This is a plot of news sites using various tools that attempt to figure out the concepts expressed in a body of text.
The source data are headlines extracted from news site web pages. Headlines are extracted daily but these graphs are only updated once a week because this kind of text processing works better with larger data sets--a single day of headlines is likely insufficient data. The viewer can decide how useful this visualization is. One possible interpretation is that whichever news site has the smallest overall shaded area may be the most "balanced" or at least uses less emotion-laden language.
Many tools are used here but noteworthy ones are The Sentiment and Emotion Lexicons produced by the National Research Council of Canada, Natural Language Toolkit and D3.js