SHARE: Shared Memory-Aware Open-Domain Long-Term Dialogue Dataset Constructed from Movie Script
- 주제(키워드) Natural Language Processing , Dialogue Systems , Long-Term Dialogue , Open-Domain Dialogue , Dialogue Dataset
- 발행기관 서강대학교 일반대학원
- 지도교수 이화란
- 발행년도 2026
- 학위수여년월 2026. 2
- 학위명 석사
- 학과 및 전공 일반대학원 인공지능학과협동과정
- 실제URI http://www.dcollection.net/handler/sogang/000000082616
- UCI I804:11029-000000082616
- 본문언어 영어
- 저작권 논문은 저작권에 의해 보호받습니다.
초록(요약문)
Shared memories between two individuals strengthen their bond and are crucial for facilitating their ongoing conversations. This dissertation 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.
more목차
Chapter 1. Introduction 1
Chapter 2. Related work 5
2.1 Long-term Dialogue Dataset 5
2.2 Long-term Dialogue Method 5
Chapter 3. Dataset Collection 6
Chapter 4. Dialogue Framework 10
4.1 Task Definition 10
4.2 Response Generation 11
4.3 Memory Management 13
Chapter 5. Experiments 15
5.1 Experimental Setup 15
5.2 Experimental Results 16
5.3 Case Study 24
Chapter 6. Conclusion 25
References 27
