Deploying Lifelong Open-Domain Dialogue Learning
This work addresses the lack of naturalness and real-world relevance in NLP datasets for dialogue systems, offering a more interactive and cost-effective learning approach.
The authors tackled the problem of static, crowdsourced datasets in NLP by deploying a role-playing game where learning agents interact with humans in an open-domain fantasy world, showing that models trained on these conversations progressively improve in automatic metrics and online engagement scores, with more efficiency and lower cost than crowdsourced data.
Much of NLP research has focused on crowdsourced static datasets and the supervised learning paradigm of training once and then evaluating test performance. As argued in de Vries et al. (2020), crowdsourced data has the issues of lack of naturalness and relevance to real-world use cases, while the static dataset paradigm does not allow for a model to learn from its experiences of using language (Silver et al., 2013). In contrast, one might hope for machine learning systems that become more useful as they interact with people. In this work, we build and deploy a role-playing game, whereby human players converse with learning agents situated in an open-domain fantasy world. We show that by training models on the conversations they have with humans in the game the models progressively improve, as measured by automatic metrics and online engagement scores. This learning is shown to be more efficient than crowdsourced data when applied to conversations with real users, as well as being far cheaper to collect.