Discrete Reasoning Templates for Natural Language Understanding
This addresses the problem of multi-part reasoning in natural language understanding for reading comprehension tasks, though it is incremental as it focuses on subtraction-based arithmetic questions.
The paper tackles the challenge of reasoning across multiple parts of a passage for reading comprehension by decomposing complex questions into simpler subquestions using single-span extraction models and predefined reasoning templates, achieving competitive state-of-the-art results on a subset of the DROP dataset with interpretability and minimal supervision.
Reasoning about information from multiple parts of a passage to derive an answer is an open challenge for reading-comprehension models. In this paper, we present an approach that reasons about complex questions by decomposing them to simpler subquestions that can take advantage of single-span extraction reading-comprehension models, and derives the final answer according to instructions in a predefined reasoning template. We focus on subtraction-based arithmetic questions and evaluate our approach on a subset of the DROP dataset. We show that our approach is competitive with the state-of-the-art while being interpretable and requires little supervision