Sarcastically

Abstract

Sarcasm is one of the most well-known but easily misunderstood parts of the English language. While sarcasm is understood well to a vast majority, the uniqueness of sarcasm is that no one thinks about it the same way, when it is understood by the majority, it does its job but when it isn’t is when questions start to arise in terms of meaning, understanding and so on. Most research on sarcasm has come in the form of a textual understanding of it using context from previous statements or blocks of statements and more recent research has led to the discovery of supplemental forms of associative data such as video and images which help to understand a machine’s understanding of sarcasm. Sound plays just as much a part as other social and visual cues which indicate sarcasm and will be the focus of the current research project. Sarcastically attempts to make use of this area of sarcasm to better classify sarcasm by making using of vocal cues accompanied by basic text sarcasm detection to better evaluate the results of machines understanding sarcasm. It will make use of a multi-model sarcasm algorithm which contains not only a text evaluator but also an audio and hybrid evaluation model that will be cross-compared and extracted to get the highest possible result. For this research, the core content has been based on acted-sarcasm for the purpose of generating generic sarcasm rather than advanced sarcasm which might take considerably longer time to prepare for. As such, Sarcastically’s job is to evaluate the benefits or the losses of using audio in accompaniment with text to justify that audio is more or less important to understanding sarcasm as other visual or social cues.