Infusing Knowledge from Wikipedia to Enhance Stance Detection
This work addresses the problem of stance detection for social media and online debates by enhancing model performance with external knowledge, representing an incremental improvement.
The paper tackled the challenge of stance detection when models lack background knowledge about targets by infusing Wikipedia knowledge into stance encoding, resulting in WS-BERT significantly outperforming state-of-the-art methods on multiple benchmark datasets.
Stance detection infers a text author's attitude towards a target. This is challenging when the model lacks background knowledge about the target. Here, we show how background knowledge from Wikipedia can help enhance the performance on stance detection. We introduce Wikipedia Stance Detection BERT (WS-BERT) that infuses the knowledge into stance encoding. Extensive results on three benchmark datasets covering social media discussions and online debates indicate that our model significantly outperforms the state-of-the-art methods on target-specific stance detection, cross-target stance detection, and zero/few-shot stance detection.