Earticle

초록

영어

Lots of recommender systems have exposed the cold-start problem. So, we propose a recommendation methodology using user emotion information to address such problem. We extract the emotion from the facial expression while a user watches a content. Our methodology is consisted of the four phase; data collection, data representation of facial expression, neighbor formation and preference prediction on contents. To evaluate the proposed methodology, we compared with the traditional collaborative filtering. From the experiment results, we can see that the proposed methodology is better performance than the traditional collaborative filtering about the new user problem.

목차

Abstract
 1. Introduction
 2. Related works
  2.1 Collaborative filtering
  2.2 Emotion Recognition
 3. Methodology
  3.1 Overall view
  3.2 Data collection
  3.3 Data Preprocessing
  3.4 Data representation of facial expression
  3.5 Neighbor formation
  3.6 Preference prediction on contents
 4. Experimental Result
  4.1 Data set and experiment design
  4.2 Experimental results
 5. Conclusion
 References

저자

• Myung Geun Oh [ School of Management, Kyung Hee University ]
• Il Young Choi [ School of Dance, Kyung Hee University ]
• Jae Kyeong Kim [ School of Management, Kyung Hee University ]

참고문헌