<iframe src="https://exploratory.io/viz/kanaugust/8641797274167691?embed=true" frameborder="0" width="100%" height="100%" > </iframe>
I have calculated TF-IDF with L1 weight for each word of all the tweets of Clinton and Trump over the last 10 days. And then, I have scored the sentiment for each word and filtered only the words with sentiment values. Lastly, I kept only the top 10 words for each candidate based on the TF-IDF values.