Good call. I will add the entire article to my training data. Just running it as is, wow, the headline came out to be 94.40655708312988 positive for clickbait! (I am using a modified keras/tensorflow model trained on clickbait headlines.)