Full-Time Supervision based Bidirectional RNN for Factoid Question Answering

TL;DR

This paper introduces FTS-BRNN, a bidirectional RNN model with supervision at every time step, significantly improving factoid question answering accuracy.

Contribution

The paper proposes a novel full-time supervision mechanism for BRNNs, enhancing information utilization in QA tasks.

Findings

FTS-BRNN outperforms baseline models in accuracy.

Supervision at every time step improves information retention.

Achieves state-of-the-art results on factoid QA.

Abstract

Recently, bidirectional recurrent neural network (BRNN) has been widely used for question answering (QA) tasks with promising performance. However, most existing BRNN models extract the information of questions and answers by directly using a pooling operation to generate the representation for loss or similarity calculation. Hence, these existing models don't put supervision (loss or similarity calculation) at every time step, which will lose some useful information. In this paper, we propose a novel BRNN model called full-time supervision based BRNN (FTS-BRNN), which can put supervision at every time step. Experiments on the factoid QA task show that our FTS-BRNN can outperform other baselines to achieve the state-of-the-art accuracy.