Attribute estimation method and attribute estimation system

Abstract

To provide an attribute estimation method and an attribute estimation system capable of estimating an attribute of a set of target persons even when attribute information of the set of target persons is insufficient.SOLUTION: The attribute estimation system 10 is a system for estimating an unknown attribute of a group of subjects, and includes a subject information acquisition unit 11, a survey data acquisition unit 12, an attribute estimation unit 13, and an affinity determination unit 14.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an attribute estimation method and an attribute estimation system for estimating unknown attributes of a group of subjects. [Background technology]

[0002] Conventionally, a system using this type of attribute estimation is known, as shown in Patent Document 1 below, which is used in online research in which an operator surveys registered monitors via the Internet, and which includes a monitor information database that stores attribute information of registered monitors, and a system that estimates an expanded population by weighted aggregation of attribute information of multiple users who are not registered monitors based on the attribute information of multiple users who are registered monitors, and creates an expanded population database that includes the attribute information and behavioral history of all users. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2016-076125 Summary of the Invention [Problem to be solved by the invention]

[0004] In this way, in systems that use conventional attribute estimation, a survey of registered monitors is conducted as a sample of the population that includes those who are not registered monitors, and attribute information of the registered monitors is obtained.The results of this sample survey of registered monitors are then applied (weighted aggregation) to the entire population, including those who are not registered monitors, thereby estimating the attribute information of those who are not registered monitors.

[0005] However, if there is insufficient attribute information on registered monitors, it is not possible to use the expanded population estimation method based on weighted aggregation.

[0006] Here, the inventors of the present application have conducted extensive research into weighted-back aggregation and have come to invent a unique method and system for estimating the attributes of a group of subjects, even when there is insufficient attribute information about the group of subjects who are monitors.

[0007] As described above, an object of the present invention is to provide an attribute estimation method and an attribute estimation system that can estimate attributes of a group of subjects even when attribute information about the group of subjects is insufficient. [Means for solving the problem]

[0008] The attribute estimation method of the first invention is an attribute estimation method for estimating unknown attributes of a set of subjects, the method comprising: a subject information acquisition step of acquiring known attributes of the subjects for the set of subjects; a survey data acquisition step of acquiring survey data including survey items corresponding to the known attributes; an attribute estimation step of estimating results of survey items other than the known attributes in the survey data as unknown attributes of the group of subjects by performing weighted aggregation so that the known attributes of the survey data acquired in the survey data acquisition step match the known attributes of the group of subjects acquired in the subject information acquisition step; The present invention is characterized by carrying out the following.

[0009] According to the attribute estimation method of the first invention, attention is focused only on the known attributes possessed by a group of subjects, and the known attributes in the survey data are weighted and aggregated to match the proportions of the known attributes, thereby estimating and assigning attributes other than the known attributes possessed by the survey data to the group of subjects.

[0010] In this way, according to the attribute estimation method of the first invention, even when attribute information of a group of subjects is insufficient, attributes of the group of subjects can be estimated.

[0011] The attribute estimation method of the second invention is the method of the first invention, The attribute estimation process is characterized in that the results of survey items other than the known attributes in the survey data are estimated as unknown attributes of the group of subjects by performing weighted aggregation using limited survey data obtained by narrowing down the survey data obtained in the survey data acquisition process to survey data having specific attributes, so as to match the attributes of the group of subjects obtained in the subject information acquisition process.

[0012] According to the attribute estimation method of the second invention, limited survey data obtained by narrowing down the survey data to survey data having specific attributes is used to perform weighted aggregation, thereby estimating and assigning attributes other than the specific attributes possessed by the survey data to a group of subjects.

[0013] In this way, according to the attribute estimation method of the second invention, even when there is insufficient attribute information for a group of subjects, it is possible to estimate the attributes of the group of subjects by using limited survey data narrowed down to survey data having specific attributes.

