Bidirectional Attention Flow for Machine Comprehension

TL;DR

This paper introduces the Bi-Directional Attention Flow (BIDAF) network, a novel model for machine comprehension that captures complex context-query interactions without early summarization, achieving state-of-the-art results on major datasets.

Contribution

The paper presents a new bi-directional attention flow mechanism and a hierarchical model that improves context-query interaction modeling in machine comprehension.

Findings

Achieves state-of-the-art results on SQuAD dataset.

Outperforms previous models on CNN/DailyMail cloze test.

Demonstrates effective multi-level context representation.

Abstract

Machine comprehension (MC), answering a query about a given context paragraph, requires modeling complex interactions between the context and the query. Recently, attention mechanisms have been successfully extended to MC. Typically these methods use attention to focus on a small portion of the context and summarize it with a fixed-size vector, couple attentions temporally, and/or often form a uni-directional attention. In this paper we introduce the Bi-Directional Attention Flow (BIDAF) network, a multi-stage hierarchical process that represents the context at different levels of granularity and uses bi-directional attention flow mechanism to obtain a query-aware context representation without early summarization. Our experimental evaluations show that our model achieves the state-of-the-art results in Stanford Question Answering Dataset (SQuAD) and CNN/DailyMail cloze test.