基于GraphX的社交网络用户推荐算法研究

Research on Recommendation Algorithm for Social Network Users Based on GraphX

针对PageRank等传统算法在分析大规模分布式集群数据过程中存在耗时长、推荐不精准等问题,提出一种基于GraphX的社交网络用户推荐算法,以期提升用户体验。综合搜索引擎中的相互超链接计算技术,采用PageRank算法和GraphX组件中的Triangle Counting算法等建立评估模型,并利用该模型用户间的活跃度和网络关联度等关键参数来获取用户好友推荐表。通过Sougou数据对模型进行验证,并与单一的PageRank算法模型进行对比分析,结果表明:算法评估模型运行速度和推荐率有显著提升,推荐用户好友更接近真实情况。

For PageRank on the analysis of the traditional methods,such as large-scale distributed cluster data in the process of some problems such as time-consuming,recommend not accurate,put forward a kind of social network users recommendation algorithm based on GraphX,in order to improve the user experience.Comprehensive calculation of mutual hyperlinks of search engine technology,using PageRank algorithm and GraphX components Triangle Counting algorithm of the evaluation model is set up,etc,the user＇s friend recommendation form is obtained by using the active degree of the model and the network relational degree.The model is validated through the Sougou data,and a single model of PageRank algorithm is compared and analyzed,the experimental results show that the speed and recommendation rate of the model are improved significantly,and the recommendation of users is closer to the real situation.