摘要
新闻视频是一种丰富的信息源,对其进行有效的挖掘将有力地辅助决策分析。本文从多媒体信息系统的角度出发,深入思考了辅助决策的新闻视频的挖掘内容和挖掘方法,主要解决了以下两个问题:即挖掘什么和如何去挖的问题。指出辅助决策的新闻视频挖掘的内容应包括结构挖掘、语义事件挖掘、趋势挖掘、关联规则挖掘以及决策者兴趣点挖掘等等;并以一个具体的例子来说明新闻视频挖掘的各种方法,如结构的分割、语义事件的探测、新闻故事重要度的判定、专题的统计分析以及模式的可视化等等。本文较好地实现了新闻视频的内容分析与挖掘的结合,初步的实验证明了提出的新闻视频挖掘方法在辅助决策方面具有有效性。
In this paper,we ponder upon the problem of news video mining supporting decision-making. Firstly,we raise two problems: mine what and how to mine? Secondly ,we answer these questions in detail. We indicate that news video mining should include mining structure ,semantic event ,tendency ,association-rule and decision-makers' interest- s,etc. Then by a lively example,we explain various news video mining methods such as structure segmentation, se- mantic event detection,important news stories determination,special topic statistical analysis,and pattern visualiza- tion,etc. This paper combines news video analysis methods with data mining methods perfectly,and the primary ex- periments prove the effectiveness of the proposed news mining methods in supporting decision-making.
出处
《计算机科学》
CSCD
北大核心
2004年第8期136-139,共4页
Computer Science
基金
国家863(2001AA115123)项目基金资助。