預測音樂於串流平台上之表現：以YouTube影音與Spotify資料為例

Predicting Performance on Media Service: Analyzing Music Video Data on YouTube

陳隆翔(Lung-Hsiung Chen)

指導教授 ： 曹承礎

共同指導教授 ： 吳玲玲(Ling-Ling Wu)

國立臺灣大學/管理學院/資訊管理學研究所/碩士(2019年)

https://doi.org/10.6342/NTU201902981

摘要

隨著近年台灣用戶影音視聽的習慣轉變，實體唱片業的利潤逐漸下滑，而數位音樂利潤逐年上升，甚至串流服務有了高幅度的增長。而像是YouTube平台等影音媒體平台，成為人們取得娛樂與資訊的重要平台，目前許多音樂公司亦會選擇在歌曲發佈時，同時將音樂影音( MV, Music Video) 上傳至YouTube，使觀眾能在最便利的管道，得到最佳的觀影品質，也就是影像與聲音的雙重體驗，進而促進未來可能更多深入的收聽行動。
而隨著串流影音的盛行，在影音平台上取得成功的用戶體驗，並轉換成具收益的下載量，對於音樂公司來說，是在歌曲表現的重要指標，而YouTube作為最容易、同時也是最可能是聽眾第一次接觸該音樂作品的第一管道，本研究透過分析YouTube官方音樂錄影帶的資料，預測在Spotify上的串流流量，並瞭解何種因子可能會與此轉換有關，提供未來相關人員進行深入研究，或作為行銷規劃的決策依據。

關鍵字

音樂錄影帶 ； 流量預測 ； 影片特徵 ； 迴歸模型

並列摘要

In recent years, the habits of watching media in Taiwan have been changing, media platforms such as YouTube have become an important platform for people to get entertainment and information. The profits of the physical musice recording industry have gradually declined, while the profits of digital music have increased year by year and even the streaming services have experienced a signicicant growth. At present, many music media companies will choose to upload MV (Music Video) to YouTube when the song recordings are released, so that viewers can experience the best viewing quality, that is, the dual experience of visual and audio, and thus become a potential motivation for their further experience with the music or singer.
With the prevalence of streaming media platform, achieving a successful user experience on the audio-visual platform and converting it into a revenue-generating download is an important indicator of song performance for music companies, and YouTube is the easiest way for listener to have first interaction with the music or singer. Our study researcher on YouTube's official music video data, attempting to predict streaming traffic volume on Spotify, and to figure out which factors may be relevant to this conversion. In the future, relevant personnel can conduct a further research into it and conduct their decision on marketing strategy based on our results.

並列關鍵字

Music Video ； Prediction ； Video Feature ； Regression Model

參考文獻

APALA, K. R., JOSE, M., MOTNAM, S., CHAN, C.-C., LISZKA, K. J. & DE GREGORIO, F. Prediction of movies box office performance using social media. Proceedings of the 2013 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, 2013. ACM, 1209-1214.

BREIMAN, L. J. M. L. 2001. Random forests. 45, 5-32.

CHA, M., KWAK, H., RODRIGUEZ, P., AHN, Y.-Y. & MOON, S. 2007. I tube, you tube, everybody tubes: analyzing the world's largest user generated content video system. Proceedings of the 7th ACM SIGCOMM conference on Internet measurement. San Diego, California, USA: ACM.

CHEN, P.-L., TSAI, C.-T., CHEN, Y.-N., CHOU, K.-C., LI, C.-L., TSAI, C.-H., WU, K.-W., CHOU, Y.-C., LI, C.-Y. & LIN, W.-S. A linear ensemble of individual and blended models for music rating prediction. Proceedings of KDD Cup 2011, 2012. 21-60.

CUNNINGHAM, S. J. & NICHOLS, D. M. 2008. How people find videos. Proceedings of the 8th ACM/IEEE-CS joint conference on Digital libraries. Pittsburgh PA, PA, USA: ACM.