Personalizing Dialogue Agents: I have a dog, do you have pets too?

TL;DR

This paper introduces a method to personalize dialogue agents by conditioning on profile information, improving engagement and consistency in chit-chat conversations, and enabling the prediction of interlocutor profiles from dialogue data.

Contribution

It presents a novel approach to personalize dialogue models using profile conditioning and demonstrates how engaging conversations can reveal profile details.

Findings

Enhanced dialogue engagement and consistency

Ability to predict interlocutor profiles from conversations

Improved next utterance prediction accuracy

Abstract

Chit-chat models are known to have several problems: they lack specificity, do not display a consistent personality and are often not very captivating. In this work we present the task of making chit-chat more engaging by conditioning on profile information. We collect data and train models to (i) condition on their given profile information; and (ii) information about the person they are talking to, resulting in improved dialogues, as measured by next utterance prediction. Since (ii) is initially unknown our model is trained to engage its partner with personal topics, and we show the resulting dialogue can be used to predict profile information about the interlocutors.