摘要
研究食品安全领域跨媒体数据的主题分析技术,融合多种媒体形式数据的语义,准确表达跨媒体文档的主题。由于食品安全事件相关多媒体数据的大量涌现,单一媒体的主题分析技术不能全面反映整个数据集的主题分布,存在语义缺失、主题空间不统一,语义融合困难等问题。提出一种跨媒体主题分析方法,首先以概率生成方法分别对文本和图像数据进行语义分析,然后利用跨媒体数据间的语义相关性进行视觉主题学习,建立视觉主题模型,进而实现视觉数据和文本主题之间的映射。仿真结果表明,跨媒体主题分析方法能够有效获取与图像语义相关的文本主题,且主题跟踪的准确度优于文本主题跟踪方法,能够为食品安全事件的监测提供依据。
Research cross - media topic analysis methods, which utilize semantic of multimedia data to describe topics of cross - media documents. As the emerging of food safety related multimedia data, topic analysis based on single media data cannot obtain full topics, causing the problem of inadequacy of semantic. In this paper, a cross - media topic analysis method was proposed. Firstly, generative methods were used to get the semantic of text and im- age data respectively. Then a visual topic learning method was presented to construct visual topic model and map vis- ual data to text topics. The results of simulations show that the proposed method can obtain topics of image data effi- ciently. And the accuracy of topic tracking based on this method is better than text topic tracking and meets the needs of food safety emergency monitoring.
出处
《计算机仿真》
CSCD
北大核心
2012年第6期1-4,36,共5页
Computer Simulation
基金
国家重点基础研究发展计划(973计划)(2012CB821206)
国家自然科学基金资助项目(91024001)
国家自然科学基金资助项目(61070142)
北京市自然科学基金资助项目(4111002)
关键词
跨媒体
主题分析
食品安全
Cross - media
Topic analysis
Food safety
