Neural Response Generation with Meta-Words

TL;DR

This paper introduces a novel open domain response generation method using meta-words, which are structured attributes that enable explicit, controllable, and explainable dialogue response modeling.

Contribution

The paper proposes a sequence-to-sequence model enhanced with a goal tracking memory network to incorporate meta-words, improving response relevance, diversity, and controllability.

Findings

Significantly outperforms state-of-the-art models in relevance and diversity

Achieves high accuracy in modeling one-to-many response variations

Demonstrates effectiveness through large-scale dataset experiments

Abstract

We present open domain response generation with meta-words. A meta-word is a structured record that describes various attributes of a response, and thus allows us to explicitly model the one-to-many relationship within open domain dialogues and perform response generation in an explainable and controllable manner. To incorporate meta-words into generation, we enhance the sequence-to-sequence architecture with a goal tracking memory network that formalizes meta-word expression as a goal and manages the generation process to achieve the goal with a state memory panel and a state controller. Experimental results on two large-scale datasets indicate that our model can significantly outperform several state-of-the-art generation models in terms of response relevance, response diversity, accuracy of one-to-many modeling, accuracy of meta-word expression, and human evaluation.