Shuailong / BlamePipeline

Blame Tie Extraction Task codebase

BlamePipeline

Implementation of the AAAI 2019 paper: Who Blames Whom in a Crisis? Detecting Blame Ties from News Articles Using Neural Networks

Task

Given a news article, extract blame ties (who blames whom) between entities in the article.

Example

*An example sentence from our dataset containing a blame tie. The red/bold words are entities involved in a blame tie, and the blue/italic words are supporting evidence that the blame tie exists.*

Dataset

• Source: New York Times/Wall Street Journal/USA Today

• Time period: 2007/10– 2010/06

	USA	NYT	WSJ
days	310	736	648
articles	132	429	438
blame ties	353	787	754

Number	value	Ratio	value
# of articles	998	Average -/+ ratio	2.19
# of samples	8562	Total -/+ ratio	3.61

Models

Context Model

Results

Model	Dev F1	Test F1
Entity	61.07	60.06
Context	73.16	66.35
Combined	76.13	69.92