SmarT: Machine Learning Approach for Efficient Filtering and Retrieval of Spatial and Temporal Data in Big Data

Spatiotemporal data has always been big data. In these days, big data analytics for spatiotemporal data is receiving considerable attention to allow users to analyze huge amounts of data. Traditional big data platforms cannot handle all the challenges of processing spatio-temporal data. Although some big data platforms have been proposed to process a massive volume of spatiotemporal data, neither is considered a clear winner for all possible scenarios. This paper presents the SmarT query engine, a machine learning-based solution that chooses the best big data platform for processing spatiotemporal queries on the fly. In a detailed experimental evaluation, considering the Apache Spark, Elasticsearch, and SciDB big data platforms, the response time decreased up to 22% when using SmarT.