摘要
针对新闻推荐中新闻文本和用户兴趣建模因子的单一性以及用户兴趣漂移的问题,提出基于内容推荐和时间函数结合的新闻推荐算法。从多个角度对用户兴趣进行建模,同时考虑用户兴趣的实时性,通过时间函数来调整长短期的用户兴趣权重。实验结果表明:论文提出的算法较之传统算法具有更好的推荐性能,在precision、recall、F1-Measure指标上分别提高5.2%、3.9%、和8.4%。
In news recommenation,to solve the problem of single element that construct the modelof news text and user inter⁃est,and the change of user interest,news recommendation algorithm based on content-based recommendation and time function is proposed.The method construct user interest form multiple perspectives,considering the real-time performance of user interest.The method adjust the weight of long-term and short-term interest of the user interest by using time function,the experimental results show that this news recommendation algorithm is more effective in recommendation than traditional algorithm,with an incresase of 5.2%in precision,an increase of 3.9%in recall and an increase of 8.4%in F1-Measure.
作者
翁海瑞
林穗
何立健
WENG Hairui;LIN Sui;HE Lijian(School of Computers,Guangdong University of Technology,Guangzhou 510006)
出处
《计算机与数字工程》
2020年第12期2973-2977,共5页
Computer & Digital Engineering
基金
广州市科技计划项目(编号:2017010160012)资助。
关键词
基于内容推荐
新闻推荐
时间函数
content-based recommendation
news recommendation
time function
