摘要
研究Web信息过载的问题,提出一种新的基于基因算法的信息免疫模型(IIM).根据免疫细胞的特异性,利用IIM不同的染色体描述用户需求,并专注于对无关信息的处理,使用户免于该类信息的入侵,并引入了特征选择和信息熵,阈值的选择也是可变的.通过实验与Rocchio方法进行了对比,结果表明,IIM的查准率比Rocchio的高27.5%,查全率比Rocchio的高47.7%.
Deals with the problem of information overload on the Web, and proposes a new information immune model (IIM) based on gene algorithm. According to the specificity of immune cells, IIM applies different chromosomes to describe user interests, emphasizes on blocking irrelevant information with variable thresholds. In order to construct efficient chromosomes, feature selection and information entropy are adopted. Finally, a prototype of IIM was developed and tested. Precision of IIM is 27\^5% higher than Rocchio, and recall of IIM is 47\^7% higher than Rocchio.
出处
《北京理工大学学报》
EI
CAS
CSCD
北大核心
2004年第12期1084-1087,共4页
Transactions of Beijing Institute of Technology
基金
国家部委预研项目(G1998030414)
关键词
信息免疫
基因算法
特征选择
熵
阈值
information immunity
gene algorithm
feature selection
entropy
threshold
