From The Washington Post
Inside the surprisingly high-stakes quest to design a computer program that ‘gets’ sarcasm online
By Caitlin Dewey
Sarcasm on the Internet, the old Web adage goes, is kind of like winking on the phone.
It’s pointless. It’s invisible. It’s almost always misunderstood.
And for linguists, pollsters, marketers, stockbrokers, law enforcement and anyone else with a vested interest in knowing what people say (and mean) online, it’s become one of modern computing’s most vexing puzzles: Could we ever teach a program to recognize sarcasm — a human quirk that even humans mess up half the time?
“Sarcasm detection is a very difficult computational problem,” stresses David Bamman, a computational linguist and an assistant professor at UC Berkeley’s School of Information. His most recent stab at solving it — sponsored by the National Science Foundation, and published earlier this year — correctly IDed sarcasm on Twitter about 85 percent of the time, which is still a long way from ideal....