Noam Chomsky at SemEval-2023 Task 4: Hierarchical Similarity-aware Model for Human Value Detection

Sumire Honda, Sebastian Wilharm

Abstract: This paper presents a hierarchical similarity-aware approach for the SemEval-2023 task 4 human value detection behind arguments using SBERT. The approach takes similarity score as an additional source of information between the input arguments and the lower level of labels in a human value hierarchical dataset. Our similarity-aware model improved the similarity-agnostic baseline model, especially showing a significant increase in or the value categories with lowest scores by the baseline model.