Viewing information analysis system and viewing information analysis method

Abstract

To provide a viewing information analysis system and a viewing information analysis method capable of accurately and efficiently estimating the household constitution of a target household.SOLUTION: A viewing information analysis system 10 comprises a viewing record data acquisition unit 11, a relation data learning part 12, a target viewing record data acquisition unit 13, a presence provability calculation unit 14, and a household constitution estimation unit 15.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a viewing information analysis system and a viewing information analysis method for obtaining information on members of a household from a viewing log recorded for the household. [Background technology]

[0002] Conventionally, as a viewing information analysis system of this type, a system has been known, such as Patent Document 1 by the applicant of the present application, which has a first viewing record data acquisition unit that acquires first viewing record data including a viewing log indicating the viewing channels and viewing dates and times recorded for each specified household and the attributes of the members included in each specified household; a relationship data learning unit that learns first relationship data indicating the relationship between the viewing log and the attributes of the members included in the household based on the first viewing record data; a second viewing record data acquisition unit that acquires second viewing record data including a viewing log indicating the viewing channels and viewing dates and times recorded for a target household; a presence probability calculation unit that calculates, for each attribute of the members in the target household based on the first relationship data and the second viewing record data, an existence probability indicating the probability that a member with each attribute is included in the target household; and a household composition determination unit that determines the attributes of the members comprising the target household based on the existence probability for each attribute of the members of the target household calculated by the existence probability calculation unit.

[0003] According to such a conventional viewing information analysis system, it is possible to determine the attributes of members who are likely to be included in the target household from the viewing log of the target household, and to accurately determine the household composition of the target household. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent No. 6433615 Summary of the Invention [Problem to be solved by the invention]

[0005] The inventors of the present application have conducted research to further improve the accuracy of the household composition of the target households in such conventional viewing information analysis systems and to determine it more efficiently, and have invented a unique system and method for accurately determining the household composition of the target households.

[0006] As described above, an object of the present invention is to provide a viewing information analysis system and a viewing information analysis method that can estimate the household composition of a target household with high accuracy and high efficiency. [Means for solving the problem]

[0007] The viewing information analysis system of the first invention comprises: a viewing record data acquisition unit that acquires viewing record data including a viewing log indicating viewing channels and viewing dates and times recorded for each predetermined household, and information on household members who viewed the programs related to each viewing log; a relational data learning unit that learns relational data indicating a relationship between the viewing log and attributes of household members based on the viewing record data; a target viewing record data acquisition unit that acquires target viewing record data including a viewing log indicating viewing channels and viewing dates and times recorded for the target household; a presence probability calculation unit that calculates the presence probability for each attribute of household members in the target household based on the relationship data and the target viewing record data; a household structure estimation unit that estimates attributes of household members that constitute the target household based on the relationship data, the target viewing record data, and the presence probability for each attribute of the household members calculated by the presence probability calculation unit; and In a viewing information analysis system comprising: The household composition estimation unit estimates the estimated number of household members of the target household from the target viewing record data acquired by the target viewing record data acquisition unit, and corresponds it to a predetermined number of person category in which differences in viewing become significant, and estimates the combination with the highest probability of existence among combinations of household members that the estimated household number corresponds to within the corresponding number of person category as the primary attribute of the household members that make up the target household.

[0008] According to the viewing information analysis system of the first invention, before estimating the attributes of the household members constituting the target household based on the probability of existence of each attribute of the household members, the estimated number of household members of the target household is estimated, and it is determined which of the predetermined number of household members categories in which differences in viewing are significant falls into which the estimated number of household members falls, and within that corresponding number of people category, the combination of household members with the highest probability of existence is estimated as the primary attribute of the household members constituting the target household.

[0009] This allows the system to not only estimate the attributes of the household members who make up the target household based on the probability of existence, but also classify them using a simple and clear filter into number categories where differences in viewing are significant, and then find the combination of household members that has the highest probability of existence among the combinations of household members that can be taken within that number category, thereby dramatically improving the accuracy of the estimation and dramatically improving processing efficiency. In this way, according to the viewing information analysis system of the first invention, it is possible to estimate the household composition of the target household with high accuracy and high efficiency.

[0010] The viewing information analysis system of the second invention is the first invention, The household composition estimation unit is characterized in that it corresponds the number of people category to a first number of people category that determines whether the target household is single or multiple, and for the target household determined to be multiple, to a second number of people category that determines whether the target household is a two-person household or a three or more person household, and estimates the combination of household members with the highest probability of existence as the primary attribute of the household members that make up the target household.

