期刊文献+

基于多维度特征的饮食健康推荐 预览

MULTI-DIMENSIONAL RECOMMENDATION ON HEALTHY DIET
在线阅读 下载PDF
收藏 分享 导出
摘要 为帮助人们判断饮食是否合理或营养是否均衡,进而改善饮食结构,合理化不同营养元素的摄入量,提出一种基于用户画像、饮食记录和即时情境多维度特征的健康饮食推荐算法。基于用户画像及用户的静态特征,制定不同人群、疾病的膳食指南,得到满足该用户身体状态的推荐列表;基于饮食记录,采用聚类方法和改进的距离计算方法,挖掘用户的营养结构和口味偏好,能够合理有效地进行饮食推荐;基于即时情境,通过外部特征对前两种算法推荐列表进行重排序,产生符合当前用户状态的菜品。实验结果表明,多维度特征推荐算法在实验平台上准确率达到实际工程应用的水平。 In order to help people judge whether the diet is reasonable or nutritionally balanced,and then improve the diet structure and rationalize the intake of different nutrients,we proposed a healthy diet recommendation algorithm based on user portraits,dietary records and multi-dimensional features of instant situations. Based on the portrait and the static characteristics of the users,we formulated the dietary guidelines for different populations and diseases,and obtained the recommended list to meet the user’s physical condition. Based on the dietary records,we used the clustering method and the improved distance calculation method to mine the user’s nutritional structure and taste preference,which could reasonably and effectively recommend diet. According to the instant situation,the first two algorithms recommendation lists were reordered by external features to produce dishes that conform to the current user’s taste. The experimental results show that the accuracy of multi-dimensional feature recommendation algorithm reaches the level of practical engineering application.
作者 喻兵 郝梓杰 罗裕升 朱珍民 Yu Bing;Hao Zijie;Luo Yusheng;Zhu Zhenmin(Xiangtan University,Xiangtan 411105,Hunan,China;Institute of Computing Technology,Chinese Academy of Sciences,Beijing 100080,China)
出处 《计算机应用与软件》 北大核心 2019年第7期110-116,共7页 Computer Applications and Software
关键词 饮食 用户画像 饮食记录 即时情境 聚类 Diet User portrait Dietary records Instant situation Cluster
作者简介 喻兵,硕士生,主研领域:普适计算,移动医疗与健康;郝梓杰,硕士;罗裕升,硕士生。
  • 相关文献

参考文献6

二级参考文献22

共引文献13

投稿分析

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部 意见反馈