Abstract : Due to the unexpected context of disaster management (with heterogeneous and non-dedicated potential data sources), the classical Big-Data approaches within this business domain (schematically focused on data storage and pattern recognition) appear to be limited, especially when trying to get a situational level of such crises from data gathered from the field. That is why this article aims at discussing a specific vision of Big-Data for data management, in two steps: (i) analyzing this business domain to identify relevant characteristics, impacted or concerned by Big-Data, and describe the new key challenges that need to be tackled, and (ii) designing an innovative Big-Data framework dedicated to this particular business domain. After having highlighted the importance to push abstraction levels and especially data interpretation as a way to perform vertical intelligence in data analysis (instead of horizontal intelligence with usual approaches), the proposed Big-Data framework brings a layered approach according to three dimensions: gathering (data level), interpretation (information level), exploitation (knowledge level).
https://hal.inria.fr/hal-01614598
Contributor : Hal Ifip <>
Submitted on : Wednesday, October 11, 2017 - 10:40:50 AM Last modification on : Wednesday, June 24, 2020 - 4:19:31 PM Long-term archiving on: : Friday, January 12, 2018 - 2:02:54 PM
Frederick Benaben, Aurelie Montarnal, Audrey Fertier, Sébastien Truptil. Big-Data and the Question of Horizontal and Vertical Intelligence: A Discussion on Disaster Management. 17th Working Conference on Virtual Enterprises (PRO-VE), Oct 2016, Porto, Portugal. pp.156-162. ⟨hal-01614598⟩