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.
@article{arxiv.1906.06050,
title = {Neural Response Generation with Meta-Words},
author = {Can Xu and Wei Wu and Chongyang Tao and Huang Hu and Matt Schuerman and Ying Wang},
journal= {arXiv preprint arXiv:1906.06050},
year = {2019}
}