Mining user daily behavior patterns from access logs of massive software and websites

ISCAS OpenIR

	Mining user daily behavior patterns from access logs of massive software and websites
	Wei, Zhao (1); Jie, Liu (2); Dan, Ye (2); Jun, Wei (2)
	2013
会议名称	5th Asia-Pacific Symposium on Internetware, Internetware 2013
会议日期	October 23, 2013 - October 24, 2013
会议地点	Changsha, China
收录类别	EI
出版地	Association for Computing Machinery, General Post Office, P.O. Box 30777, NY 10087-0777, United States
ISBN	9781450323697
部门归属	(1) Institute of Software, Chinese Academy of Sciences, University of Chinese Academy of Sciences, Beijing 100190, China; (2) Chinese Academy of Sciences, Institute of Software, Beijing 100190, China
摘要	Everyone has a characteristic pattern of daily activities. This study applies cluster analysis to identify a computer user's daily behavior patterns based on 1000 China users' 4-weeks software and web usage. Clustering models are built for 4 different behavior definition methods with different time period divisions and feature measurement selections. With these patterns, we build classification models to predict new users' daily behavior pattern with their half day activity logs. For example, if we know one user use computer for entertainment in the morning, we can predict his behavior in the afternoon and evening. The prediction model can be used to recommend suitable items to users according to their current behavior status. Our method can get 92.5% prediction correctness for the best.; Everyone has a characteristic pattern of daily activities. This study applies cluster analysis to identify a computer user's daily behavior patterns based on 1000 China users' 4-weeks software and web usage. Clustering models are built for 4 different behavior definition methods with different time period divisions and feature measurement selections. With these patterns, we build classification models to predict new users' daily behavior pattern with their half day activity logs. For example, if we know one user use computer for entertainment in the morning, we can predict his behavior in the afternoon and evening. The prediction model can be used to recommend suitable items to users according to their current behavior status. Our method can get 92.5% prediction correctness for the best.
语种	英语
内容类型	会议论文
URI标识	http://ir.iscas.ac.cn/handle/311060/16676
专题	中国科学院软件研究所
推荐引用方式 GB/T 7714	Wei, Zhao ,Jie, Liu ,Dan, Ye ,et al. Mining user daily behavior patterns from access logs of massive software and websites[C]. Association for Computing Machinery, General Post Office, P.O. Box 30777, NY 10087-0777, United States,2013.