Methods and apparatus to generate audience metrics for connected television
Methods, apparatus, systems, and articles of manufacture are disclosed to generate audience metrics for connected television. An example system includes at least one memory, programmable circuitry, and instructions to cause the programmable circuitry to obtain media access data corresponding to connected television media and a user identifier corresponding to a media access device, generate, using a machine learning model, probability values for corresponding audience demographics in a household composition corresponding to the user identifier, the probability values indicative of likelihoods that corresponding ones of the audience demographics are accessing the connected television media, determine a person-level characteristic based on the probability values of the audience demographics, the person-level characteristic corresponding to an audience member of the connected television media, and assign the media access data to the person-level characteristic.
1 . A computing system comprising:
a processor; and
at least one memory storing instructions that, when executed by the processor, cause the computing system to perform operations comprising:
obtaining a user identifier and media access data corresponding to a media presentation device, wherein the media access data characterizes media accessed by one or more users through the media presentation device;
obtaining, based on the user identifier, a device type of the media presentation device associated with the media access data, wherein the device type comprises a single-user device type or a multi-user device type;
providing as an input to a machine learning model, the media access data, the user identifier, and the device type;
processing the input using the machine learning model to generate a set of probability values for corresponding demographic categories in a household composition associated with the user identifier, wherein the probability values are numerical values, and wherein the probability values are indicative of probabilities that corresponding ones of the demographic categories were accessing the media characterized by the media access data; and
transmitting the set of probability values and the user identifier to a remote server, wherein transmitting is in response to determining whether the probability values are above a threshold.
2 . The system of claim 1 , wherein the operations further comprise:
selecting a target probability value from the set of probability values; and
generating a person-level characteristic based on the target probability value, wherein the person-level characteristic corresponds to a particular user associated with the media presentation device.
3 . The system of claim 2 , wherein the target probability value is the largest probability value from the set of probability values.
4 . The system of claim 2 , wherein the person-level characteristic comprises an age or a gender.
5 . The system of claim 2 , wherein the operations further comprise:
transmitting the person-level characteristic and the user identifier to the remote server.
6 . The system of claim 1 , wherein the machine learning model comprises a neural network configured to generate a respective probability value for each demographic category.
7 . The system of claim 1 , wherein the operations further comprise:
selecting one or more target probability values from the set of probability values, wherein the one or more target probability values comprise numerical values that are above the threshold; and
assigning the demographic categories associated with the one or more target probability values to the user identifier to generate household composition parameters for the user identifier.
8 . The system of claim 7 , wherein the set of probability values is a first set of probability values, and wherein the operations further comprise:
obtaining real-time media access data corresponding to the media presentation device and the user identifier; and
processing the real-time media access data, the device type, and the user identifier using the machine learning model to generate a second set of probability values that characterizes media accessed by the corresponding demographic categories in real-time,
wherein the machine learning model is conditioned on the household composition parameters.
9 . The system of claim 8 , wherein the operations further comprise:
transmitting the second set of probability values and the user identifier to the remote server in real-time.
10 . A non-transitory machine readable storage medium comprising instructions that, when executed, cause a processor to perform operations comprising:
obtaining a user identifier and media access data corresponding to a media presentation device, wherein the media access data characterizes media accessed by one or more users through the media presentation device;
obtaining, based on the user identifier, a device type of the media presentation device associated with the media access data, wherein the device type comprises a single-user device type or a multi-user device type;
providing as an input to a machine learning model, the media access data, the user identifier, and the device type;
processing the input using the machine learning model to generate a set of probability values for corresponding demographic categories in a household composition associated with the user identifier, wherein the probability values are numerical values, and wherein the probability values are indicative of probabilities that corresponding ones of the demographic categories were accessing the media characterized by the media access data; and
transmitting the set of probability values and the user identifier to a remote server, wherein transmitting is in response to determining whether the probability values are above a threshold.
11 . The non-transitory machine readable storage medium of claim 10 , wherein the operations further comprise:
selecting a target probability value from the set of probability values; and
generating a person-level characteristic based on the target probability value, wherein the person-level characteristic corresponds to a particular user associated with the media presentation device.
12 . The non-transitory machine readable storage medium of claim 11 , wherein the target probability value is the largest probability value from the set of probability values.
13 . The non-transitory machine readable storage medium of claim 11 , wherein the person-level characteristic comprises an age or a gender.
14 . The non-transitory machine readable storage medium of claim 11 , wherein the operations further comprise:
transmitting the person-level characteristic and the user identifier to the remote server.
15 . The non-transitory machine readable storage medium of claim 10 , wherein the machine learning model comprises a neural network configured to generate a respective probability value for each demographic category.
16 . The non-transitory machine readable storage medium of claim 10 , wherein the operations further comprise:
selecting one or more target probability values from the set of probability values, wherein the one or more target probability values comprise numerical values that are above the threshold; and
assigning the demographic categories associated with the one or more target probability values to the user identifier to generate household composition parameters for the user identifier.
17 . The non-transitory machine readable storage medium of claim 16 , wherein the set of probability values is a first set of probability values, and wherein the operations further comprise:
obtaining real-time media access data corresponding to the media presentation device and the user identifier; and
processing the real-time media access data, the device type, and the user identifier using the machine learning model to generate a second set of probability values that characterizes media accessed by the corresponding demographic categories in real-time, wherein the machine learning model is conditioned on the household composition parameters.
18 . The non-transitory machine readable storage medium of claim 17 , wherein the operations further comprise:
transmitting the second set of probability values and the user identifier to the remote server in real-time.
19 . A method comprising:
obtaining a user identifier and media access data corresponding to a media presentation device, wherein the media access data characterizes media accessed by one or more users through the media presentation device;
obtaining, based on the user identifier, a device type of the media presentation device associated with the media access data, wherein the device type comprises a single-user device type or a multi-user device type;
providing as an input to a machine learning model, the media access data, the user identifier, and the device type;
processing the input using the machine learning model to generate a set of probability values for corresponding demographic categories in a household composition associated with the user identifier, wherein the probability values are numerical values, and wherein the probability values are indicative of probabilities that corresponding ones of the demographic categories were accessing the media characterized by the media access data; and
transmitting the set of probability values and the user identifier to a remote server, wherein transmitting is in response to determining whether the probability values are above a threshold.
20 . The method of claim 19 , further comprising:
selecting a target probability value from the set of probability values; and
generating a person-level characteristic based on the target probability value, wherein the person-level characteristic corresponds to a particular user associated with the media presentation device.