南京大学学报（自然科学版）Issue(4)：826-833,8.DOI:10.13232/j.cnki.jnju.2015.04.020
一种结合文化和因子分解机的快速评分预测方法
A fast rating predicting method based on the culture and factorization machine
摘要
Abstract
With the development of big data,the traditional recommendation filtering techniques are confronted with some difficulties,such as high dimensional data,low recommendation accuracy,and sparse data.In this light,we build a model called CFM,which is the combination of factorization machine and the user’s cultural background mined from the social media.Culture difference means that in different cities,regions or nation,the user’s interests are different in the same things affected by the cultural background.And with the advantages of factorization machine such as linear time complexity and nice commonality,CFM is a kind of general score prediction method, which is not limited to specific areas without expert knowledge.In a large number of experimental results on real data set,the accuracy of CFM is significantly higher than the traditional methods.关键词
文化/大数据/兴趣推荐/因子分解机Key words
culture/big data/interesting recommendation/factorization machine分类
信息技术与安全科学引用本文复制引用
胡亚慧,李石君,余伟,杨莎,方其庆..一种结合文化和因子分解机的快速评分预测方法[J].南京大学学报（自然科学版）,2015,(4):826-833,8.基金项目
国家自然科学基金(61272109),中央高校基本科研业务费专项资金(2042014kf0057),湖北省自然科学基金(2014CFB289),空军预警学院青年创新基金(2013ZDJC0101) （61272109）
