音樂的本質是傳達人類心靈深處的情感，可使人生氣或溫柔，鼓勵或節制，且具有淨化情緒與昇華情感的作用（方銘健，1997）。亦即音樂能夠喚起強烈的情緒經驗，在聆賞中每個人建構獨特的音樂經驗，影響其腦部、身體與情感的發展（Reimer, 2005）。現今多媒體數位化的環境下，音樂資源在網路上的分享已成為Web2.0 時代常見的應用。如何有效的管理與分享這些大量產生的音樂資源以利同好社群的形成，創造有利的商機，便成為一件十分重要的工作。本研究提出一個以心情分類與標籤的雛型系統呈現一種以心情為基礎的音樂管理方式。 本研究提出混和式的分類模式，並透過不同的相似度比對方式來探討歌曲之間相似度比對的結果。而發現考量標籤計算的相似度比對比傳統的相似度比對更適用於標籤資源間的相似度比對。 The essence of music is to convey the emotion which is deeply within human soul, and it can purify and raise human’s mood and feeling, such as angry, gentle, encouraged, and restrained (M. J. Fang, 1997). In another word, music can evoke strong emotional experience, and build unique musical experience to affect the brain, physical and emotional development when enjoying music (Reimer, 2005). It is not rare under the digital multimedia environment nowadays that the applying shared music resource to internet in Web 2.0. How to effectively manage and share these great deals of generated music resource to facilitate the community of music aficionados and create a favorable business opportunities becomes a very important work. This study proposes a prototype system of mood classification and labeling to present a mood-based music management. This study raises the mixed-type classification model, and discusses the result of songs similarity comparison with different modes of similarity comparison. The experiment result shows that the calculated similarity of labeling based comparison is more suitable for the labeled resource than traditional similarity comparison.
