指導教授：林守德臺灣大學：資訊工程學研究所顏妤安Yan, Yu-AnYu-AnYan2014-11-262018-07-052014-11-262018-07-052014http://ntur.lib.ntu.edu.tw//handle/246246/261407隨著網路服務例如搜尋引擎、社群網路的發達，我們能夠從網路上取得的資訊愈來愈多。本篇論文提出利用網路資訊來增進電視節目收視率預測的方法。我們從網路上取出了例如Google搜尋關鍵字趨勢、Facebook上的貼文、情緒等資料，並以兩個常見的電視收視率預測模型，支持向量機回歸及高斯過程回歸，來結合我們所提出的資訊。實驗結果證實，我們提出的資訊確實能提高收視率預測模型的精準度。除此之外，我們比較了未來預測(forecasting)與當下預測(nowcasting)兩種不同情況下的實驗結果，並加以分析。Because of rising of web services such as search engine and social network, there are more and more information we can extract from the internet. In this paper, we propose a　method to utilize information from the internet to enhance the performance of TV rating prediction models. We extract features such as trends of search query, posts and opinion on Facebook; apply two common models for TV rating prediction, Support Vector Regression and Gaussian Process Regression, with these features, and demonstrate that they can be useful for TV rating prediction. Furthermore, we compare the performance of our model under two different settings, namely forecasting and nowcasting, and provide several discussions.Contents 口試委員審定書………………………………………………………………………………i 誌謝…………………………………………………………………………………………………ii 摘要………………………………………………………………………………………………iii Abstract……………………………………………………………………………………iv Contents………………………………………………………………………………………v List of Figures…………………………………………………………………vi List of Tables………………………………………………………………viii Chapter 1 Introduction…………………………………………………1 Chapter 2 Related Work…………………………………………………2 Chapter 3 Frameworks and Features……………………3 Chapter 4 Methodology…………………………………………………12 4.1 Support Vector Regression (SVR)……………12 4.2 Gaussian Process Regression (GPR)………13 Chapter 5 Experiments…………………………………………………15 5.1 Dataset and Evaluation Metric…………………15 5.2 Experiments of Daily Drama…………………………17 5.2.1 Experiment Results…………………………………………17 5.2.2 Feature Analysis………………………………………………21 5.3 Experiments of Weekly Drama………………………23 5.3.1 Experiment Results…………………………………………23 5.3.2 Feature Analysis………………………………………………27 5.4 Forecasting and Nowcasting…………………………29 5.4.1 On Daily Dramas…………………………………………………30 5.4.2 On weekly Dramas………………………………………………31 Chapter 6 Conclusion……………………………………………………32 Chapter 7 Future Work…………………………………………………32 References………………………………………………………………………………341929679 bytesapplication/pdf論文公開時間：2016/08/16論文使用權限：同意有償授權(權利金給回饋學校)電視收視率預測支持向量機回歸高斯過程回歸時間序列趨勢社群資訊thesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/261407/1/ntu-103-R01922127-1.pdf