基于重引力搜索链接预测和评分传播的大数据推荐系统

大数据推荐系统的搜索空间较大导致推荐的响应时间过长。为权衡大数据推荐系统的时间效率和推荐性能,提出一种基于重引力搜索链接预测和评分传播的大数据推荐系统。采用相对相似性指数度量用户的相似性,采用广义Meta Path模型建立相似图;引入社区信息来提高局部链接预测的准确率,从强社区提取优化的子图来实现局部链接的预测,通过重引力搜索对子图做优化处理,从而缩小搜索空间;设计基于传染病模型的网络传播策略,根据已有的模式探索隐藏的模式。基于公开数据集的实验结果表明,该算法有效地提高了推荐系统的准确率和覆盖率,并且响应时间在可接受的范围内。

BIG DATA RECOMMENDER SYSTEM BASED ON GRAVITATIONAL SEARCH FOR LINK PREDICTION AND RATINGS PROPAGATION

The large search space of big data recommender systems causes long response time of recommendations.To balance the time efficiency and performance of big data recommender systems,we propose a big data recommender system based on gravitational search for link prediction and ratings propagation.We adopted the relative similarity index community information to improve the accuracy of local links prediction,and predicted local links according to sub-graphs extracted from strong communities,and it optimized sub-graphs by gravitational search to reduce the search patterns according to the existing patterns.Experimental results based on public datasets indicate that the proposed algorithm improves the accuracy and coverage rate for the recommender systems,and realizes an acceptable response to measure the similarities of users,and used generalized Meta Path model to construct similarity graph;it introducedspace;we designed a network propagation strategy based on the infectious susceptible model,and explored the hidden time.