Multi-modal Explanations

In our dataset, the users can select both textual and visual explanations according to their preferences for each video.

Multi-aspect Explanation Ground Truth

In our dataset, we provide the explanation ground truths from the perspectives of persuasiveness, informativeness and satisfaction.

Real User Labeled Dataset

In our dataset, the ground truths annotators are exactly the people who produce the user-item interactions in the dataset.

Comprehensive User Information

In our dataset, we have collected user profiles like the age, gender, income, hobby and physiology information.