I love your chain mail! Making knights smile in a fantasy game world: Open-domain goal-oriented dialogue agents
This addresses the challenge of creating more natural and effective dialogue agents for interactive gaming or similar domains, though it appears incremental in bridging two existing dialogue types.
The paper tackles the problem of combining chit-chat and goal-oriented dialogue in a fantasy game environment, showing that models trained with reinforcement learning outperform a baseline and can converse naturally to achieve goals.
Dialogue research tends to distinguish between chit-chat and goal-oriented tasks. While the former is arguably more naturalistic and has a wider use of language, the latter has clearer metrics and a straightforward learning signal. Humans effortlessly combine the two, for example engaging in chit-chat with the goal of exchanging information or eliciting a specific response. Here, we bridge the divide between these two domains in the setting of a rich multi-player text-based fantasy environment where agents and humans engage in both actions and dialogue. Specifically, we train a goal-oriented model with reinforcement learning against an imitation-learned ``chit-chat'' model with two approaches: the policy either learns to pick a topic or learns to pick an utterance given the top-K utterances from the chit-chat model. We show that both models outperform an inverse model baseline and can converse naturally with their dialogue partner in order to achieve goals.