MultiModal Sarcasm Detection: A Survey
Aruna B. Bhat, Aditya Chauhan · 2022 IEEE Delhi Section Conference (DELCON) · 2022
Deciphering sarcasm can be arduous at times as the literal meaning is contrary to what is intended. Sarcasm in very simple terms is a form of expression where one expresses himself/herself while trying to poke fun, take a jibe or simply make a joke as the intended meaning is in stark contrast to what has been actually said. Sarcasm detection is a trickier task as some background information or the context is required in order to get the intended meaning. In today's fast paced world where social media is scaling great heights day by day, sarcasm detection has become a need of the hour as it can have a lot of use cases and can eventually take drastic forms like bullying, harassment etc. In that case detection of sarcasm on various social media platforms is one of the important issues which needs to be handled. So far most of the work has been done using only textual data, but as per the present day scenario where users can express themselves via multiple modes of data, in order to understand the message clearly it is not enough to focus on just texts as other modes can provide important clues that aren't clear from the text. In this paper we will explore various researches in the field of sarcasm detection using multimodal data. We will analyze different datasets and go through the various methodologies proposed. In the end, we show a comparative analysis of the results from various models which use accuracy and F1-Score as metrics to measure the performance.