Researchers conducted a series of experiments to better understand when people are expressing sarcasm through text. By training their computers, the team was able to detect Twitter sarcasm using several factors that include keywords, hyperbole and hashtags.