Efficient Processing of Spatio-Temporal-Textual Queries

Abstract

Devices with built-in GPS (e.g. smartphones) are producing a huge amount of data objects with spatial, temporal and textual information. For example, a significant part of Twitter messages sent from smartphones has spatial location (latitude and longitude), temporal information (timestamp) and textual information (the message itself). Therefore, there is a growing interest for new approaches that are able to select the data objects that are spatially, temporally and textually relevant from huge datasets. In this paper, we specify the spatio-temporal-textual query that returns the relevant data objects considering these three criteria simultaneously, presenting new indexes and algorithms to process such query efficiently. The proposed approaches are evaluated taking real datasets, potentially providing more accurate results.