As Facebook continues to battle allegations of voter fraud and the manipulation of the 2016 US election, it seems that trust in the ubiquitous social media platform could be flagging.
Behind the headlines, however, Facebook are continuing to file patents to increase its ability to utilise personal data for advertising purposes. On 15 November, a new patent application from Facebook was published, showing technology that would make it simpler for the platform to target families and extended connections with advertising.
Detailed identification tools
The application was filed on 10 May and describes an algorithm that can identify different elements in photos. Details covered include faces, shapes and other specifics. Facebook aims to harvest this data and use it to build a profile of the whole household.
If they go ahead with the system, it would work with the family targeting programme the network announced in 2017. This already allows Facebook to analyse information to work who lives with the user. By checking the relationships users list on their Facebook profile, analysing surnames and locations, event check-ins and shared life events it decides who lives in groups and households.
The new patent describes a system that would take this even further. It talks about an extremely sophisticated way of mining data, by cross-referencing tagged photos, descriptions typed by the user, the user’s IP address, the number of Facebook user also posting from that IP address and possibly even more details. Facebook will use this to work out who lives in a household and will then categorise the householders into different demographics. These are based on age, gender, socioeconomic status and more.
Piecing together household members
What does this mean? Essentially, that Facebook may be able to work out who is in your household and family even if you never mention them on the platform. An example given by Facebook describes a man who posts lots of pictures of two women who repeatedly appear in either his photos or those his friends tag him in.
One of the pictures includes a single young girl, described by the user as ‘my angel’. Facebook’s proposed algorithms would use this to work out that this household comprises three people, one man and two women. It would further deduce that the women are most likely his wife and daughter.
While this deduction may or may not be accurate, it would still allow advertisers to target the user with ads aimed at those demographics. This applies even if he had never listed his wife and daughter on his profile at all, or that they don’t even use the platform themselves.
The claims in the patent are very broad, making it difficult to know how specific the system will be. It could be very detailed, and involve cross-referencing faces, or cover a more comprehensive level of data by analysing photographs for other evidence.
Facebook applied for an earlier patent that talked about using smartphone microphones to work out the TV shows users watch and use that to target ads. This was published in June 2018 and shows just how many ways the platform is trialling to mine user data.
In the latest patent, Facebook describes ‘demographic composition’, but doesn’t explain what this covers. It does describe targeting users based on the size of their household, the characteristics of the household members, any shared interests and which household members use electronic devices.
Facebook heavily advertises and emphasises how it ‘brings families together’. Its recently launched Portal home video chat platform is aimed at connecting family members. This smart display device features a microphone that’s always listening and a camera that’s always watching. It also features advanced AI tracking in the ‘smart camera’, which identifies users with a system called ‘2D Pose’ rather than facial recognition.
The company has said that it may use call and app data from Portal users to target advertising on other platforms, which further complicates things for those concerned about data privacy.
The newest patent covers photographs posted by users, rather than any content from private messaging systems or videos. And, of course, we mustn’t forget that many tech companies file patents that never come to fruition. While the latest patent suggests that Facebook more than likely could use your Bonfire Night photos to target you with Christmas sales, we don’t know whether it actually will.
About Dawn Ellmore Employment
Dawn Ellmore Employment was incorporated in 1995 and is a market leader in intellectual property and legal recruitment.