Discovering Highly Influential Shortcut Reasoning: An Automated Template-Free Approach
This addresses the issue of irrational inference processes in NLP models, improving robustness for applications like text classification and reasoning, though it is incremental as it builds on prior shortcut identification work.
The paper tackles the problem of identifying shortcut reasoning in NLP models, which degrades robustness, by proposing a novel method that quantifies severity using out-of-distribution data and discovers both known and unknown shortcuts in Natural Language Inference and Sentiment Analysis tasks.
Shortcut reasoning is an irrational process of inference, which degrades the robustness of an NLP model. While a number of previous work has tackled the identification of shortcut reasoning, there are still two major limitations: (i) a method for quantifying the severity of the discovered shortcut reasoning is not provided; (ii) certain types of shortcut reasoning may be missed. To address these issues, we propose a novel method for identifying shortcut reasoning. The proposed method quantifies the severity of the shortcut reasoning by leveraging out-of-distribution data and does not make any assumptions about the type of tokens triggering the shortcut reasoning. Our experiments on Natural Language Inference and Sentiment Analysis demonstrate that our framework successfully discovers known and unknown shortcut reasoning in the previous work.