The software is a sophisticated machine translation system (this is a sub-field of computational linguistics that investigates the use of software to translate text or speech from one language to another). The primary aim is to interpret sarcastic statements made on social media, be they Facebook comments, tweets or some other form of digital communication. The software continues to undergo testing; the primary aim is to help people with learning difficulties or even for situations where the intention of the sender is unclear. In other words, to aid any person who might experience difficulties with interpreting sarcasm, irony and humor.
The development is from the Technion-Israel Institute of Technology Faculty of Industrial Engineering and Management and the software is called Sarcasm SIGN (sarcasm Sentimental Interpretation GeNerator). the device is based on machine translation; here the software turns sarcastic sentences into honest (non-sarcastic) ones.
As an example, the software can turn a sarcastic sentence like: “The new ‘Fast and Furious’ movie is awesome. #sarcasm” into the honest sentence, and one which is less ambiguous: “The new Fast and Furious movie is terrible.”
Behind the project is lead researcher is Professor Roi Reichart, together with Lotam Peled. Explaining the usefulness of the application, Peled states in a research note: “There are a lot of systems designed to identify sarcasm, but this is the first that is able to interpret sarcasm in written text.” The researcher adds: “We hope in the future, it will help people with autism and Asperger’s, who have difficulty interpreting sarcasm, irony and humor.” To instruct the software, which works on a form of machine learning, the researchers compiled a database of 3,000 sarcastic tweets that were tagged with #sarcasm. Whether the ability to detect ‘sarcastic’ tweets without the tag is possible is uncertain at this stage of machine learning development. This is because describing how we pick up on sarcasm is often difficult because it depends on shared knowledge, social customs and norms.
Nonetheless the researchers feel that as social media becomes more universal and, despite the predominance of videos, it remains largely text driven there is an increasing need for social media intelligence and tools to facilitate this.