電子商務已成為現今勢不可擋的風潮,但是在網路虛擬的世界中購物,消費者不但要面對五花八門龐雜的資料內容,更要面對比實體世界更多的風險及不確定性。因此,網站上雖然資訊垂手可得,但消費者卻更加需要資訊的有效過濾與其他相同偏好者的意見及推薦,以作為是否購買的依據。 本研究嘗試將「協同過濾」機制應用在網站商品推薦之上,希望藉由此機制推薦出使用者所喜愛的商品,以減少消費者在網路上所浪費的時間。本研究之進行主要可以分成兩部分,第一部份乃是實際開發一套以「協同過濾」為基礎的商品推薦系統,其內容包括「建立使用者偏好資料庫」、「建立分群模組」、「建立推薦模組」三個過程。由於本研究選擇以「手機」作為商品推薦之標的物,因此,在「分群模組」中,特別採用「最適合男生使用」、「最適合女生使用」、「外型」、「功能」、「娛樂」和「鈴聲」作為「分群」之過濾變數,「推薦模組」則是以推薦出消費者心目中的最佳手機為目標。第二部分則是透過實地實驗來驗證、比較此系統的推薦成效。 從本研究的實地實驗結果中,歸納出以下主要結論: 1.「協同過濾」機制對於線上商品之推薦確實有其成效。 2.不同分群過濾變數對於商品的推薦成效皆佳。 3.在本研究中不同「干擾變數」對於「分群過濾變數」有差異化之成效。 本研究的研究成果將有助於電子商務業者在進行線上商品推薦時,一個有用的參考機制。而對學界而言,將有助於學界了解「協同過濾」對「商品推薦」的影響效果,並可供發展「商品推薦」系統之理論基礎。
The tide of electronic commerce is very powerful . When consumers shop in the virtual world,they face not only a lot of information but risk and uncertain. Although we can find information easily one the internet,consumers need more effective information filtering and recommendation which other people have similar interest for purchasing. Collaborative filtering can solve the problem above. In order to reduce the time consumer waste on the internet,this study apply collaborative filtering to develop a commodity recommendation metric. We divide this study into two.First,it is develop a commodity recommendation system based on collaborative filtering. The processes of this stage are built user profile database、cluster model and recommendation model. Because the study’s recommendation object is mobile phone,we adopt sex、shape、entertainment and bell as cluster filtering variables. The object of recommendation model is to recommend best mobile phone for consumer . Second,this study use field experiment to compare and verify recommendation system. According to the result of this study , the conclusions are listed as follows: 1.Collaborative filtering takes effect on commodity recommendation online. 2. The different cluster filtering variables have a good effect for commodity recommendation. 3. The cluster filtering variables have different effects on the moderator variables. The achievement of this study support electronic commerce operator to recommend commodity for consumer online. For the academic community,it helps them understand the effect between commodity recommendation and collaborative filtering.