Researchers develop an A.I based 'sarcasm detector' for social media

Developing a deeper understanding of consumer input from social media sites is essential for the growth of brands, but it is very work-intensive. However, a recent study by computer scientists at Central Florida University, who created a "sarcasm detector," may have made it easier.

In a social media text, a UCF crew created a technique that accurately detects sarcasm.

The team successfully taught the computer system to identify patterns that often signal sarcasm by properly selecting indicative terms in sarcasm sequences. They did so with huge data sets and also with consistency verification.

"Sarcasm in this text is the main impediment to feelings research," said Ivan Garibay, Assistant Professor of Engineering.

"Sarcasm in speech is not really easy to detect, but you might think that it's pretty difficult for a computer system to do that and do it very well. A deep learning model with multi-head focus and gated recurrent units was developed. The multi-head attention module supports the identification of essential sarcastic input cue words, and the recurrent units learn long-range dependence between the cue words to help classify the input text," he added.