Data Collection for Interactive Learning through the Dialog
This addresses a key limitation in dialog AI by providing a resource for improving fact reasoning, though it is incremental as it focuses on data collection rather than novel methods.
The paper tackles the problem of fact sparsity in open-domain dialog systems by introducing a dataset of 1900 dialogs that enables testing interactive learning, where systems can acquire new facts from user utterances during conversations.
This paper presents a dataset collected from natural dialogs which enables to test the ability of dialog systems to learn new facts from user utterances throughout the dialog. This interactive learning will help with one of the most prevailing problems of open domain dialog system, which is the sparsity of facts a dialog system can reason about. The proposed dataset, consisting of 1900 collected dialogs, allows simulation of an interactive gaining of denotations and questions explanations from users which can be used for the interactive learning.