CLES

• project for paper "Harvesting Events from Multiple Sources: Towards a Cross-Document Extraction Paradigm"

1. The process of collecting the CLES dataset involved several steps:

1. Firstly, we gather document-level event data from Wikipedia. The original dataset is located in "dataset/document_level_dataset".
2. Secondly, we extract events from the documents using OmniEvent as our tool to obtain the raw dataset.
3. We utilize human validation to obtain the final dataset, as described in the paper.

2. CDEE model