No abstract

No abstract

This article explores the use of deep learning to choose an appropriate spatial partitioning technique for big data. The exponential increase in the volumes of spatial datasets resulted in the development of big spatial data frameworks. These systems need to partition the data across machines to be able to scale out the computation. Unfortunately, there is no current method to automatically choose an appropriate partitioning technique based on the input data distribution. This article addresses this problem by using deep learning to train a model that captures the relationship between the data distribution and the quality of the partitioning techniques. We propose a solution that runs in two phases, training and application. The offline training phase generates synthetic data based on diverse distributions, partitions them using six different partitioning techniques, and measures their quality using four quality metrics. At the same time, it summarizes the datasets using a histogram and well-designed skewness measures. The data summaries and the quality metrics are then use to train a deep learning model. The second phase uses this model to predict the best partitioning technique given a new dataset that needs to be partitioned. We run an extensive experimental evaluation on big spatial data, and we experimentally show the applicability of the proposed technique. We show that the proposed model outperforms the baseline method in terms of accuracy for choosing the best partitioning technique by only analyzing the summary of the datasets.

This article describes the UCR Spatio-temporal Active Repository (UCR-STAR). UCR-STAR is a visual catalog for big spatial datasets. Rather than a boring tabular listing of datasets, it provides an interactive map interface that allows users to explore these datasets to assess their coverage, quality, and distribution. This article describes both the functionality of UCR-STAR as well as the underlying system architecture. We believe that this article can help the research community by explaining how to realize research ideas into a workable product.

The rapid growth of big spatial data urged the research community to develop several big spatial data systems. Regardless of their architecture, one of the fundamental requirements of all these systems is to spatially partition the data efficiently across machines. The core challenges of big spatial partitioning are building high spatial quality partitions while simultaneously taking advantages of distributed processing models by providing load balanced partitions. Previous works on big spatial partitioning are to reuse existing index search trees as-is, e.g., the R-tree family, STR, Kd-tree, and Quad-tree, by building a temporary tree for a sample of the input and use its leaf nodes as partition boundaries. However, we show in this paper that none of those techniques has addressed the mentioned challenges completely. This paper proposes a novel partitioning method, termed R*-Grove, which can partition very large spatial datasets into high quality partitions with excellent load balance and block utilization. This appealing property allows R*-Grove to outperform existing techniques in spatial query processing. R*-Grove can be easily integrated into any big data platforms such as Apache Spark or Apache Hadoop. Our experiments show that R*-Grove outperforms the existing partitioning techniques for big spatial data systems. With all the proposed work publicly available as open source, we envision that R*-Grove will be adopted by the community to better serve big spatial data research.