific viewing log information to a server (not shown) of the manufacturer of the television or recording device that constitutes the television device 23 via an external network NW, such as the Internet or a telephone line network. In this case, the viewing data transmitted from the viewing data output device 22 includes not only device-specific viewing log information, but also location information (e.g., information indicating the first three digits of the postal code) that has been pre-registered on the television or recording device that constitutes the television device 23, and identification information for the television device 23. Note that the identification information for the television device 23 can also be used as identification information for the device-specific household 20 that owns the television device 23.

The viewer attribute estimation device 1 includes a viewing data acquisition unit 11 and an attribute configuration estimation unit 12 as functions realized by the implemented hardware configuration and program (software configuration). Below, we will explain the overall processing of the viewer attribute estimation device 1, along with details of each of these functional units.

The viewing data acquisition unit 11 is capable of communicating with a server (not shown) of the manufacturer of each television device 23 in the device-specific household 20, and by communicating in this manner, is able to acquire viewing data for each television device 23 of that manufacturer. The viewing data acquisition unit 11 functions as a viewing log information acquisition unit in the present invention.

In addition, if the viewer attribute estimation device 1 is able to communicate with the viewing data output device 22 of the device-specific household 20, the viewing data acquisition unit 11 may acquire viewing data of the television device 23 of the device-specific household 20 directly from the viewing data output device 22 of that television device 23. In addition, the viewer attribute estimation device 1 may acquire viewing data of the television device 23 from each manufacturer via an appropriate storage device.

In this embodiment, the viewer attribute estimation device 1 executes the processing of the attribute composition estimation unit 12 using the viewing data acquired by the viewing data acquisition unit 11 for each device-specific household 20 in the survey area.

The attribute composition estimation unit 12 is a functional unit that estimates the attribute composition indicating which members of which attributes are included in the device-specific household 20 (in other words, the attribute composition of viewers of the television devices 23 of the device-specific household 20) from device-specific viewing log information for a predetermined period (e.g., one month, several months, one year, etc.) output from the television devices 23 of each device-specific household 20.

Here, the attributes of the members (viewers of television devices 23) of each device-specific household 20 are classified into multiple types of attributes, for example, according to the age and gender of the members. For example, they are classified into multiple types of attributes such as children under y1, men aged y1 or older but under y2, women aged y1 or older but under y2, men aged y2 or older, and women aged y2 or older. Hereinafter, the number of types of attributes will be designated as N, and each of the N types of attributes will be appropriately represented as At(i) (i = 1, 2, ..., N). Note that the attributes of the members of each device-specific household 20 can be classified according to various parameters other than age and gender, such as occupation and educational background.

As shown in FIG. 2, the attribute composition estimation unit 12 estimates the attribute composition of the members of each device-specific household 20 using an attribute composition estimation model that has been previously subjected to machine learning processing. Here, the attribute composition estimation model is a model that has been previously subjected to machine learning processing so that the attribute composition of the members of any device-specific household 20 belonging to the survey area can be estimated from the device-specific viewing logs for a predetermined period of that device-specific household 20. In the machine learning processing for the attribute composition estimation model, viewing log information for each sample household obtained from people meters or the like installed in each of multiple sample households with known attribute compositions and the attribute composition of each sample household are used as learning data. Furthermore, a publicly known algorithm can be used as the algorithm for this machine learning processing.

In this case, in this embodiment, the attribute composition estimation model is configured to be able to determine, for each of N types of attributes At(i) (i = 1, 2, ..., N), an attribute-specific presence probability, which is the probability that a member of each attribute At(i) is present in the device-specific household 20, from the device-specific viewing log information for a predetermined period of time for each device-specific household 20. Such an attribute composition estimation model may be, for example, a model similar to the mathematical model for determining household composition in Patent Document 1. However, the attribute composition estimation model may be of another form, as long as it can determine an attribute-specific presence probability (or an index value similar thereto) from the device-specific viewing log information for a predetermined period of time for the device-specific household 20.

The attribute composition estimation unit 12 then estimates the attributes of the members of each device-specific household 20 based on the attribute-specific presence probability identified by the attribute composition estimation model for each device-specific household 20. Specifically, the attribute composition estimation unit 12 calculates the total value of the attribute-specific presence probability identified by the attribute composition estimation model for each device-specific household 20, and sets the number of members of the device-specific household 20 according to the calculated total value.

