Skipping Word: A Character-Sequential Representation based Framework for Question Answering

TL;DR

This paper introduces a character-sequential representation framework for question answering that simplifies the modeling process and achieves stable, competitive results across benchmarks by using CNNs on character sequences instead of word embeddings.

Contribution

The paper proposes a novel character-sequential representation approach combined with CNNs for question answering, avoiding complex corpus and dictionary transformations associated with word embeddings.

Findings

Competitive performance on benchmark datasets

Simpler modeling procedure compared to word embedding methods

More stable performance across different benchmarks

Abstract

Recent works using artificial neural networks based on word distributed representation greatly boost the performance of various natural language learning tasks, especially question answering. Though, they also carry along with some attendant problems, such as corpus selection for embedding learning, dictionary transformation for different learning tasks, etc. In this paper, we propose to straightforwardly model sentences by means of character sequences, and then utilize convolutional neural networks to integrate character embedding learning together with point-wise answer selection training. Compared with deep models pre-trained on word embedding (WE) strategy, our character-sequential representation (CSR) based method shows a much simpler procedure and more stable performance across different benchmarks. Extensive experiments on two benchmark answer selection datasets exhibit the competitive performance compared with the state-of-the-art methods.