網路感知之以社群理論為基礎之同儕式電影推薦系統

Network-aware Social-based Movie Recommender for Peer-to-Peer System

簡嘉瑩(Chia-Ying Chien)

指導教授 ： 周承復

國立臺灣大學/電機資訊學院/資訊網路與多媒體研究所/碩士(2010年)

https://doi.org/10.6342/NTU.2010.01660

摘要

推薦系統向來是個熱門的焦點，目前已經在許多平台上普遍的被使用，例如像電子商務這種龐大的系統，推薦機制可以幫助使用者更快速找到自己有興趣的項目。此外，在這個資訊流通的時代，點對點檔案分享系統，一直深受使用者的歡迎。結合兩者，便是一個對使用者而言便利又有效率的系統。對於這樣的應用，過去的推薦系統，並沒有考慮假若使用者去下載系統推薦的項目，是否能夠快速有效的得到檔案。因此在這篇論文中，我們使用了一個分散式的電影搜尋推薦系統，能夠搜尋和分享電影，為了充分利用網路頻寬，減少使用者下載檔案的時間，我們以混合式推薦系統為基礎，再加入網路因素來過濾出網路傳輸品質較好的影片，作為我們最後的推薦項目。最後經由模擬實驗可以看到我們的網路過濾機制，在不影響準確度的情形下，對於檔案傳輸時間確實有所改善。

關鍵字

分散式推薦系統 ； 電影推薦系統 ； 下載效率 ； 點對點檔案分享系統 ； 可用上傳頻寬

並列摘要

Recommendation system is popular and well-known by people, and also used in many platforms, like e-commerce which is large and complex. Recommendation mechanism can help users quickly finding the items that they are interested in. In addition, today, more and more information is transferred in the world, peer-to-peer file sharing system is preferred by users. For all the users, it is a strong and convenient service to combine these two ideas. For this kind of applications, existed recommendation systems didn’t consider if users could get the files as quick as possible. Therefore, in this thesis, we propose a distributed movie search recommendation system which can search and share movies. In order to use users’ bandwidth adequately and decrease the time for users to download the files, we base on a hybrid recommendation systems, and consider network factors to filter out the movies which can be downloaded efficiently in network as our final recommendation list. Finally,according to the simulation result , it shows that our network filter can help decreasing the time to transfer files without effecting the accuracy.

並列關鍵字

Distributed recommendation system ； Movie recommendation system ； Download efficiency ； P2P file sharing system ； Available upload bandwidth

參考文獻

[4] G. Adomavicius and A. Tuzhilin. Toward the next generation of recommender systems:

[5] E. Aimeur, G. Brassard, J. M. Fernandez, and F. S. Mani Onana. Alambic: a

privacy-preserving recommender system for electronic commerce. Int. J. Inf. Secur.,

7(5):307–334, 2008.

[6] M. Balabanovi’c and Y. Shoham. Fab: content-based, collaborative recommendation.