[0014] The attribute estimation method of the third invention is the method of the first invention, The attribute estimation process is characterized in that it limits the results of survey items other than the known attributes in the survey data to a portion of the known attributes, and performs weighted aggregation using the survey data acquired in the survey data acquisition process to match the limited portion of known attributes of the set of subjects acquired in the subject information acquisition process, thereby estimating the results of survey items other than the known attributes in the survey data as unknown attributes of the set of subjects.

[0015] According to the attribute estimation method of the third invention, only characteristic attributes among the known attributes possessed by a group of subjects are focused on, and weighted aggregation is performed using survey data (or, if necessary, limited survey data narrowed down to survey data having specific attributes corresponding to the characteristic attributes) to match the characteristic attribute ratio, thereby estimating and assigning attributes (other than the characteristic attributes) possessed by the survey data to the group of subjects.

[0016] In this way, according to the attribute estimation method of the third invention, even when there is insufficient attribute information for a group of subjects, it is possible to estimate the attributes of the group of subjects based on characteristic attributes that feature some of the known attributes.

[0017] An attribute estimation method of a fourth invention is any one of the first to third inventions, an affinity determination step of determining an affinity of the group of subjects to an event related to the unknown attribute of the group of subjects estimated by the attribute estimation step; The present invention is characterized by carrying out the following.

[0018] According to the attribute estimation method of the fourth invention, the unknown attributes of a group of subjects estimated by the attribute estimation process are used to determine affinity, such as whether the group of subjects can accept events related to the unknown attributes or whether they are compatible.

[0019] In this way, according to the attribute estimation method of the fourth invention, even when there is insufficient attribute information about a group of subjects, it is possible to estimate the attributes of the group of subjects and predict how they will respond to events related to the unknown attributes.

[0020] An attribute estimation system of a fifth aspect of the present invention is an attribute estimation system that estimates unknown attributes of a set of subjects, the system comprising: a subject information acquisition unit that acquires known attributes of the subjects for the group of subjects; a survey data acquisition unit that acquires survey data including survey items corresponding to the known attributes; an attribute estimation unit that estimates results of survey items other than the known attributes in the survey data as unknown attributes of the group of subjects by performing weighted aggregation so that the known attributes of the survey data acquired by the survey data acquisition unit match the known attributes of the group of subjects acquired by the subject information acquisition unit; The present invention is characterized by comprising:

[0021] According to the attribute estimation system of the fifth invention, the system focuses only on the known attributes possessed by a group of subjects, and by weighting and aggregating the known attributes in the survey data to match the proportions of the known attributes, it estimates and assigns attributes other than the known attributes possessed by the survey data to the group of subjects.

[0022] In this way, according to the attribute estimation system of the fifth aspect of the present invention, even when attribute information of a group of subjects is insufficient, attributes of the group of subjects can be estimated. [Brief explanation of the drawings]

[0023] [Figure 1] 1 is a system configuration diagram showing a configuration of an attribute estimation system according to an embodiment of the present invention. [Figure 2] FIG. 2 is an explanatory diagram showing the processing content of the attribute estimation system of FIG. 1. [Figure 3A] FIG. 2 is an explanatory diagram showing the processing content of the attribute estimation system of FIG. 1. [Figure 3B] FIG. 2 is an explanatory diagram showing the processing content of the attribute estimation system of FIG. 1. DETAILED DESCRIPTION OF THE INVENTION

[0024] An attribute estimation system according to one embodiment of the present invention will be described with reference to FIG.

[0025] As shown in Figure 1, the attribute estimation system 10 is a system that estimates unknown attributes possessed by a group of subjects, and includes a subject information acquisition unit 11, a survey data acquisition unit 12, an attribute estimation unit 13, and an affinity determination unit 14.

[0026] Here, the target person is a user for whom only limited attributes are acquired (a user for whom only a small profile is acquired).

[0027] The subject information acquisition unit 11 acquires known attributes of a set of subjects from a subject information database DB1. The subject information database DB1 may be configured inside the attribute estimation system 10 or may be configured by an external server or the like.

[0028] The survey data acquisition unit 12 acquires survey data including survey items corresponding to known attributes of a group of subjects (the survey items do not need to be identical, as long as they correspond in some way) from the survey data DB 2. Note that the survey data DB 2 may be configured inside the attribute estimation system 10 or may be configured by an external server or the like.

