All projects

Sentiment Analysis Enginein progress

Tracks what people say about a campaign online, scores each mention, groups mentions into stories, and sends anything ambiguous, like sarcasm or attacks on an opponent, to a person to check.

What it does

The engine reads public mentions of a campaign, scores how positive or negative each one is, and says how confident it is in that score. Related mentions are grouped into stories, so a team sees what's building instead of a pile of posts.

Where people step in

Software misreads politics in predictable ways. Sarcasm, attacks on an opponent, crisis words like "arrested", and reshared news all trip it up. When the engine sees one of these, it sets its confidence to zero and sends the mention to a person to review.