[0011] According to the viewing information analysis system of the second invention, the first number category for determining whether a household is single or multiple is determined as a predetermined number category in which differences in viewing become significant, for example, from the viewing minute distribution (an analytical method for looking at viewing households (viewers) for a program or time period from the distribution of viewing time), and for target households determined to be single, the primary attributes can be estimated with high accuracy from the combination of single members using the existence probability.

[0012] Furthermore, for target households determined to have multiple members, the number of members can be classified as two or three or more, for example, by determining the number of members from the distribution of viewing minutes or the distribution of presence probability.For target households determined to have two members, the primary attributes can be estimated with high accuracy from combinations of two members using the presence probability, and for target households determined to have three or more members, the primary attributes can be estimated with high accuracy from combinations of three or more members using the presence probability.

[0013] In this way, according to the viewing information analysis system of the second invention, it is possible to estimate the actual household composition of the target household with high accuracy and high efficiency.

[0014] The viewing information analysis system of the third invention is the first or second invention, The household composition estimation unit is characterized in that it estimates the secondary attribute with the highest probability of existence as the attribute of the household members constituting the target household in a secondary characteristic classification that further subdivides the primary attribute.

[0015] According to the viewing information analysis system of the third invention, since primary attributes can be estimated with high accuracy by the first or second invention, even when such primary attributes are further classified into secondary characteristic categories, secondary attributes can be estimated by existence probability.

[0016] In this way, according to the viewing information analysis system of the third invention, estimation can be performed with high accuracy and efficiency even when the household structure of the target household is subdivided.

[0017] The viewing information analysis system of the fourth invention is the first invention, As a pre-processing step, the household composition estimation unit determines whether the target viewing record data acquired by the target viewing record data acquisition unit exceeds a viewing volume threshold for a predetermined viewing time, and if the viewing volume threshold is exceeded, executes a process to estimate the attributes of the household members who make up the target household.

[0018] According to the viewing information analysis system of the fourth invention, if the viewing volume of the target viewing record data is small, the estimation accuracy may decrease. However, by determining whether the viewing volume exceeds a predetermined viewing time threshold, and if the viewing volume threshold is exceeded, the attributes of the household members who make up the target household are estimated, thereby ensuring high estimation accuracy.

[0019] Here, if the viewing volume threshold is not exceeded, it is possible to ensure a certain level of estimation accuracy by, for example, assigning primary attributes to the attributes of the household members making up the target household on a rule-based basis (assigning them so as to reproduce the family composition ratio of the target household).

[0020] In this way, according to the viewing information analysis system of the fourth aspect of the present invention, it is possible to estimate the household composition of a target household with high accuracy and efficiency, taking into account the viewing volume.

[0021] The viewing information analysis method of the fifth invention comprises: a viewing record data acquisition step of acquiring viewing record data including a viewing log indicating viewing channels and viewing dates and times recorded for each predetermined household, and information on household members who viewed the programs related to each viewing log; a relational data learning step of learning relational data indicating a relationship between the viewing log and attributes of household members based on the viewing record data; a target viewing record data acquisition step of acquiring target viewing record data including a viewing log indicating viewing channels and viewing dates and times recorded for the target household; a presence probability calculation step of calculating a presence probability for each attribute of household members in the target household based on the relationship data and the target viewing record data; a household composition estimation step of estimating attributes of household members constituting the target household based on the relationship data, the target viewing record data, and the presence probability for each attribute of the household members calculated by the presence probability calculation step; In the viewing information analysis method, The household composition estimation process is characterized in that it estimates the estimated number of household members of the target household from the target viewing record data acquired by the target viewing record data acquisition process, corresponds to a predetermined number of household members category in which differences in viewing become significant, and estimates the combination with the highest probability of existence among combinations of household members that the estimated household number corresponds to within the corresponding number of household members category as the primary attribute of the household members that make up the target household.

[0022] According to the viewing information analysis method of the fifth invention, before estimating the attributes of the household members constituting the target household based on the probability of existence for each attribute of the household members, the estimated number of household members in the target household is estimated, and it is determined which of the predetermined number-of-person categories in which differences in viewing are significant falls into which the estimated number of household members falls, and within that corresponding number-of-person category, the combination of household members with the highest probability of existence is estimated as the primary attribute of the household members constituting the target household.

