Denotation Extraction for Interactive Learning in Dialogue Systems

TL;DR

This paper introduces a new interactive denotation extraction task using real user dialogue data to improve training data collection for question answering systems, demonstrating initial neural network model evaluations.

Contribution

It proposes a novel interactive denotation extraction task from human-machine dialogues and provides initial neural network model evaluations for this purpose.

Findings

Attention-based neural models show promising results

Interactive data collection can enhance dialogue system training

Initial evaluation indicates feasibility of the proposed approach

Abstract

This paper presents a novel task using real user data obtained in human-machine conversation. The task concerns with denotation extraction from answer hints collected interactively in a dialogue. The task is motivated by the need for large amounts of training data for question answering dialogue system development, where the data is often expensive and hard to collect. Being able to collect denotation interactively and directly from users, one could improve, for example, natural understanding components on-line and ease the collection of the training data. This paper also presents introductory results of evaluation of several denotation extraction models including attention-based neural network approaches.