Modeling Relational Information in Question-Answer Pairs with Convolutional Neural Networks

TL;DR

This paper introduces CNN-based models that incorporate relational word-matching information between question-answer pairs, significantly improving answer selection accuracy.

Contribution

It presents a novel way to encode relational information in CNNs for question-answer matching, enhancing performance on benchmark datasets.

Findings

Relational encoding improves answer selection accuracy.

The proposed models approach state-of-the-art results.

Relational information is effectively captured by the CNN architecture.

Abstract

In this paper, we propose convolutional neural networks for learning an optimal representation of question and answer sentences. Their main aspect is the use of relational information given by the matches between words from the two members of the pair. The matches are encoded as embeddings with additional parameters (dimensions), which are tuned by the network. These allows for better capturing interactions between questions and answers, resulting in a significant boost in accuracy. We test our models on two widely used answer sentence selection benchmarks. The results clearly show the effectiveness of our relational information, which allows our relatively simple network to approach the state of the art.