In this case, the number of constituent members of a device-specific household 20 is set to be greater the greater the total value of the attribute-specific presence probability. For example, if the total value is 1.8, which exceeds 1, the number of constituent members is set to 1; if the total value is 2.5, which exceeds 2, the number of constituent members is set to 2; if the total value is 3.4, which exceeds 3, the number of constituent members is set to 3; and so on. The number of constituent factors is set according to the total value.

The attribute composition estimation unit 12 then selects attributes for the set number of people from the N types of attributes At(i) (i = 1, 2, ..., N) for which the attribute-specific presence probability has been specified, in descending order of the attribute-specific presence probability. As an example, Figure 2 illustrates the processing of the attribute composition estimation unit 12 for one device-specific household 20 (in Figure 2, the device-specific household 20 whose identification information ID is x1). Note that in the block diagram shown in Figure 2, the processing execution unit is indicated by a thick frame, and data is indicated by a thin frame.

In the example shown in Figure 2, the attribute-specific presence probability of members of a device-specific household 20, whose identification information ID is x1, is determined by an attribute composition estimation model from device-specific viewing log information for a predetermined period of time. In this example, the total value of the attribute-specific presence probability across all N types of attributes At(i) (i = 1, 2, ..., N) is, for example, 3.4, and the number of members is set to 3.

Of the N types of attributes At(i) (i = 1, 2, ..., N), the attribute-specific existence probability of each of attributes At(1), At(3), and At(N) is higher than the attribute-specific existence probability of the other attributes. In this case, the attribute composition estimation unit 12 estimates that the attributes At(1), At(3), and At(N) for three people whose attribute-specific existence probability is higher than the threshold value are the attributes of the members of the device-specific household 20 (viewers of the television device 23), and estimates that the other attributes At(2), At(4) to At(N-1) are not the attributes of the members of the device-specific household 20. In this way, the attribute composition of the members of the device-specific household 20 is estimated.

In this way, the attribute composition estimation unit 12 estimates the attribute composition of the members of the device-specific household 20 by selecting attributes for the number of people set according to the total value of the attribute-specific presence probability, in order from highest to lowest attribute-specific presence probability.

In this case, the reliability of the number of constituent factors can be improved by setting the number of constituent members of each device-specific household 20 to be greater as the total value of the attribute-specific presence probability increases (or set the number of constituent members to be smaller as the total value decreases). This in turn makes it possible to improve the reliability of the attribute composition of each device-specific household 20.
The attribute configuration estimated for each device-specific household 20 as described above can be utilized for investigating the viewing status of the television device 20 for each attribute.

1... viewer attribute estimation device, 21... receiver, 22... viewing data output device, 23... television device

Claims (2)

A viewer attribute estimation device that estimates the attribute composition of viewers of a television device based on viewing log information output from the television device, the television device comprising: a television broadcast receiver; and a viewing data output device that is capable of outputting viewing log information indicating when and which channel or broadcast station television broadcast was viewed via the receiver, the viewer attribute estimation device comprising:
a viewer attribute estimation device configured to execute the following processes based on the viewing log information: a process of identifying, for each type of attribute of a viewer of the television device, an attribute-specific presence probability, which is the probability that a viewer with that attribute exists; a process of calculating a total value of the attribute-specific presence probability values; and a process of selecting, based on the total value, attributes of a predetermined number of people set in descending order of the attribute-specific presence probability values; and to estimate the selected attributes as attributes of the constituent members of the viewers of the television device. A viewer attribute estimation method for estimating an attribute configuration of viewers of a television device based on viewing log information output from the television device, the television device including a television broadcast receiver and a viewing data output device capable of outputting viewing log information indicating when and which channel or broadcast station television broadcast was viewed via the receiver, the method comprising:
A viewer attribute estimation method characterized by: based on the viewing log information, identifying an attribute-specific presence probability, which is the probability that a viewer with that attribute exists, for each type of attribute of the viewer of the television device; calculating the sum of the identified attribute-specific presence probability values; selecting attributes of a predetermined number of people set based on the calculated sum in descending order of the attribute-specific presence probability values; and estimating the selected attributes as the attributes of the constituent members of the viewers of the television device.