Higher-order features bring significant accuracy gains in semantic dependency parsing. However, modeling higher-order features with exact inference is NP-hard. Graph neural networks (GNNs) have been demonstrated to be an effective tool for solving NP-hard problems with approximate inference in many graph learning tasks. Inspired by the success of GNNs, we investigate building a higher-order semantic dependency parser by applying GNNs. Instead of explicitly extracting higher-order features from intermediate parsing graphs, GNNs aggregate higher-order information concisely by stacking multiple GNN layers. Experimental results show that our model outperforms the previous state-of-the-art parser on the SemEval 2015 Task 18 English datasets.
@article{arxiv.2201.11312,
title = {A Higher-Order Semantic Dependency Parser},
author = {Bin Li and Yunlong Fan and Yikemaiti Sataer and Zhiqiang Gao},
journal= {arXiv preprint arXiv:2201.11312},
year = {2022}
}