Detecting incongruent news headlines with auxiliary textual information

Joonwon Jang, Yoon-Sik Cho, Minju Kim, Misuk Kim

Abstract: Highlights•We proposed a model with an attention mechanism for predicting incongruent headlines.•We generated incongruent headline dataset from a real-world news.•We found subtitle and image caption achieves better predictions.•We further investigated the impact of the components in our proposed model.•Our results reveals the possibility of real-world applications.