[0029] The attribute estimation unit 13 performs weighted aggregation so that the known attributes of the group of subjects in the survey data acquired by the survey data acquisition unit 12 match the known attributes of the group of subjects acquired by the subject information acquisition unit 11, thereby estimating the results of survey items other than the known attributes in the survey data as unknown attributes of the group of subjects.

[0030] The affinity determination unit 14 predicts, from the unknown attributes of the group of subjects estimated by the attribute estimation unit 13, how the group of subjects will respond to an event related to the unknown attributes.

[0031] The above is the system configuration of the attribute estimation system 10. In the above configuration, each of the processing units 11 to 14 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 unit (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 14 may be configured with other servers (external servers), and the attribute estimation system 10 may be realized by distributed processing.

[0032] Next, the processing content (the attribute estimation method of the present invention) by the attribute estimation system 10 will be described in detail.

[0033] First, in the attribute estimation system 10, the subject information acquisition unit 11 acquires known attributes of a group of subjects from the subject information database DB1 (STEP 11, which corresponds to the subject information acquisition step of the present invention).

[0034] In this embodiment, the set of subjects is, for example, readers of novel A, and only gender has been acquired as a known attribute from a reader questionnaire or the like.

[0035] Next, in the attribute estimation system 10, the survey data acquisition unit 12 acquires survey data from the survey data database DB2 (STEP 12, which corresponds to the survey data acquisition step of the present invention).

[0036] Here, the survey data acquired in this embodiment is consumer survey data, which is a wide-ranging questionnaire survey related to daily life, including at least gender, which is a known attribute, as a survey item.

[0037] The survey data may be media contact data obtained by investigating the contact status between users and media, in addition to or instead of the consumer survey data.

[0038] Next, the attribute estimation system 10 estimates the results of survey items other than gender, which is a known attribute in the survey data, as unknown attributes of the group of subjects by performing weighted aggregation so that gender, which is a known attribute in the survey data acquired in STEP 12, matches the known attributes of the group of subjects acquired in STEP 11 (STEP 13 corresponds to the attribute estimation process of the present invention).

[0039] For example, in the case shown in FIG. 2, for the set of subjects who are readers of novel A obtained in STEP 11, the proportion of gender, which is a known attribute, is 37% male and 63% female.

[0040] Meanwhile, the gender breakdown of the survey data obtained in STEP 12 was 51.5% male and 48.5% female, and the percentage of people who "care about what they wear" in the lifestyle awareness section of the survey data was 51.8% male and 74.7% female, for an overall figure of 62.9%. Note that in Figure 2, N is the denominator and ON is the numerator.

[0041] At this time, the attribute estimation unit 13 calculates weighting values ​​(reduced values) for men as 0.7184 and women as 1.2990 to convert the gender ratios of the survey data, 51.5% men and 48.5% women, into the reader data, 37% men and 63% women, respectively, and performs weighting aggregation by multiplying the calculated weighting values ​​by the aggregated value of the lifestyle consciousness item "I pay attention to what I wear."

[0042] According to this weighted-back calculation, it is estimated that 66.2% of readers of Novel A "pay attention to what they wear" overall, and the 62.9% of the total in the original survey data can be assigned the unknown attribute of "pay attention to what they wear" in a way that is tailored to readers of Novel A.

[0043] Regarding lifestyle consciousness in the survey data, it is possible to add not only "paying attention to what you wear" but also attributes such as "going to see plays," "going to events," and "buying merchandise," along with the percentage (in percentages) of each.

[0044] Similarly, if weighted aggregation is performed using media contact data obtained by investigating user contact with media as survey data, it is possible to assign attributes (profiles) to readers of Novel A, such as "which media, at what time, and in what location they are in contact with," and this can be used for media planning. In other words, this can be used to plan the media (television, newspapers, magazines, internet, etc.) to advertise in order to efficiently deliver advertisements to target customers, as well as the timing and method of exposure.

