摘要： 提出一种基于符号图的高维时间序列数据库索引方法，通过等分空间法对高维时间序列数据进行预处理，将其转化为高维符号序列，利用符号状态转换图对数据库进行索引。应用CMU Graphics lab发布的Motion Capture Database动作数据库进行实验，取得了较好的结果。

关键词: 索引, 符号化, 高维, 熵

Abstract: This paper presents an index method for high-dimensional time series databases based on symbol graph. It symbolizes the high- dimensional data sequence into high-dimensional symbol sequence based on labanotation as a pre-process. The database is indexed by using symbol state transformation graph. Experiments on Motion Capture Database provided by CMU Graphics Lab show the effectiveness of this method.

Key words: index, symbolization, high-dimensional, entropy

中图分类号: 

• TP393