[0023] This allows the system to not only estimate the attributes of the household members who make up the target household based on the probability of existence, but also classify them using a simple and clear filter into number categories where differences in viewing are significant, and then find the combination of household members that has the highest probability of existence among the combinations of household members that can be taken within that number category, thereby dramatically improving the accuracy of the estimation and dramatically improving processing efficiency. In this way, according to the viewing information analysis method of the fifth aspect of the present invention, the household composition of the target household can be estimated with high accuracy and high efficiency. [Brief explanation of the drawings]

[0024] [Figure 1] 1 is a system configuration diagram showing the configuration of an audience information analysis system according to an embodiment of the present invention. [Figure 2] FIG. 2 is an explanatory diagram showing the processing content of the audience information analysis system of FIG. DETAILED DESCRIPTION OF THE INVENTION

[0025] An audience information analysis system according to one embodiment of the present invention will be described with reference to FIG.

[0026] As shown in Figure 1, the viewing information analysis system 10 is a viewing information processing device that performs processing related to investigating the viewing status of television broadcasts, etc., and is connected to a television device S20 installed in a sample household S and a television device T30 installed in a target household T via an external network NW composed of the Internet, a telephone line network, etc.

[0027] Sample household S is a household randomly selected from households in a specified area for a viewer rating survey. For sample household S, history information (viewing log) of channels viewed by each member of the household on a receiver such as a television can be recorded. In other words, for sample household S, both the program viewing log for each household (household viewing log) and the program viewing log for each individual member of the household (individual viewing log) can be obtained.

[0028] It should be noted that sample household S includes multiple households, but for the sake of explanation, only one household in sample household S, the first sample household, will be used as an example below.

[0029] A personal viewing log recording system capable of recording personal viewing logs of each member of the household is installed in the television device S20 of the first sample household S. The personal viewing log recording system includes a receiver S21 and a viewing data output device S22 that has a channel sensor for identifying individuals and their viewing channels, a personal identifier (specifically, a people meter), and an online meter, and is capable of outputting viewing data related to viewing of television broadcasts.

[0030] As a result, while the receiver S21 is powered on, first viewing record data is generated at regular time intervals (for example, one-minute intervals), and the first viewing record data and personal viewing record data, which consist of logs generated every minute, are sequentially transmitted to the viewing information analysis system 10 via the network NW.

[0031] The target household T is a household extracted from households in a specified area for an audience rating survey. The target household T may be extracted from the same area as the sample household S, or may be extracted from an area different from that of the sample household S. Note that in this embodiment, only one target household T is shown for the sake of explanation, but there may be multiple target households T.

[0032] The viewing log recording system of the television device T30 of the target household T is capable of recording only the viewing log of the household's programs. In other words, the viewing log recording system of the television device T30 of the target household T is equipped with a receiver T31 and a viewing data output device T32 having a channel sensor and an online meter for identifying the household's viewing channels, and differs from the viewing data output device S22 of the sample household S in that it does not have a personal identifier.

[0033] In this way, for the target household T, the program viewing logs of each member of the household are not recorded.

[0034] The viewing information analysis system 10 is a device that performs analysis based on viewing record data and personal viewing record data obtained from a sample household S, and viewing record data obtained from a target household T. Specifically, the viewing information analysis system 10 performs processing to estimate the household composition of a target household T, whose composition is unknown, based on the viewing record data and personal viewing record data obtained from the sample household S.

[0035] The viewing information analysis system 10 includes a viewing record data acquisition unit 11, a relationship data learning unit 12, a target viewing record data acquisition unit 13, a presence probability calculation unit 14, and a household composition estimation unit 15. Note that the basic functions of each processing unit are the same as those in Patent Document 1, and therefore detailed explanations will be omitted.

[0036] The viewing record data acquisition unit 11 acquires viewing record data from the sample household S, which includes a viewing log indicating the viewing channel and viewing date and time recorded for each household, and information on the household members who viewed the content related to each viewing log.

[0037] The relational data learning unit 12 learns relational data indicating the relationship between the viewing log and the attributes of the household members based on the viewing record data acquired from the sample household S.

[0038] Specifically, the relational data learning unit 12 learns relational data indicating the relationship between the viewing log and the attributes of the members based on the viewing record data acquired from the sample household S.

[0039] The relationship data is data for calculating the probability of a member having each attribute in a household based on a viewing log recorded over a predetermined period (e.g., one month). For example, the relationship data is expressed as an equation showing the relationship between an explanatory variable and a target variable in a regression analysis.

[0040] The target viewing record data acquisition unit 13 acquires, from the target household T, target viewing record data including a viewing log indicating the viewing channels and viewing dates and times recorded for the target household T.

