Opinion Mining for Emergency Case Risk Analysis in Spark Based Distributed System.

Xiao LongDeng,Ya Qi Tang,Yi Hua Huang
DOI: https://doi.org/10.1145/2835596.2835608
2015-01-01
Abstract:The 2014 West African Ebola Outbreak rised from Sierra Leone, Guinea, and Liberia at December in 2013 has been reported to cause 21296 cases and 8429 deaths until now. Besides that, Typhoon Haiyan and Typhoon Hagupit are also the serious disaster accidents for emergency management in recnt years. In order to collect and analyze these accidents in more details, we need to design and adopt some new ideas and models. In this paper, we used Spark based Distributed mining system for online opinion data collecting and mining. We crawled related online social media data of Ebola from the most four favorite online social networks (including SINA, TENCENT) in China from June to November in 2014. By analyzing these attained social media data and airline data of GEM model, we found some interesting results. Furthermore, we collected the online social media data of Typhoon Haiyan and Hagupit before and after their landing. And we also showed analysis of combining social network data with geographical demonstration and Chinese citizen sentiment towards these disasters. As a result, we offered some suggestions to the government to handle opinion situation in these disasters.
What problem does this paper attempt to address?