Hybrid Multimodal Sarcasm Detection System

Hybrid Multimodal Sarcasm Detection SystemSarcasm detection is a challenging problem in natural language processing because the intended meaning of a sentence is often opposite to the literal meaning of the words used. Traditional sarcasm detection systems mainly rely on textual data, which limits their ability to correctly interpret sarcastic expressions. To address this issue, the Hybrid Multimodal Sarcasm Detection System is designed to use multiple types of data such as text, images, and contextual information, making the detection process more accurate and reliable.The main aim of this system is to improve sarcasm detection by combining different data modalities through a hybrid approach. In this system, data is collected from sources like social media, where sarcasm is commonly used in the form of captions, posts, or comments along with images. The collected data undergoes preprocessing, where textual data is cleaned by removing unnecessary elements such as stop words and special characters, and image data is normalized to ensure consistency.After preprocessing, feature extraction is performed on both text and images.