Authors:
Song-Yi Shin
1
;
Joonwhoan Lee
1
;
Kyoung-bae Eum
2
and
Eun-Jong Park
3
Affiliations:
1
Chonbuk National University, Korea, Republic of
;
2
Kunsan National University, Korea, Republic of
;
3
Electronics and Telecommunications Research Institute, Korea, Republic of
Keyword(s):
Features of MPEG-7, Emotion-based music retrieval, Multi-query method, Consistency principle.
Related
Ontology
Subjects/Areas/Topics:
Multimedia and User Interfaces
;
Searching and Browsing
;
Web Information Systems and Technologies
;
Web Interfaces and Applications
Abstract:
In this paper, we propose the construction of multi - queries and consistency principal for emotion-based music retrieval. Existing content-based music retrievals which use a single query reflect the retrieval intention of user by moving the query point or updating the weights. However, these methods have the limitation to represent the complicated factors of emotion. In the proposed method, additional queries of music are taken in each feedback process to express the user’s emotion. We classify the music by the emotions. And the music is clustered by the MKBC(Mercer Kernel-Based Clustering) method. After that, the inclusion degree of each descriptor is obtained. This means the weight that represents the importance of each descriptor for each emotion in order to reduce the computation. We got the excellent result within the 2nd retrieval through the feedback. In the feedback process, we used the consistency principle and multi- queries.