摘要
为了提升雷达情报系统的情报分发性能,研究个性化推荐算法在雷达情报按需推送上的应用.针对传统相似度算法对所有特征一视同仁,造成重要特征作用不突出的问题,提出一种基于Relief F加权内容相似度的雷达情报推荐算法.该算法选择主要特征,形成基于加权内容相似度的最近邻集合,再通过推荐预测算法并设定评价阈值形成雷达情报推荐.实验结果表明,基于加权内容相似度的情报推荐算法能较好地完成雷达情报推荐,其推荐性能优于传统内容相似度算法;合理设置算法中最近邻数目和评价阈值可使准确率和召回率均较高.
In order to improve the intelligence distribution performance of radar intelligence system, thispaper does a research on the application of personalized recommendation algorithm in radar intelligenceon-demand push. Aiming at the problem that the traditional similarity algorithm treats all features equally so as tomake the important features indistinctive, the paper proposes a radar intelligence recommendation algorithm basedon ReliefF weighted content similarity. The proposed algorithm selects the main features to form the nearestneighbor set based on the similarity of the weighted content, and then uses the radar intelligence recommendationand sets the evaluation threshold to form radar intelligence recommendation. The experimental result shows thatthe proposed algorithm can better complete the radar intelligence recommendation and its recommendationperformance is better than the traditional content similarity algorithm;setting the nearest neighbor number andevaluation threshold in the algorithm makes accuracy and recall rate higher.
出处
《空军预警学院学报》
2017年第6期427-431,共5页
Journal of Air Force Early Warning Academy
关键词
雷达情报分发
个性化推荐
内容相似度
加权内容相似度
内容特征
radar intelligence distribution
personalized recommendation
content similarity
weighted contentsimilarity
content characteristics
