Decoding Political Advertising Inferring Messages and Predicting Impacts

Original article was published on Deep Learning on Medium

Decoding Political Advertising Inferring Messages and Predicting Impacts

Advertising is basically buying attention. Selling information as return, now how the information is made and color scheme, etc is what determines the QUALITY of the information. (besides the information itself!).

And there are different types of argument, some might be awareness others may be forceful. But at the end of the day, communicating information, that can have some effect on the reader/viewer/listeners feeling, that in turn will be a conversion.

This is fucking sickening, but it did happen. Racist motherfuckers, what a waste of humanity. This is exactly what I mean.

Engagement of emotions!

But now it is anger rather than the excitement of buying that fucking shitty soap. Looking through this, we need to be VERY careful about how we want to formulate.

If we want our grandsons/daughters to see the ad and have the same feeling, we should sell things very well.

And in the age of the internet as well as longer healthcare, we are going to be held accountable for a long period of time.


Understanding culture is another thing, how can we encode that information into a model. How can we even CREATE a dataset to cover all of the distribution?

This is one hell of a problem.

And how can ML be used in this, we know that it is going to be used. Not right now, but in the upcoming era. So the question is HOW?

Very difficult indeed, but if there is a software that aids this? Maybe we can?

This is a good approach, divide this problem into smaller ones that are more ‘manageable’.

So there is some work done here, and obviously classification or regression is the first task to be done. This is understandable, a lot of tutorials (or I like to call it infrastructure of ML) are based on classification or regression.