A Multi-Type Multi-Span Network for Reading Comprehension that Requires Discrete Reasoning
This addresses the challenge of improving reading comprehension for tasks involving diverse answer types and discrete reasoning, representing a strong specific gain rather than a foundational advancement.
The paper tackles the problem of reading comprehension models performing poorly in realistic scenarios requiring discrete reasoning and multiple answer types, and introduces the Multi-Type Multi-Span Network (MTMSN) which achieves 79.9 F1 on the DROP dataset, setting a new state-of-the-art result.
Rapid progress has been made in the field of reading comprehension and question answering, where several systems have achieved human parity in some simplified settings. However, the performance of these models degrades significantly when they are applied to more realistic scenarios, such as answers involve various types, multiple text strings are correct answers, or discrete reasoning abilities are required. In this paper, we introduce the Multi-Type Multi-Span Network (MTMSN), a neural reading comprehension model that combines a multi-type answer predictor designed to support various answer types (e.g., span, count, negation, and arithmetic expression) with a multi-span extraction method for dynamically producing one or multiple text spans. In addition, an arithmetic expression reranking mechanism is proposed to rank expression candidates for further confirming the prediction. Experiments show that our model achieves 79.9 F1 on the DROP hidden test set, creating new state-of-the-art results. Source code\footnote{\url{https://github.com/huminghao16/MTMSN}} is released to facilitate future work.