Incongruity Detection

Incongruity detection focuses on identifying discrepancies between expected and actual information, a key element in understanding humor, affect, and deception. Current research employs various approaches, including hierarchical fusion models that dynamically weigh multimodal inputs and graph neural networks that analyze relationships between textual components like news headlines and bodies. These methods are applied across diverse domains, from humor recognition to detecting deceptive news articles, highlighting the broad applicability of incongruity detection in both scientific understanding and practical applications like automated content analysis.