[0041] The presence probability calculation unit 14 calculates the presence probability for each attribute of household members in the target household T based on the relationship data and the target viewing record data.

[0042] The household configuration estimation unit 15 estimates the attributes of the household members constituting the target household T based on the relationship data, the target viewing record data, and the presence probability for each attribute of the household members calculated by the presence probability calculation unit.

[0043] The above is the system configuration of the viewing information analysis system 10. In the above configuration, each of the processing units 11 to 15 is configured with hardware such as a CPU (Central Processing Unit), a ROM (Read Only Memory), and a RAM (Random Access Memory), and functions as an arithmetic device (sequencer) for executing various processes by storing and holding programs for executing various processes described below in memory (not shown) and executing the programs. Also, some or all of the processing units 11 to 15 may be configured with other servers (external servers), and the matching platform 10 may be realized by distributed processing.

[0044] Next, the processing content (audience information analysis of the present invention) by the household structure estimation unit 15 of the audience information analysis system 10 will be described in detail.

[0045] As a premise, the viewing information analysis system 10 performs a viewing record data acquisition process in which the viewing record data acquisition unit acquires viewing record data from the sample household S, the viewing log indicating the viewing channel and viewing date and time recorded for each household, and information on the household members who viewed the content related to each viewing log, and the target viewing record data acquisition process in which the target viewing record data acquisition unit acquires target viewing record data including the viewing log indicating the viewing channel and viewing date and time recorded for the target household.

[0046] First, the household composition estimation unit 15 determines whether the viewing volume of the target viewing record data of the target household T acquired from the target household T exceeds a viewing volume threshold for a predetermined viewing time (for example, a predetermined viewing time in one month) (STEP 10).

[0047] Then, if the viewing volume threshold is not exceeded, the household composition estimation unit 15 determines that the viewing volume is low, and assigns 10 primary attributes to the attributes of the household members constituting the target household T on a rule-based basis (assigning them so as to reproduce the family composition ratio of the target household) (STEP 11).

[0048] Here, the 10 characteristics are information represented by CHILD (children: 4-12 years old), TEEN (children: 13-19 years old), M1 (men: 20-34 years old), M2 (men: 35-49 years old), M3 (men: 50-64 years old), M4 (men: 65 years old and over), F1 (women: 20-34 years old), F2 (women: 35-49 years old), F3 (women: 50-64 years old), and F4 (women: 65 years old and over), which classify members by their age or gender, and correspond to the primary attributes of the ``member attributes.''

[0049] Furthermore, the household structure estimation unit 15 assigns secondary attributes by further dividing the 10 characteristics assigned in STEP 11 into 40 characteristics (STEP 12). Specifically, while maintaining the results of the 10 characteristics, secondary attributes of the 40 characteristics are assigned on a rule basis (assigning so as to reproduce the ratio of the 40 characteristics of the entire audience rating panel survey).

[0050] Specifically, the 40 characteristics are divided into CHILD (children: 4-12 years old) and TEEN (children: 13-19 years old) by gender and by generation, roughly divided into 5-year increments: M4-5 (males: 4-5 years old), M6-12 (males: 6-12 years old), M13-15 (males: 13-15 years old), M16-19 (males: 17-19 years old), M20-24 (males: 20-24 years old), M25-29 (males: 25-29 years old). ·M95-99 (Male: 95-99 years old), similarly, F4-5 (Female: 4-5 years old), F6-12 (Female: 6-12 years old), F13-15 (Female: 13-15 years old), F16-19 (Female: 16-19 years old), F20-24 (Female: 20-24 years old), F25-29 (Female: 25-29 years old) ··F95-99 (Female: 95-99 years old) is information represented and corresponds to the secondary attribute of the "member attributes".

[0051] On the other hand, if the viewing volume threshold is exceeded, the household composition estimation unit 15 estimates the number of household members constituting the target household T and determines whether the target household T is a single-person (single-person household) or multiple-person (multiple-person household) as the first number-of-person category (STEP 20).

[0052] Specifically, the household structure estimation unit 15 causes the relationship data learning unit 12 to learn relationship data indicating a relationship between the viewing record data acquired from the sample household S as an explanatory variable, the viewing log and member attributes as an objective variable, and whether the family structure of the household (device) is single or multiple (corresponding to the relationship data learning step of the present invention), and determines whether the target household T is a single-person or multiple-person household based on the relationship data (corresponding to the existence probability calculation step of the present invention). Note that by using the sample household S as the result of a nationwide audience rating panel survey, this determination is the same as determining whether the household is single or multiple based on a threshold value that matches the distribution of single / multiple people in the nationwide audience rating panel survey based on the probability value of the number of people estimated.

