Cross-topic Argument Mining from Heterogeneous Sources Using Attention-based Neural Networks
This addresses the challenge of automating argument search across diverse text types for applications like information retrieval, though it is incremental in improving generalization.
The paper tackled the problem of argument mining from heterogeneous texts by proposing a new annotation scheme and attention-based neural network, achieving a 6% accuracy and 11% F-score improvement over baseline models in cross-topic experiments.
Argument mining is a core technology for automating argument search in large document collections. Despite its usefulness for this task, most current approaches to argument mining are designed for use only with specific text types and fall short when applied to heterogeneous texts. In this paper, we propose a new sentential annotation scheme that is reliably applicable by crowd workers to arbitrary Web texts. We source annotations for over 25,000 instances covering eight controversial topics. The results of cross-topic experiments show that our attention-based neural network generalizes best to unseen topics and outperforms vanilla BiLSTM models by 6% in accuracy and 11% in F-score.