RobustLR: Evaluating Robustness to Logical Perturbation in Deductive Reasoning
This work addresses the need for better evaluation methods in AI for logical reasoning over natural language, though it is incremental as it focuses on assessment rather than new model development.
The authors tackled the problem of evaluating whether transformers truly understand logical semantics in deductive reasoning by proposing RobustLR, a suite of datasets to test robustness to logical perturbations, and found that models like RoBERTa and T5 perform inconsistently, especially struggling with logical negation and disjunction.
Transformers have been shown to be able to perform deductive reasoning on a logical rulebase containing rules and statements written in English natural language. While the progress is promising, it is currently unclear if these models indeed perform logical reasoning by understanding the underlying logical semantics in the language. To this end, we propose RobustLR, a suite of evaluation datasets that evaluate the robustness of these models to minimal logical edits in rulebases and some standard logical equivalence conditions. In our experiments with RoBERTa and T5, we find that the models trained in prior works do not perform consistently on the different perturbations in RobustLR, thus showing that the models are not robust to the proposed logical perturbations. Further, we find that the models find it especially hard to learn logical negation and disjunction operators. Overall, using our evaluation sets, we demonstrate some shortcomings of the deductive reasoning-based language models, which can eventually help towards designing better models for logical reasoning over natural language. All the datasets and code base have been made publicly available.