摘要
现有网页分类技术忽略用户个性行为的差异。为此,提出一种结合用户行为特征分析的网页分类技术。运用知识规则发现、页面特征提取等方法,分析Web用户的访问历史和个性化定制信息,学习并掌握用户的行为和兴趣。针对用户的认知特征,提供合适的Web页面分类模式,能在一定程度上改进单纯统计学网页分类方法在自然语言理解上的不足。实验数据表明,该分类方法与多种统计学方法相结合实施网页分类均能有效地提高分类准确率,使网页分类结果更接近分类的真实情形和要求。
This paper introduces a kind of Web page categorization technology through analysis of characters of users' behavior,along with current hotspot of researching on Web pages categorization.Trough grasping users' behavior and interest by analyzing the history of Web user's access,and by concluding knowledge rules out also with pages' characters distilled.It provides a kind of appropriate categorization pattern on Web pages based on users' knowledge level,and surely improves classifying effect without language meanings understood contrast with pure statistic categorization.Experimental results indicate that this pattern of categorization combining kinds of statistic algorithm can improve accuracy of categorization,and make the classifying results more closer to real facts and people's knowledge desire.
出处
《计算机工程》
CAS
CSCD
2012年第20期179-183,共5页
Computer Engineering
基金
国家自然科学基金资助项目(60473142)
安徽省高校省级自然科学研究基金资助重点项目(KJ2010A051
KJ2011A039)
安徽省高校省级优秀青年人才基金资助项目(2009SQRZ076)
关键词
网页分类
行为特征
数据挖掘
逆向推理
关联规则
序列模式
Web page categorization
behavior characteristic
data mining
reverse-reasoning
association rule
sequence pattern