This video discusses the use of active learning, deep learning, Bayesian inference, and causality in Project Feels. This project, developed by the Data Science Group at the New York Times, sought to predict how likely a given article was to evoke a range of emotions. Thus project crowdsourced data from a pool of hundreds of thousands of articles and modeled this data using a range of state-of-the-art text models. The model predictions were in turn used to create premium advertising spaces, which showed quantitative improvements in randomized control trials.
[Related Article: An Introduction to Natural Language Processing (NLP)]
