We wrote about a Semi-Supervised Algorithm for Sarcasm Identification (SASI) AKA a sarcasm detector a few years back, but there's a new one out and it's just sooo much cooler.
So, here's the scoop: Spotter is a French company that developed an algorithm-based analytics tool that makes reports for companies wishing to track their online reputation. So if a company is concerned about how they're being perceived — or just wants to know how consumers feel about their products or services, Spotter might come in and monitor customer web comments.
Maybe you've had a delayed flight before and tweeted something like this: 'thanks Air France for getting us into London two hours late.' Spotter's algorithm picks up on the heavy sarcasm here so that analysts don't have to sit around reading whiny complaints all day. It then distills the information for the company to review.
Richard May from Spotter's UK sales team said they only evaluate "publicly available" comments and use linguistics, semantics and heuristics to determine customer sentiment. They claim that their automated system gets it right 80% of the time (that's up from 50% five years ago). The algorithms works in 29 languages and start at roughly $1,500 a month, which seems so affordable to us.