Learning Recurrent Span Representations for Extractive Question Answering
This work addresses the challenge of extracting arbitrary answer strings from text for reading comprehension, offering a significant but incremental improvement in a specific domain.
The paper tackles the problem of extractive question answering on the SQuAD dataset by proposing a model that builds fixed-length representations of all spans in a document using a recurrent network, improving performance by 5% over the best published results and reducing error by over 50% compared to a baseline.
The reading comprehension task, that asks questions about a given evidence document, is a central problem in natural language understanding. Recent formulations of this task have typically focused on answer selection from a set of candidates pre-defined manually or through the use of an external NLP pipeline. However, Rajpurkar et al. (2016) recently released the SQuAD dataset in which the answers can be arbitrary strings from the supplied text. In this paper, we focus on this answer extraction task, presenting a novel model architecture that efficiently builds fixed length representations of all spans in the evidence document with a recurrent network. We show that scoring explicit span representations significantly improves performance over other approaches that factor the prediction into separate predictions about words or start and end markers. Our approach improves upon the best published results of Wang & Jiang (2016) by 5% and decreases the error of Rajpurkar et al.'s baseline by > 50%.