地学时序大数据的相似性度量与索引方法综述

Review on geoscience time series big data similarity measurement and index method

地学时序大数据具有多传感器、多目标、多分辨率、多类型的多源异构特征,是地学领域机器学习与数据挖掘的重要数据来源,分为基于时点和基于时段的两大类时序数据。现有时序数据的相似性度量与索引研究主要聚焦在前者。时序数据表达方法的核心思想是降维处理,是相似性度量与索引方法的基础,主要包括基于域变换和模型的表达方法和基于极限分段思想的表达方法。相似性度量的核心是相似性距离计算,主要分为锁步度量和弹性度量。它为时序数据索引中索引项的聚合与划分提供了基本准则。多源异构地学时序大数据的高效相似性度量与分布式索引方法是地学大数据领域未来的重要研究方向。

Geoscience time series big data is a kind of multi-sensor,multi-target,multi-resolution,multi-type,and multi-source heterogeneous time data,and is an important data source for machine learning and data mining in the field of geosciences.Geosciences time series data has two categories:the time point data set,and the time interval data set.The main data representation methods,similarity measurements and data indexing methods of existing time series data focus on time-points-based time series data.The core idea of the representation method for time series data is dimensionality reduction.It is the basis of similarity measurement and indexing method,including domain-transformation-based,model-based and pieces-based representation methods.The key of similarity measurement is the similarity distance including lock-step measurement and elasticity measurement.It provides a basic guideline for the aggregation and division of index items in the index of time series data.The efficient similarity measurement and distributed indexing method of multi-source heterogeneous time series big data will be an important further direction in the field of geosciences big data.