[0045] In addition, the attribute estimation unit 13 may use limited survey data, which is obtained by narrowing down the survey data obtained in STEP 12 to survey data having specific attributes, to perform weighted aggregation to match the attributes of the group of subjects obtained in STEP 11, thereby estimating the results of survey items in the survey data other than the known attributes as unknown attributes of the group of subjects.

[0046] For example, the survey data obtained in STEP 12 may be narrowed down to "consumers who regularly read manga" and then weighted using the narrowed down survey data. This allows the attributes of a group of subjects to be estimated using limited survey data narrowed down to survey data with specific attributes.

[0047] In addition, the attribute estimation unit 13 may limit the known attributes acquired in STEP 11 to a portion thereof, and use the survey data acquired in STEP 12 (or limited survey data obtained by narrowing down the survey data to survey data having specific attributes, as described above) to perform weighted aggregation to match the limited portion of the known attributes, thereby estimating the results of survey items other than the known attributes in the survey data (limited survey data) as unknown attributes of the group of subjects.

[0048] For example, as shown in Figure 3A, the known attributes obtained in STEP 11 for the set of subjects who are readers of Novel A are as follows: the age distribution is 15% for teens, 20s, 30s, 40s, and 50s, and 25% for those in their 60s.

[0049] On the other hand, the age breakdown of the survey data obtained in STEP 12 was 9% for teenagers, 16% for people in their 20s, 18% for people in their 30s, 21% for people in their 40s and 50s, and 15% for people in their 60s.Regarding other survey items in the survey data, such as lifestyle consciousness, the percentage of people who "pay attention to what they wear" was 64.6% for teenagers, 68.2% for people in their 20s, 61.9% for people in their 30s, 61.0% for people in their 40s, 59.0% for people in their 50s, and 65.7% for people in their 60s.

[0050] In this case, as shown in FIG. 3B, when known attributes are used as they are, the attribute estimation unit 13 first calculates the weight values ​​(reduced values) to be 1.7050 for teens, 0.9490 for 20s, 0.8290 for 30s, 0.7080 for 40s, 0.7210 for 50s, and 1.6340 for 60s based on the age distribution in the survey data, which is 9% for teens, 16% for 20s, 18% for 30s, 21% for both 40s and 50s, and 15% for 60s, in the reader data, which are 15% for teens, 20s, 30s, 40s, and 50s, and 25% for 60s, and then performs weighted aggregation by multiplying the calculated weight values ​​by the aggregated value of the lifestyle consciousness item "I care about what I wear."

[0051] If known attributes are used as they are in this way, then according to such weighted aggregation, it is estimated that 63.6% of readers of Novel A overall "pay attention to what they wear," and the 62.9% of the total in the original survey data can be assigned the unknown attribute "pay attention to what they wear" in a way that is tailored to the readers of Novel A.

[0052] On the other hand, when the attribute estimation unit 13 first uses only the teens, twenties, and thirties as part of the known attributes, it performs weighted aggregation by multiplying the aggregated value of the lifestyle consciousness "I pay attention to what I wear" by the weighted values ​​(reduced values) of 1.7050 for teens, 0.9490 for twenties, and 0.8290 for thirties.

[0053] In this way, when using only teenagers, people in their twenties, and people in their thirties as part of the known attributes, such weighted aggregation estimates that 64.9% of readers of Novel A in the specific age group of teens to thirties "care about what they wear," and the 62.9% of the total in the original survey data can be assigned the unknown attribute of "care about what they wear" in a way that is tailored to readers of Novel A in the specific age group.

[0054] In the above-described embodiment, the case where a specific age group is used as part of the known attributes has been described, but the present invention is not limited to this. For example, if the known attribute acquired in STEP 11 is gender, the known attribute may be limited to male only, and weighted aggregation may be performed using the survey data acquired in STEP 12 for only males to assign the attribute.

