基于概念网的媒体大数据分析和结构化描述方法

[摘要] 提出基于概念网的媒体大数据结构化描述和分析的技术框架，该框架可以针对不同的数据获取来源，通过层次式多角度概念描述模型融合数据的视觉特征、实例和概念关联的语义，并提出面向单一媒体和多媒体文档的跨媒体概念提取及基于结构的语义对齐方法，从而有效支持媒体大数据的语义关联分析及多领域的智能应用。

[关键词] 概念网；媒体大数据分析；概念抽取；结构化描述；可视化

[Abstract] In this paper, we propose that a topic network-based enabling technology framework for big media analysis and structural description. And it proposes a hierarchical concept description model with multiple perspectives for different sources data to integrating semantic of visual, instance and concept correlation. And cross-media concept extraction method for single media and multimedia document and their structure-based semantic alignment method are also proposed, which can efficiently support the big media analysis and smart application in many domain.

[Keywords] topic network; big media analysis; concept extraction; structural description; visualization