科研社交网络中基于联合概率矩阵分解的科技论文推荐方法研究

近年来随着科研社交网络中科技论文数量爆炸式的增长,科研人员很难高效地找到与之相关的科技论文,因此面向科研工作者的科技论文推荐方法应运而生。然而,传统的科技论文推荐方法没有充分挖掘科研社交网络中广泛存在的社会化信息,导致科技论文推荐质量不高。为此,提出了一种科研社交网络中基于联合概率矩阵分解的科技论文推荐方法,在传统概率矩阵分解的基础上,融入了社会化标签信息和社会化群组信息来进行科技论文推荐。为了验证所提方法的有效性,抓取了科研社交网络CiteULike上的数据进行了实验。实验结果表明,与其它传统推荐方法相比较,所提方法在Precision和Recall两个评价指标上均取得了较好的推荐结果,并且能够应用于大规模数据集,具有良好的可扩展性。

Study of Scientific Paper Recommendation Method Based on Unified Probabilistic Matrix Factorization in Scientific Social Networks

In recent years,the number of scientific papers in scientific social networks has grown at an explosive rate.It is difficult for researchers to find scientific papers related to their research.Therefore,the paper recommendation for researchers was proposed to solve this problem.However,many problems exist in traditional paper recommendation methods,especially for the fact that a lot of social information in scientific social network are not fully used,resulting in poor quality of paper recommendation.Therefore,this research proposed a new paper recommendation method for researchers in scientific social networks based on the unified probability matrix factorization.This method incorporates social tag information and group information into traditional matrix factorization.In order to verify the validity of the proposed method,we crawled data from a famous scientific social network,i.e.CiteULike,to conduct experiments.Experimental results show that the proposed method gets the best recommendation results at the two evaluation metrics,i.e. Precision and Recall,compared to other traditional recommendation methods.The proposed method is linear with respect to the number of observed data,and performs well in scalability.