SHARE: Shared Memory-Aware Open-Domain Long-Term Dialogue Dataset Constructed from Movie Script

TL;DR

This paper introduces SHARE, a novel long-term dialogue dataset from movie scripts, and EPISODE, a framework that leverages shared memories to enhance dialogue engagement and sustainability.

Contribution

The paper presents a new dataset SHARE derived from movie scripts and a framework EPISODE that utilizes shared memories for improved long-term dialogue management.

Findings

Shared memories increase dialogue engagement.

EPISODE effectively manages shared experiences.

SHARE dataset enables long-term dialogue research.

Abstract

Shared memories between two individuals strengthen their bond and are crucial for facilitating their ongoing conversations. This study aims to make long-term dialogue more engaging by leveraging these shared memories. To this end, we introduce a new long-term dialogue dataset named SHARE, constructed from movie scripts, which are a rich source of shared memories among various relationships. Our dialogue dataset contains the summaries of persona information and events of two individuals, as explicitly revealed in their conversation, along with implicitly extractable shared memories. We also introduce EPISODE, a long-term dialogue framework based on SHARE that utilizes shared experiences between individuals. Through experiments using SHARE, we demonstrate that shared memories between two individuals make long-term dialogues more engaging and sustainable, and that EPISODE effectively manages shared memories during dialogue. Our dataset and code are available at https://github.com/e1kim/SHARE.