Effective fake news video detection using domain knowledge and multimodal data fusion on youtube

Hyewon Choi, Youngjoong Ko

Abstract: Highlights•Leveraging domain knowledge and ”hit-likes” of comments to detect fake news videos effectively.•Multi-modal encoding reflecting important characteristics of the image and text features.•Technology to detect the stance difference between title/description/video and user’s comments.•We achieve an F1-score of 0.93, which is higher than those of other comparison models in all the test datasets.