Towards conversational information seeking
- Zhaochun Ren
- maandag 27 februari 2023
Niels Bohrweg 1
2333 CA Leiden
Recent years have witnessed a surge of interest in conversational artificial intelligence. In particular, the rapid development of pre-trained language models and their applications to dialogue tasks, as exemplified by ChatGPT, offers a new perspective on how people can seek information through conversations. Conversational information seeking aims to acquire information through conversation to meet users' needs by tracking their preferences and generating appropriate responses. In this talk, I will introduce my research on conversational information seeking, with a specific focus on representing knowledge, modeling user preferences, and generating textual information. It's worth noting that the increased use of large language models presents new challenges for future research in conversational information seeking. Building on existing research efforts, I will provide a comprehensive overview of the challenges and potential visions for future research in this field.