In classification-based event extraction, authors usually predefine event types and their corresponding argument roles. A classifier will classify the event into an event type and then classifies each entity to a predefined argument 

Carleton College

There are several ways of implementing event extraction:

- Classification-based
- Machine Reading Comprehension (MRC)-based
- Sequence labelling-based
 - Sequence-to-structure generation-based

Different implementations of event extraction

Li, Q., Li, J., Sheng, J., Cui, S., Wu, J., Hei, Y., Peng, H., Guo, S., Wang, L., Beheshti, A., & Yu, P. (2021). A Compact Survey on Event Extraction: Approaches and Applications. https://arxiv.org/abs/2107.02126.

A Compact Survey on Event Extraction: Approaches and Applications

Classification-based event extraction

In MRC-based event extraction, the event type is first determined along with the corresponding argument roles according to the event type. Next, a question schema is designed for each argument role, and a MRC model is applied to the text to extract the different event arguments

MRC-based event extraction

Sequence labelling-based event extraction is a word-level tagging classification task. Event arguments are directly matched based on word level event type extraction

Sequence labelling-based event extraction

Sequence-to-structure generation-based event extraction uniformly models all tasks in a single model and universally predicts different labels. From there, the model directly generates all arguments and their roles. An encoder-decoder structure is typically used.

Learn Before

Related