社會網路服務推薦機制之研究

Abstract

隨著社交網路的蓬勃發展，許多利基於社交網路上的服務，例如： 社交網站上的application、部落格或個人入口首頁上的widgets和gadgets等等皆成長快速且多元。為了能有效地為社交網路上的使用者篩選出適合的服務，我們透過分析服務的熱門度與信譽、使用者個人喜好與其社交關係等三個面向，並利用倒傳遞類神經網路來模擬使用者的決策條件，建構出一個系統化的社會網路服務推薦機制。本實驗實作於全球著名的社交網路平台Facebook上；實驗結果顯示所提出的機制優於其他的方法，同時發現社交關係在社會網路服務的推薦上比使用者自身的喜好與服務的熱門度和信譽更為重要。

Social network based services, such as applications on the social network websites, widgets on blogs, and gadgets on personal portals have grown dramatically in a tremendous amount. In order to efficiently recommend suitable and attractive social network based services to users, a systematical recommendation mechanism composing of service’s popularity and reputation, user’s preference and social relationship is proposed. A back-propagation neural network is applied to optimally model general users’ decision making criteria of using social network based services. This recommender service is implemented to one of the most famous social network websites- Facebook. The experimental result shows that the proposed model outperforms than any other methodology, including Analytic Hierarchy Process. It is also found that social relationship plays the most important part in recommendation of social network based service, instead of user’s preference or service’s popularity and reputation.