Metric-temporal access methods

Abstract

Metric-temporal databases are a new database model that combines metric spaces with temporal databases to process similarity queries within a time interval or snapshot. The Historical FHQT is a metric-temporal index which has shown to be competitive answering this type of queries. This index store a list of valid snapshots where each one contains an Fixed Height Queries Tree that indexes all objects existing at that instant. In this paper we present an improvement to this access method that consists in using different sets of pivots for the Fixed Height Queries Tree that correspond to consecutive time instants. The experimental results show this modification improves the filtering capacity of the index.