热点问题情报过滤模型与算法

热点问题情报过滤模型与算法 Hot Problem Intelligence Filter Model and Algorithm 陈磊 CHEN Lei 1 2 first-author 李德华 LI Dehua 1 2 华中科技大学图像识别与人工智能研究所 华中科技大学图像识别与人工智能研究所 Institute of Pattern Recognition and Artificial Intelligence,Huazhong University of Science and Technology Institute of Pattern Recognition and Artificial Intelligence,Huazhong University of Science and Technology 华中科技大学图像识别与人工智能研究所,湖北,武汉,430074 华中科技大学图像识别与人工智能研究所,湖北,武汉,430074 针对热点问题跨越时间长以及含重复情报、过时情报和无实用价值信息量大等问题,提出了一种新的热点问题情报获取模型与分级过滤算法,以解决情报的去粗取精、去伪存真问题.首先将无序关联情报结构化,然后采用SVM（space vector model）方法,根据热点问题情报的特殊性,提出了新的特征选择方法,在特征向量的相似性计算中引入去除重复和过时情报的判别参数,达到优化情报的目的.实验结果表明,过滤算法除去原始情报库中的无关情报、重复情报和过时情报后剩余情报数约为原始情报数的59%. Because hot problem intelligence spans a long period of time and contains much repeated intelligence,dated intelligence,huge amount of information that has no practical value,this paper put forward a new hot problem intelligence obtaining model and a ratings-filter algorithm to get rid of coarse and false information while keeping refined and true intelligence.Firstly unordered but related intelligence was arranged to be structural,and then SVM（space vector model） was applied to make a new feature selection method according to the specificity of hot problem intelligence.This method took into consideration repeated parameter and outdated parameter in calculating the similarities between feature vectors,thus intelligence could be optimized.The experiment results showed that the filtering algorithms could effectively remove repeated information and outdated information from the original intelligence library,and the remaining piece number of intelligence information was about 59% of the original number. 热点问题 情报获取模型 分级过滤 hot problem intelligence obtain model ratings filter TP301.6 国家高技术研究发展计划（863）项目（2003AA001032,2002AA111010） 2008-03-01 2021-04-01 3