隨著社群網站的日益興起，在網路世界進行交流的人數急速增長，也逐漸成為將產品推薦給使用者的媒介之一。 過去傳統微網誌(Microblog)的產品推薦方式，大多只在乎使用者個人喜好與興趣，卻忽略了其他可能的影響因素，例如使用者對廠商的喜好、產品受歡迎程度、產品類別喜好、廠商名聲及產品上市時間，並且在過去文獻當中也較少同時將上述變數進行研究分析。因此本文擬針對上述五項不同影響因素，去探討其對使用者的偏好是否造成顯著影響進而提升產品推薦的效果。 本文實證結果顯示， Plurk的使用者對於產品的廠商喜好、產品受歡迎程度與產品類別喜好具有顯著影響關係，會去影響到產品頁面的點擊數。此外，本研究先以回歸模型進行顯著性分析，挑選變數。再透過類神經網路來預測使用者對推薦商品的點擊率是否具有穩健性，實證結果發現類神經具有較好的預測效果。回歸分析所挑的變數也的確有很好的預測力。;With the growing popularity of the social network, the number of people using the social network to communicate and interactive with others increased steadily. As a result, social commerce has become a new phenomena.
In the past, most of the product recommendation in Microblog only deal with personal preferences and interests, and ignores other possible factors such as crowd Interest, Popularity of products, reputation of creators, types of preference and recency. These variables are used by facebook to recommend posts to users. Therefore, this research adapted the five aspects and analyzed their effectiveness to recommend products on social media sites.
The empirical results show that the Interest, Popularity and Type have significant impacts on recommendation effectivness. In addition, this studies also utilized Artificial Neural Networks to predict the click through rates of recommended web pages. The results show that the Artificial Neural Networks have better predictive effect then Linear Regression. However, the three variables identified by Linear Regression indeed outperform the other variables.
