[ad_1]
Facebook wants to know who you live with.
The social media giant has filed a patent for a technology that would look at your family's photos to determine who else lives in your household.
In return, this would allow Facebook to broadcast targeted ads to the entire family.
Scroll down for the video
Facebook has filed a patent for a technology that would review your family's photos to determine who else lives in your household. This would allow the company to serve targeted ads.
The patent, titled "Predicting Household Demographics from Image Data," was filed on May 10 but was released only on Thursday.
It describes the use of in-depth learning algorithms trained to look for details in photos, such as faces, and reference them with textual data in the person's profile that "suggest relationships "between people in the photo.
The algorithm would also examine other available data, such as comments, captions, tags, IP addresses to form conclusions.
"For example, a trained image analysis model identifies each individual shown in the user's photos, a trained text analysis model derives information about relationships between members of the user. household from the user's profile data and tags associated with photos, "says the patent.
"The online system uses forecasts to generate more information about the user and his or her home in the online system, and to provide improved and targeted content delivery to the user and to his or her audience. home. "
The algorithm detects who is on the photo and crosses it with other data such as IP addresses, captions and tags to establish a family relationship between people in the photo.
The use of this technology would give Facebook all kinds of information about households and demographics, such as age, socio-economic data, gender and geographical location, as many as 15,000 people. essential elements for advertisers.
The algorithm is able to deduce granular information such as the number of people marked on a photo, the people in the photo and the frequency with which these people appear simultaneously on the same photo.
He also examines labels to determine family relationships, giving the example of a photo including family hashtags, Christmas, husband, wife, mother or children.
The algorithm studies the email history, tag history and even web browsing history to see if multiple people share the same IP address.
In addition, he would also use Instagram images to gather information.
The algorithm can deduce granular information such as the number of people marked on a photo, the people in the photo and the frequency with which these people appear simultaneously on the same photo.
In another example, Facebook describes a man who publishes several photos of two women.
On one of the photos, he is photographed with a girl and wears the subtitle "my angel".
The algorithm deduces that he is photographed with what is probably his daughter, while the photo of the other woman is his wife.
As a result, the algorithm can also indicate that it is a three-person household, creating a richer profile so that brands can target more relevant ads to the business. # 39; user.
This is not the first patent published by Facebook regarding ways to collect more robust demographic information from users.
Earlier this year, he patented a technology that would allow him to predict "life-changing events" such as marriage, birth and even death.
Source link
