Problem:
?? mining relationships in time series data; A new class of relationships in time series data.
traditional methods: discover pair-wise relationships.
Introduction:
Challenge: discovery of complex patterns of relationships between individual time-series. ?? what does this mean?
A common type of relationship in time series data is pairs of time-series with strong Pearson correlations.
In this paper, we define a novel relationship across three time series.
tripole: a group of three time-series, one root node/ two leaf nodes.
Note that a tripole is interesting only if correlation ofT0 withT1+2 is much stronger than correlation of T0 with T1 or T2 individually。
a formal study of tripole and explore their utility in different domains:
需要思考清楚，这个tripole pattern适用于什么domains之下，比如他可以适用于交通数据transportation data/ sea level pressure data 都和区域有关，那如果对于天气，生产，消费的time series关系是否适用，需要认真思考一下。但是我奇怪为什么要衡量T0 and T1+T2的相关性，这种现象在我们的数据中，有什么特殊的代表特征吗。
这种tripole现象，以期并没有进行系统的研究，因此，如何定义tripole,如何发现tripole，如何评估/解释发现的tripole成为了重要的问题。
define; measures to assess its interestingness; discovery with efficiency;
DEFINITIONS:
N time-series, {T1,T2,T3...Tn}
tripole: (T0: T1, T2); root; leaves;
Strength;
Note that while there could be other ways to combine the information in the two leaves, we chose the sum for its simplicity and ease of interpretation.
Jump; 用来判断这个tripole的有趣程度。
PROPOSED APPROACH:
how to discover tripoles in time-series data.
主要描述提出了节约计算的发现tripoles的方法。
EXPERIMENTAL RESULTS AND EVALUATION
两个数据库，海平面压力数据库，location, time series data. 神经影像学数据库，
用到了假设检验，p-value;
PHYSICAL INTERPRETATION OF TRIPOLES
Further validation and study of these tripoles by domain experts could possibly explain the phenomenon that results in the manifestation of these tripoles.
问题：tripole的物理意义应该由该领域的专家进行解释，而且是could possibly explain the phenomenon, 比较弱势的描述。
解释实例：大气中的遥相关，