This is a page that curates AI-related papers published worldwide. All content here is summarized using Google Gemini and operated on a non-profit basis. Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.
This paper provides a comprehensive review of multi-turn dialogue systems based on large-scale language models (LLMs). We summarize LLMs and their application to subtasks. We detail recent advances in multi-turn dialogue systems, including LLM-based open-domain dialogue (ODD) and task-oriented dialogue (TOD) systems, as well as datasets and evaluation metrics. We also discuss future research directions and recent research issues arising from LLM development and the growing demands of multi-turn dialogue systems.
Takeaways, Limitations
•
Takeaways:
◦
This comprehensive guide provides researchers with useful information by summarizing the latest trends and technologies in LLM-based multi-session conversation systems.
◦
In-depth analysis of both ODD and TOD systems enhances understanding of various applications.
◦
We present future research directions for LLM development and multi-session dialogue systems, suggesting future research focuses.
•
Limitations:
◦
This paper focuses on a comprehensive review of existing research and may not include new research findings.
◦
Due to the rapidly evolving nature of the LLM and multi-session conversation systems fields, new technologies and research may emerge after the paper is published.
◦
There may be a bias towards certain LLMs or approaches.
