Tribrid: Stance Classification with Neural Inconsistency Detection
This work addresses the problem of improving stance classification accuracy for social media analysis, with potential applications in fact-checking, but it is incremental as it builds on existing neural methods like BERT.
The paper tackles stance classification on social media by introducing a neural architecture that incorporates automatically generated negated perspectives to improve performance, achieving human-like performance on retained data after filtering doubtful predictions.
We study the problem of performing automatic stance classification on social media with neural architectures such as BERT. Although these architectures deliver impressive results, their level is not yet comparable to the one of humans and they might produce errors that have a significant impact on the downstream task (e.g., fact-checking). To improve the performance, we present a new neural architecture where the input also includes automatically generated negated perspectives over a given claim. The model is jointly learned to make simultaneously multiple predictions, which can be used either to improve the classification of the original perspective or to filter out doubtful predictions. In the first case, we propose a weakly supervised method for combining the predictions into a final one. In the second case, we show that using the confidence scores to remove doubtful predictions allows our method to achieve human-like performance over the retained information, which is still a sizable part of the original input.