[0053] Then, if the target household T is determined to be single-person in STEP 20, the household composition estimation unit 15 estimates as the primary attribute the attribute that has the highest probability of existence for each attribute of the household members in the first number category (excluding CHILD) (STEP 21, which corresponds to the household composition estimation process of the present invention).

[0054] Specifically, the household structure estimation unit 15 causes the relationship data learning unit 12 to learn relationship data indicating the relationship between the viewing record data acquired from the sample household S as an explanatory variable, the viewing log and the attributes of the members as an objective variable, and the presence or absence of the 10 characteristics as an objective variable, and estimates the primary attribute based on the presence probability of the 10 characteristics of the target household T on the basis of the relationship data. For example, in Figure 2, since the presence probability of M2 is the highest, M2 is estimated as the primary attribute.

[0055] On the other hand, if the household is determined to be a multiple household in STEP 20, the household composition estimation unit 15 determines whether the target household T is a two-person household or a household with three or more people as the second number of people category (STEP 40) based on the probability of presence of each attribute of the household members in the target household T using 10 characteristics (similar to STEP 21) (STEP 31).

[0056] Specifically, the household structure estimation unit 15 causes the relationship data learning unit 12 to learn relationship data indicating a relationship in which the existence probability values ​​of the 10 characteristics of the viewing record data acquired from the sample household S are used as explanatory variables and whether the number of family types is two or three or more is used as the objective variable (this corresponds to the relationship data learning step of the present invention, and based on this relationship data, it determines whether the target household T is a two-person household or a household of three or more people (this corresponds to the existence probability calculation step of the present invention).

[0057] Then, the household composition estimation unit 15 estimates the combination of household members with the highest probability of existence as the primary attribute of the household members that make up the target household T, corresponding to whether the household is a two-person household or a three or more person household determined in STEP 40 (STEP 41, which corresponds to the household composition estimation process of the present invention).

[0058] That is, when it is determined in STEP 40 that the household is a two-person household, the household structure estimation unit 15 estimates the attribute of the combination with the highest existence probability in STEP 31 from among the combinations of the 10 characteristics of a two-person household as the primary attribute.

[0059] Similarly, if the household is determined to be a household of three or more people in STEP 40, the household structure estimation unit 15 estimates the attribute of the combination with the highest probability of existence in STEP 31 among the combinations of the 10 characteristics of households of three or more people as the primary attribute.

[0060] The household composition estimation unit 15 estimates the attribute with the highest probability of existence as the secondary attribute in the second secondary characteristic classification, which is a further subdivision of the primary attributes estimated in STEP 21 and STEP 41 (STEP 42, which corresponds to the household composition estimation process of the present invention).

[0061] Specifically, the household composition estimation unit 15 has the relationship data learning unit 12 learn relationship data that shows the relationship between the viewing record data acquired from the sample household S as an explanatory variable, and the viewing log and member attributes as the objective variable, with the presence or absence of the 40 characteristics as the objective variable (this corresponds to the relationship data learning process of the present invention), and estimates secondary attributes based on the presence probability of the target household T in the 40 characteristics (the presence probability estimation process of the present invention) based on this relationship data.

[0062] For example, in Figure 2, in STEP 41, the primary attributes of the household members making up target household T are estimated to be M2, F1, and TEEN. M2 (men: 35-49 years old) is further subdivided into M35-39 (men: 35-39 years old), M40-44 (men: 40-44 years old), and M45-49 (men: 45-49 years old), and M35-39 is estimated as the secondary attribute, as it has the highest probability of existence.

[0063] Similarly, F2 (women: 35-49 years old) is further subdivided into F35-39 (women: 35-39 years old), F40-44 (women: 40-44 years old), and F45-49 (women: 45-49 years old), and F35-39, which has the highest probability of existence, is estimated as the secondary attribute.

[0064] Similarly, when TEEN (children: 13-19 years old) is subdivided, M13-15 (male: 13-15 years old), M16-19 (male: 17-19 years old), F13-15 (female: 13-15 years old), and F16-19 (female: 16-19 years old), F16-19, which has the highest probability of existence, is estimated as the secondary attribute.

[0065] Then, the household structure estimation unit 15 estimates the combination of secondary attributes estimated in STEP 42 as the attributes of the household members that make up the target household T (STEP 50, which corresponds to the household structure estimation step of the present invention).