[0055] In this way, by limiting the attributes to a portion of known attributes and then performing weighted aggregation using survey data (or, if necessary, limited survey data narrowed down to survey data having the specific attributes) so as to match the limited attribute ratios, it is possible to estimate and assign attributes (other than the limited attributes) possessed by the survey data to a group of subjects. This makes it possible to focus on a portion of known attributes and, using them as features, estimate the attributes possessed by a group of subjects based on specific attributes.

[0056] Next, the affinity determination unit 14 of the attribute estimation system 10 determines the affinity of the group of subjects to an event related to the unknown attribute of the group of subjects estimated in STEP 13 (STEP 14 corresponds to the affinity determination process of the present invention).

[0057] Specifically, the affinity determination unit 14 performs the determination by replacing the proportion of unknown attributes with the affinity of events related to the unknown attributes.

[0058] For example, if a high percentage of readers of novel A are inferred to have the unknown attribute "often go to the movies," it is determined that they have a high affinity to the event related to the unknown attribute, that is, to have novel A made into a movie, in STEP 13. On the other hand, if a low percentage of readers of novel A are inferred to have the unknown attribute "often watch anime," it is determined that they have a low affinity to the event related to the unknown attribute, that is, to have novel A made into an anime.

[0059] The attribute estimation system and attribute estimation method according to this embodiment have been described above, and it is possible to estimate attributes of a group of subjects even when attribute information of the group of subjects is insufficient.

[0060] In this embodiment, the group of subjects is described as readers of novel A, and the only known attributes that can be obtained are gender and its proportion. However, this is not limited to this, and if there are attributes such as gender, age, hobbies, and annual income, unknown attributes can be assigned, including their proportions, by performing weighted aggregation of the survey data to match these. [Explanation of symbols]

[0061] 10...attribute estimation system, 11...subject information acquisition unit, 12...survey data acquisition unit, 13...attribute estimation unit, 14...affinity determination unit, DB1...subject information database, DB2...survey data database.

Claims

1. An attribute estimation method for estimating unknown attributes of a group of subjects, the method comprising: a subject information acquisition step of acquiring known attributes of the subjects for the set of subjects; a survey data acquisition step of acquiring survey data including survey items corresponding to the known attributes; an attribute estimation step of estimating results of survey items other than the known attributes in the survey data as unknown attributes of the group of subjects by performing weighted aggregation so that the known attributes of the survey data acquired in the survey data acquisition step match the known attributes of the group of subjects acquired in the subject information acquisition step; An attribute estimation method, characterized by performing the following.

2. 2. The attribute estimation method according to claim 1, The attribute estimation method is characterized in that the attribute estimation step uses limited survey data obtained by narrowing down the survey data acquired in the survey data acquisition step to survey data having specific attributes, and performs weighted aggregation to match the attributes of the group of subjects acquired in the subject information acquisition step, thereby estimating the results of survey items in the survey data other than the known attributes as unknown attributes of the group of subjects.

3. 2. The attribute estimation method according to claim 1, The attribute estimation method is characterized in that the attribute estimation step limits the results of survey items in the survey data other than the known attributes to a portion of the known attributes, and performs weighted aggregation using the survey data acquired in the survey data acquisition step to match the limited portion of known attributes of the group of subjects acquired in the subject information acquisition step, thereby estimating the results of survey items in the survey data other than the known attributes as unknown attributes of the group of subjects.

4. 4. The attribute estimation method according to claim 1, further comprising: an affinity determination step of determining an affinity of the group of subjects to an event related to the unknown attribute of the group of subjects estimated by the attribute estimation step; An attribute estimation method, characterized by performing the following.

5. An attribute estimation system that estimates unknown attributes of a group of subjects, the system comprising: a subject information acquisition unit that acquires known attributes of the subjects for the group of subjects; a survey data acquisition unit that acquires survey data including survey items corresponding to the known attributes; an attribute estimation unit that estimates results of survey items other than the known attributes in the survey data as unknown attributes of the group of subjects by performing weighted aggregation so that the known attributes of the survey data acquired by the survey data acquisition unit match the known attributes of the group of subjects acquired by the subject information acquisition unit; An attribute estimation system comprising:

Images