I think you have met Mark (if no, read the full story here or just note that Mark is our artificial intelligence project learning to understand tv content and even to help us track propaganda threats). His growing process, and the challenges we face as his “parents” are worth keeping the record of, so we decided to start a journal. Let this first entry mark a beginning of a remarkable biography.
Since November 2016, Mark has been watching TV and burying us under tons of data. Part of the data was easy to interpret, as it was done in accordance with the media industry standards such as airings, ratings, the share of voice. Another part, which concerns objects and scenes found, people and emotions identified, is yet to be processed and formalized.
Mark is a clever guy, and like all clever guys, he tries to make his life easier. It took an effort to explain to him that some people are smoking and some – just holding hands near their faces. It went a bit faster with explaining that a microphone is not a huge cigar.
We were astonished to learn from Mark that a TV news studio looks exactly like a bathroom. After some fierce brainwork, we came up with a solution we could be proud of: if the human on the screen is dressed (usually wearing a suit), it is most probably not a bathroom.
Regardless of what makes you feel better – being surrounded by happy people or observing those who have a worse time than you, – TV can give it to you. Just find the right channel.
Mark recognizes visual expressions of human emotions, and his findings show that some channels have a more negative visual content than the others. Apart from emotions, Mark identifies objects, and the “negative” ones like “a ruin”, “a military person”, or “an army tank” are usually central in the news, films, and cartoons aired by the channel.
Such focus on negativity may be a coincidence, but – hello to the conspiracy church! – it may also be deliberate if the channel is state-sponsored. “Conspiracy” seems even more real when one knows that domestic content differs from what is broadcasted to the audience abroad.
Another interesting story is animals. Like the good old PC vs Mac fight in the IT sphere, the Cat vs Dog dispute is going on among animal lovers (needless to say, both cases involve time-worn arguments and a high degree of subjectivity).
There are multiple ways to read the Cats vs Dogs chart Mark has kindly provided us with. It can indicate any of the following:
media industry thinks there are more dog lovers watching TV;
cats sell better, therefore one needs fewer cat airings to get the same effect as with dogs;
cats do not sell anymore (like sex, according to recent findings);
nothing, as statistics, concern something else.
We are still improving Mark’s abilities to recognize objects, scenes, and people – TV & VOD content, pictures from social media, and even the contents of our mobile phones, which Mark has been given a chance to play with, – for it is a never-ending process. But… Mark has already been taught to recognize Latvian politicians and parties, which is an invaluable feature now when municipal elections are near and new campaigning regulations need to be observed!