[0066] As described above, before estimating the attributes of the household members who make up the target household T based on the probability of existence for each attribute of the household members, the estimated number of household members in the target household T is estimated, and it is determined which of the pre-set number-of-persons categories in which the estimated number of household members falls into which the difference in viewing is significant is determined. Within that corresponding number-of-persons category, the attributes of the household members who make up the target household are estimated based on the combination of household members that has the highest probability of existence, thereby dramatically improving the accuracy of the estimation and dramatically improving processing efficiency.

[0067] As described above, the viewing information analysis system and viewing information analysis method according to this embodiment can estimate the household composition of a target household with high accuracy and high efficiency. [Explanation of symbols]

[0068] 10...viewing information analysis system, 11...viewing record data acquisition unit 11, 12...relationship data learning unit, 13...target viewing record data acquisition unit, 14...existence probability calculation unit, 15...household composition estimation unit, S...sample household, S20...television device of sample household, S21...receiver of sample household, S22...viewing data output device of sample household (with personal identifier), T...target household, T30...television device of target household, T31...receiver of target household, T32...viewing data output device of target household (without personal identifier).

Claims

1. a viewing record data acquisition unit that acquires viewing record data including a viewing log indicating viewing channels and viewing dates and times recorded for each predetermined household, and information on household members who viewed the programs related to each viewing log; a relational data learning unit that learns relational data indicating a relationship between the viewing log and attributes of household members based on the viewing record data; a target viewing record data acquisition unit that acquires target viewing record data including a viewing log indicating viewing channels and viewing dates and times recorded for the target household; a presence probability calculation unit that calculates the presence probability for each attribute of household members in the target household based on the relationship data and the target viewing record data; a household structure estimation unit that estimates attributes of household members that constitute the target household based on the relationship data, the target viewing record data, and the presence probability for each attribute of the household members calculated by the presence probability calculation unit; and In a viewing information analysis system comprising: The household composition estimation unit estimates the estimated number of people in the target household from the target viewing record data acquired by the target viewing record data acquisition unit, and corresponds it to a predetermined number of people category in which differences in viewing are significant, and estimates the combination of household members that has the highest probability of existence among the combinations of household members that the estimated household number corresponds to within the corresponding number of people category as the primary attribute of the household members that make up the target household.

2. 2. The viewing information analysis system according to claim 1, The household composition estimation unit corresponds the number of people classification to a first number of people classification that determines whether the target household is single or multiple, and for target households determined to be multiple, to a second number of people classification that determines whether the target household is a two-person household or a household of three or more people, and estimates the combination of household members with the highest probability of existence as the primary attribute of the household members making up the target household. This is a viewing information analysis system characterized by the above.

3. 3. The viewing information analysis system according to claim 1, The household composition estimation unit estimates the secondary attribute with the highest probability of existence in a secondary characteristic classification that further subdivides the primary attribute as the attribute of the household members that make up the target household, in a viewing information analysis system characterized by the above.

4. 2. The viewing information analysis system according to claim 1, A viewing information analysis system characterized in that, as a preprocessing step, the household composition estimation unit determines whether the target viewing record data acquired by the target viewing record data acquisition unit exceeds a viewing volume threshold for a predetermined viewing time, and if the viewing volume threshold is exceeded, executes a process to estimate the attributes of the household members who make up the target household.

5. a viewing record data acquisition step of acquiring viewing record data including a viewing log indicating viewing channels and viewing dates and times recorded for each predetermined household, and information on household members who viewed the programs related to each viewing log; a relational data learning step of learning relational data indicating a relationship between the viewing log and attributes of household members based on the viewing record data; a target viewing record data acquisition step of acquiring target viewing record data including a viewing log indicating viewing channels and viewing dates and times recorded for the target household; a presence probability calculation step of calculating a presence probability for each attribute of household members in the target household based on the relationship data and the target viewing record data; a household composition estimation step of estimating attributes of household members constituting the target household based on the relationship data, the target viewing record data, and the presence probability for each attribute of the household members calculated by the presence probability calculation step; In the viewing information analysis method, The household composition estimation step estimates the estimated number of household members of the target household from the target viewing record data acquired by the target viewing record data acquisition step, and corresponds it to a predetermined number of people category in which differences in viewing are significant, and estimates the combination with the highest probability of existence among combinations of household members that the estimated household number corresponds to within the corresponding number of people category as the primary attribute of the household members that make up the target household.

Images