摘要
在信息检索服务中跨媒体检索技术日益重要,为了提高其检索的准确度,需要加强对不同模态之间语义信息的相互关系的学习和分析.早期的跨媒体检索技术侧重于对多媒体信息的底层特征的分析,而忽略了多媒体信息的底层特征与高层语义方面存在的联系.本文分析了多媒体信息在底层特征与高层语义之间的关联,根据不同模态对象的底层特征空间构造出同构的高层语义空间,将集成学习的方法应用到跨媒体检索之中.提出了Bagging-SM的方法对不同模态的多媒体对象进行语义匹配.实验结果表明该方法相比于其他方法,对跨媒体检索结果的准确度有很大的提升.
In information retrieval service,the cross-media retrieval technology has become increasingly important. In order to improve the accuracy of retrieval,it is necessary to strengthen the learning and analysis of the interrelationship of semantic information among different modalities. cross-media retrieval early focuses on the analysis of the low-level features of multimedia information,and ignores the relationship between the low-level features of multimedia information and high-level semantic aspects. In this paper,we analyze the association of multimedia information between low-level features and high-level semantics,construct high-level semantic spaces according to the low-level feature spaces of different modal objects,and apply ensemble learning methods based on different modalities to cross-media retrieval. The Bagging-SM method is proposed for semantic matching of multimedia objects with different modes. The experimental results show that the proposed method has a great improvement on the accuracy of cross-media retrieval results.
作者
陈祥
于治楼
Chen Xiang Yu Zhilou(School of Information Science and Engineering, Shandong Normal University, 250014, Jinan, China Inspur Group Ltd, 250101, Jinan, China Shandong Provincial Key Laboratory for Novel Distributed Computer Software Technology, 250014, Jinan, China)
出处
《山东师范大学学报（自然科学版）》
CAS
2017年第3期9-15,共7页
Journal of Shandong Normal University(Natural Science)
关键词
跨媒体检索
语义匹配
特征提取
集成学习
cross-media retrieval
semantic matching
feature extraction
ensemble learning
