中文摘要：

社交网络和网络购物的发展普及导致了社会化导购的产生和发展,也催生了通过在社交网络中推荐产品从而获取利润的“橙领”.通过对橙领相关技术的研究,能更透彻地了解基于社会网络的产品营销机制以及探索社会化导购的底层模式.目前国内外少有这方面研究.因此,文中针对橙领的自身定位问题和面向用户或商家的橙领推荐问题,提出一种针对社会化导购的橙领推荐方法,主要包括3个算法：橙领定位算法、面向用户的橙领推荐算法OCRA4U（Orange Collar Recommending Algorithm for User）和面向商家的橙领推荐算法OCRA4S（Orange Collar Recommending Algorithm for Shop）.橙领定位算法依据橙领的推荐历史对橙领进行定位特征向量化描述,最终转化为一个聚类问题进行解决.OCRA4U考虑了橙领在社交网络中的影响力和橙领与用户需求的匹配度,得到橙领推荐列表.OCRA4S结合橙领在网络中的影响力以及橙领的历史推荐产品,推荐出最符合商家产品需求的橙领列表.基于新浪微博数据集和DBLP数据集,文中设计并实现了3个相关实验：橙领定位算法实验、橙领推荐实验以及社会化数据影响实验,实验结果验证了所提算法的准确性和可行性.

英文摘要：

Social shopping guiders appear and develop with the development of the social network and online shopping. This also leads to the emerging of orange-collar who gets income by recommending products to other persons in SNS in China. Through studying orange-collar related technologies, we can not only learn the mechanism of online product marketing thoroughly, but also explore the deep level patterns of social sales. However,current researches seldom focus on this issue. Hence, to solve the problem of orientating the orange-collars and recommending them to users or shops, this paper proposes an orange-collar recommending method, which includes three algorithms that are orange-collar positioning algorithm, OCRA4U algorithm and OCRA4S algorithm. The orange-collar positioning algorithm describes orange-collar by orientation vector based on the varieties of the products in its recommending history and finally transforms to a clustering problem to solve. OCRA4U takes the orange-collar＇s network influence and the matching degree between the orange-collar and user＇s need into consideration and returns an orange-collar recommending list to user. OCRA4S combines the orange-collar＇s influence and its recommending history, and finally gains the most satisfied orange-collar recommendation for the product need. Based on Sina microblog dataset and DBLP dataset, this paper has designed and implemented three experiments, the orange-collar positioning experiment, the orange-collar recommending experiment and the social data influence experiment, whose results have proved the correctness and effectiveness of the proposed algorithms.