Machine Reading Comprehension as Data Augmentation: A Case Study on Implicit Event Argument Extraction

Machine Reading Comprehension as Data Augmentation: A Case Study on Implicit Event Argument Extraction Jian Liu, Yufeng Chen, Jinan Xu Beijing Jiaotong University, School of Computer and Information Technology, China {jianliu, chenyf, jaxu}@bjtu.edu.cn Abstract Implicit event argument extraction (EAE) is a crucial document-level information extraction task that aims to identify event arguments be- yond the sentence level. Despite many efforts for this task, the lack of enough training data has long impeded the study. In this paper, we take a new perspective to address the data sparsity issue faced by implicit EAE, by bridg- ing the task with machine reading compre- hension (MRC). Particularly, we devise two data augmentation regimes via MRC, includ- ing: 1) implicit knowledge transfer, which en- ables knowledge transfer from other tasks, by building a unified training framework in the MRC formulation, and 2) explicit data aug- mentation, which can explicitly generate new training examples, by treating MRC models as an annotator. The extensive experiments have justified the effectiveness of our approach — it not only obtains state-of-the-art performance on two benchmarks, but also demonstrates su- perior results in a data-low scenario. 1 Introduction Textual event descriptions may span over multi- ple sentences. Implicit event argument extraction (EAE) (Ebner et al., 2020; Zhang et al., 2020), a crucial task for event information extraction, aims to identify event arguments beyond the sentence level. For example, in a document describing an AirstrikeMissileStrike event (shown in Figure 1), implicit EAE requires a model to recognize a global event argument “Syria", fulfilling the semantic role of PLACE. Note the argument is one-sentence away from the event trigger bombarding. One key challenge faced by implicit EAE is data sparsity — owing to the complex interdependen- cies between triggers and arguments, it is expen- sive to label training data for the task. The existing datasets, which typically contain several dozens of documents, are too small to train a model for capturing regularities underlying how event argu- … This was indeed an IS attack, rather than an accidental explosion. New satellite imagery appears to reveal extensive damage to [an airbase]Victim in [central Syria]Place. ... Q1: What is the victim in the attack event? Question Answer A1: an airbase Document Q2: Where does the attack event happen? A2: central Syria When Russian aircraft bombed a remote garrison in southeastern Syria last month, alarm bells sounded at the Pentagon and the Ministry of Defense in London. The Russians weren’t bombarding a run-of-the-mill rebel outpost, according to U.S. officials. AirstrikeMissileStrike attacker target instrument place Q1: Who is the attacker in the bombarding event? A1: Russians Q2: Where does the bombarding event take place? A2: Syria Figure 1: An example of implicit EAE (Above), and the illustration of framing implicit EAE as MRC (Bottom). ments appear in texts (Li et al., 2021). For example, even the state-of-the-art model, trained on the full corpus of RAMS (Ebner et al., 2020), attains only 5% in F1 when an event argument is two-sentence away from the trigger (Zhang et al., …

Content truncated to 500 words