Abstract
Due to their multiple sources and structures, big spatial data require adapted tools to be efficiently collected, summarized and analyzed. For this purpose, data are archived in data warehouses and explored by spatial online analytical processing (SOLAP) through dynamic maps, charts and tables. Data are thus converted in data cubes characterized by a multidimensional structure on which exploration is based. However, multiple sources often lead to several data cubes defined by heterogeneous dimensions. In particular, dimensions definition can change depending on analyzed scale, territory and time. In order to consider these three issues specific to geographic analysis, this research proposes an original data cube metamodel defined in unified modeling language (UML). Based on concepts like common dimension levels and metadimensions, the metamodel can instantiate constellations of heterogeneous data cubes allowing SOLAP to perform multiscale, multi-territory and time analysis. Afterwards, the metamodel is implemented in a relational data warehouse and validated by an operational tool designed for a social economy case study. This tool, called “Racines”, gathers and compares multidimensional data about social economy business in Belgium and France through interactive cross-border maps, charts and reports. Thanks to the metamodel, users remain independent from IT specialists regarding data exploration and integration.
Highlights
Big data has become a very active research field considering the fast increasing of data sources diversity: sensors, smartphones, crowdsourcing, social networks, open databases, etc
This research aimed at developing a business intelligence tool dealing with three important issues of geographic analysis related to heterogeneous and multidimensional data: multiscale analysis, multi-territories analysis and time analysis
Based on a review of literature related to economic geography, business intelligence, (S)online analytical processing (OLAP) and more its modeling aspects, we formulated this hypothesis to meet our research objective: the design of a unified modeling language (UML) metamodel able to instantiate spatial data cubes sharing common dimension levels
Summary
Big data has become a very active research field considering the fast increasing of data sources diversity: sensors, smartphones, crowdsourcing, social networks, open databases, etc These numerous and large datasets are characterized by heterogenous semantics, structures and formats leading to time-consuming processes for management and analysis purposes. Around 80% have a spatial component [1] This opens the door to geographic analysis and its specific issues related to heterogeneous data. Geographic analysis is very interdisciplinary because it can be performed in numerous fields involving spatial data: marketing, criminology, archeology, ecology, oceanography, urban planning, etc All these fields have their own experts who might need to analyze and explore big geospatial data.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have