Event Process Typing via Hierarchical Optimal Transport

Understanding intention behind event processes in texts is important to many applications. One challenging task this line process typing, which aims tag the with one action label and object describing overall of likely affects respectively. To tackle task, existing methods mainly rely on matching level representation, ignores two characteristics: Process Hierarchy Label Hierarchy. In paper, we propose a Hierarchical Optimal Transport (HOT) method address above problem. Specifically, first explicitly extract hierarchy hierarchy. Then HOT optimally matches types Experimental results show that our model outperforms baseline models, illustrating effectiveness model.