ReliefF内容相似度的雷达情报按需分发  

Radar Intelligence Distribution on Demand Based on Relief F Content Similarity

作　　者：古秦弋 杨瑞娟[2] 黄美荣[2] 杨云飞[1] 叶伟 李玥 GU Qin-yi;YANG Rui-juan;HUANG Mei-rong;YANG Yun-fei;YE Wei;LI Yue(Air Force Early Warning Acadamy, a. Department of Graduate Management;b. No. 4 Department Hubei Wuhan 430019, Chin;2. PLA, No. 92261 Troop, Hainan Haikou 570100, Chin;3. PLA, No. 93986 Troop, Xinjiang Hetian 848000, China)

机构地区：[1]空军预警学院研究生管理大队,湖北武汉430019 [2]空军预警学院4系,湖北武汉430019 [3]中国人民解放军92261部队,海南海口570100 [4]中国人民解放军93986部队,新疆和田848000

出　　处：《现代防御技术》2018年第3期184-190,共7页Modern Defence Technology

摘　　要：基于内容相似度的情报按需分发通过相似度计算形成用户的信息推荐,但是传统的相似度算法假定所有特征具有相同的重要程度,与目前已知事实相悖。为此,提出一种特征加权的内容相似度计算方法,该方法考虑各维特征对结果的不同影响,利用Relief F算法赋予特征权值,并根据特征的重要程度进行特征选择,最后通过计算情报之间的内容相似度,形成基于加权内容相似度的最近邻区,在此基础上预测新情报对用户的推荐度,并按兴趣程度实现雷达情报的按需分发。实验表明,该算法具有较好的推荐性能,预测效果优于传统的内容相似度推荐方法。Radar intelligence distribution on demand based on content similarity forms user information recommendation by computing the similarity. But the traditional similarity algorithms assume that all characteristics have the same importance,which conflicts with known facts. To this end,a content similarity calculation method based on feature weighting is proposed. The method considers the different influence on the results by various dimensional characteristics,uses Relief F algorithm to give the feature weights,selects features according to the characteristics of the importance,and forms the nearest neighbor area based on weighted content similarity through calculating the content of the similarity. On this basis,the recommendation of new information to the user is predicted,the radar intelligence distribution on demand according to the degree of interest is realized. The simulation results show that the algorithm has better performance and better prediction effect than the traditional content similarity recommendation method.

关 键 词：雷达情报按需分发 内容特征 内容相似度 加权 欧几里德距离 特征选择 

分 类 号：TN957.51[电子电信—信号与信息处理]