Speakers Bio & Abstract

Biography
Andreas Abecker holds a diploma in Computer Science from Technical University Kaiserslautern and a PhD in Applied Informatics from Karlsruhe Institute of Technology. He worked 9 years as a researcher and project manager at the DFKI German Research Center for Artificial Intelligence and 6 years as the department manager for ?Knowledge Management and Semantic Technologies? at the FZI Research Center for Information Technologies. Since 2010, he coordinates the research and innovation projects of Disy.Abstract
Toward an Architecture for Processing Spatal Big DataThe project 'BigGIS - Prescriptive and Predictive GIS Based on High-Dimensional Spatio-Temporal Data Structures' investigates methods and tools for management and analysis of heterogeneous spatial data. The project is driven by three exemplary usage scenarios: (1) Smart City and Human Health (considering urban heat islands and particulate matter); (2) Environment Monitoring (with a focus on invasive species); (3) Civil Security (regarding toxic gas clouds). Based on these scenarios' requirements and a first analysis of available geodata sources, a project architecture has been defined as a Spatial Big Data Infrastructure. The architecture is based on the idea of lambda architecture for big data solutions. It allows for specific data pre-processing pipelines for geodata ETL processes using the R library. The architecture combines a bulk-data store for long-term and high-volume data persistency with a high-end in-memory database for high-speed real-time processing. Special emphasis is laid upon (i) pragmatic and efficient treatment of huge remote-sensing data and (ii) on powerful semantic metadata mechanisms for dealing with varying levels of data reliability (e.g., when dealing with user-generated data) and for dealing with heterogeneous data. In the talk, we will present the BigGIS project, discuss the project architecture and explain the major design decisions and underlying rational.