Towards Question Answering with Multi-hop Reasoning over Knowledge using a Neural Network Model with External Memories

Abstract

The Differentiable Neural Computer (DNC), a neural network model with an addressable external memory, can solve algorithmic and question answering tasks. As improved versions of DNC, rsDNC and DNC-DMS have been proposed. However, how to integrate structured knowledge into these DNC models remains a challenging research question. We incorporate an architecture for knowledge into such DNC models, i.e. DNC, rsDNC and DNC-DMS, to improve the ability to generate correct answers for questions using both contextual information and structured knowledge. Our improved rsDNC model outperformed the other models with the mean top-l accuracy and top-10 accuracy in GEO dataset. In addition, our improved rsDNC model achieved the best performance with the mean top-10 accuracy in augmented GEO dataset.