Overall structure（troduction part）

paragraph 1 ：

what is EAE：

Event Argument Extraction (EAE) is a longstanding information extraction task to extract
event structures composed of arguments from unstructured text (Xiang and Wang, 2019).

Application

Event structures can serve as an intermediate semantic representation and be further used for improving
downstream tasks, including machine reading comprehension (Han et al., 2021), question answering(Costa et al., 2020), dialog system (Zhang et al., 2020), and recommendation system (Li et al., 2020).

Challenging

Despite the large performance boost by Pre-trained Language Models (PLMs), extracting complex event structures across sentences is still challenging (Ebner et al., 2020).

paragraph 2

real-life scenario

In real-world text, event structures are usually distributed in multiple sentences (Li et al., 2021).

previous method

To capture cross-sentence and multi-hop structures, Xu et al. (2022) introduces Abstract Meaning Representation (AMR) graphs to assist the model in understanding the document. Their main idea is to take AMR as additional features to enrich span representations. Xu and Huang (2022) and Wang et al. (2021) utilize AMR graphs to provide training signals via self-training and contrastive learning, respectively.