第四章 實驗結果與討論
4.2 實驗結果
4.2.1 分群結果
實驗共收集 8307 個家戶的收視行為,以隨機的方式將百分之八十的家戶歸為訓練 資料,另外百分之二十的家戶歸為測試資料;訓練資料為 6645 戶以及測試資料 1662 戶。
本實驗的目的是了解頻道收視行為對隨選視訊的解釋性,因此藉由家戶的頻道收視行為 建立相似度後的分群結果。根據家戶相似表分別建立三群、四群、五群的分群數,各群 的結果是根據隨選視訊類型區分,如表 4-1。
訓練資料 家戶相似度計算
合併分群
各群的頻道 代表態樣
測試資料入群
測試資料 各群的隨選視
訊統計
推薦清單
14
15
群 5 426 212 638 4.2.2 權重分配實驗結果
實驗共分五種方式,分別為:方法一、完全以頻道收視分群。方法二、完全以隨選 視訊分群。方法三、兩種收視行為權重均等,各為百分之五十比重分群。方法四、頻道 收視與隨選視訊,分別為百分之二十與百分之八十比重分群。方法五、分群比例為頻道 收視佔百分之八十與隨選視訊佔百分之二十。藉由這五種方式,比較實驗推薦成效的差 異性。並了解是否會因為頻道收視行為解釋用戶的隨選視訊收視行為。
分別將三種分群數進行前述的五種實驗方式,各分群的推薦準確率差異如圖 4-2。
其結果顯示完全以頻道分群的準確率高於其他方法。推論頻道收視行為的群間家戶行為 類似,造成隨選視訊代表態樣的區隔力不明顯。所以若加上隨選視訊的權重,則準確率 會較低。回應率的衡量會與準確率相關,當準確率越高時,回應率會較低;因此,實驗 結果的回應率如圖 4-3。F1 指標則平衡準確率與回應率的結果,如圖 4-4。
16
17
18
19
第五章 結論
以往研究偏重於單一收視行為的分析,直接將隨選視訊收視紀錄進行傳統式推薦或 改良。本研究突破從單一行為進行推薦,進而紀錄收視戶二十四小時頻道收視習慣來推 論其喜好性、偏好性,並結合隨選視訊推薦強化既有研究並了解用戶實際習性。
本研究在商業用途上更可作為隨選視訊頻道商在商業分析或產品包裝上的參考。以 頻道行為結合隨選視訊紀錄瞭解收視戶對於免費與付費的收視習慣、喜好性、偏好性等 之相關性進行有效推薦。未來研究更可結合用戶端外部的使用行為,例如,線上影音收 視習慣、社群影響等習性來更幫助推薦系統有更準確的推薦。
20
第六章 參考文獻
彭心儀、呂書賢、邱懷祖(2008),〈WiMAX 服務對於視聽產業的影響(PART-II:歐 盟影音服務管制之變革)〉,刊載於台灣服務業聯網,2008 年 3 月 31 日,網址:
http://www.twcsi.org.tw/topic_detail.php?Iid=122.
褚怡涵,2012,植基於情境感知的 IPTV 節目推薦系統,東吳大學。
Adomavicius, Gediminas, & Tuzhilin, Alexander. (2005). Toward the next generation of recommender systems: A survey of the state-of-the-art and possible extensions.
Knowledge and Data Engineering, IEEE Transactions on, 17(6), 734-749.
Balabanović, Marko, & Shoham, Yoav. (1997). Fab: content-based, collaborative recommendation. Communications of the ACM, 40(3), 66-72.
Bambini, Riccardo, Cremonesi, Paolo, & Turrin, Roberto. (2011). A recommender system for an iptv service provider: a real large-scale production environment Recommender systems handbook (pp. 299-331): Springer.
Breese, John S, Heckerman, David, & Kadie, Carl. (1998). Empirical analysis of predictive algorithms for collaborative filtering. Paper presented at the Proceedings of the Fourteenth conference on Uncertainty in artificial intelligence.
Burke, Robin. (2002). Hybrid recommender systems: Survey and experiments. User modeling and user-adapted interaction, 12(4), 331-370.
Burke, Robin. (2007). Hybrid web recommender systems The adaptive web (pp. 377-408):
Springer.
De Campos, Luis M, Fernández-Luna, Juan M, Huete, Juan F, & Rueda-Morales, Miguel A.
(2010). Combining content-based and collaborative recommendations: A hybrid
approach based on Bayesian networks. International Journal of Approximate Reasoning, 51(7), 785-799.
Ha, Sung Ho. (2002). Helping online customers decide through web personalization. IEEE Intelligent systems, 17(6), 34-43.
Hand, Stacey, & Varan, Duane. (2008). Interactive narratives: Exploring the links between empathy, interactivity and structure Changing television environments (pp. 11-19):
Springer.
Jensen, Jens F. (2008). Interactive television-a brief media history Changing Television Environments (pp. 1-10): Springer.
Kamahara, Junzo, Asakawa, Tomofumi, Shimojo, Shinji, & Miyahara, Hideo. (2005). A community-based recommendation system to reveal unexpected interests. Paper
presented at the Multimedia Modelling Conference, 2005. MMM 2005. Proceedings of the 11th International.
Kazienko, P, & Kiewra, M. (2004). Personalized recommendation of web pages. Chapter, 10, 163-183.
Lang, Ken. (1995). Newsweeder: Learning to filter netnews. Paper presented at the In Proceedings of the Twelfth International Conference on Machine Learning.
Melville, Prem, Mooney, Raymond J, & Nagarajan, Ramadass. (2002). Content-boosted collaborative filtering for improved recommendations. Paper presented at the AAAI/IAAI.
Mladenic, Dunja. (1996). Personal WebWatcher: design and implementation.
Montaner, Miquel, López, Beatriz, & De La Rosa, Josep Lluís. (2003). A taxonomy of recommender agents on the internet. Artificial intelligence review, 19(4), 285-330.
21
Pazzani, Michael J, Muramatsu, Jack, & Billsus, Daniel. (1996). Syskill & Webert: Identifying interesting web sites. Paper presented at the AAAI/IAAI, Vol. 1.
Resnick, Paul, Iacovou, Neophytos, Suchak, Mitesh, Bergstrom, Peter, & Riedl, John. (1994).
GroupLens: an open architecture for collaborative filtering of netnews. Paper presented at the Proceedings of the 1994 ACM conference on Computer supported cooperative work.
Shardanand, Upendra, & Maes, Pattie. (1995). Social information filtering: algorithms for automating “word of mouth”. Paper presented at the Proceedings of the SIGCHI conference on Human factors in computing systems.
Symeonidis, Panagiotis, Nanopoulos, Alexandros, & Manolopoulos, Yannis. (2008).
Providing justifications in recommender systems. Systems, Man and Cybernetics, Part A:
Systems and Humans, IEEE Transactions on, 38(6), 1262-1272.