Lecture Recommendation Chatbot

Developed a hybrid chatbot that combines task-oriented dialogue (TOD) with open-domain chit-chat to provide personalized lecture recommendations.
Successfully deployed in SK Telecom internal systems for real-world use.
Achieved 1st place in the SKT AI Fellowship program.
Published and presented at the Korea Institute of Information and Communication Engineering (KICS) Conference, Jeju (2022).
- Yerin Choi, Yeen-heui Yeen, and Dong-Geun Kim (2022). Dialogue System for User Customized Lecture Recommendation. Proceedings of the KICS Fall Conference.

Led data collection and augmentation for task-specific TOD dialogues
Adapted generic task-oriented dialogue structures to the lecture recommendation domain via controlled style transfer
Developed and integrated the full chatbot pipeline, from intent handling to response generation and deployment