基于权威度的指导者挖掘与个性化推荐方法

Authority-based approach for advisor mining and individual recommendation

在学术搜索引擎Arnetminer提供的数据中,对师生关系进行了挖掘,并在此基础上,结合用户信息,进行个性化的指导者推荐。计算出指导者的权威度和申请者的申请成功率。针对这两方面的内容,分别基于排序支持向量机模型和概率模型设计了基于权威度的推荐模型和基于个性化的推荐模型。研究成果成功应用于Arnetminer系统中,并可实时收集用户反馈信息以提高师生关系挖掘的准确率和推荐模型质量。

We mine advisor-advisee relations based on data of Arnetminer,an academic search engine.Furthermore,we make individual recommendations combining user＇s information.This paper concentrates on calculating two values,authority scores and success rates.We separately build a recommendation model based on authority scores and a recommendation model based on success rates.The application of the paper is practically applied in Arnetminer,and collects user feedback to increase the accuracy of advisor-advisee relation mining and improve two recommendation models＇ quality.