Development of SNS Stream Analysis Based on Forest Disaster Warning Information Service System
Abstract
Forest disasters, such as landslides and wildfires, cause huge economic losses and casualties, and the cost of recovery is increasing every year. While forest disaster mitigation technologies have been focused on the development of prevention and response technologies, they are now required to evolve into evacuation and border evacuation, and to develop technologies fused with ICT. In this study, we analyze the SNS (Social Network Service) stream and implement a system to detect the message that the forest disaster occurred or the forest disaster, and search the keyword related to the forest disaster in advance in real time. It is possible to detect more accurate forest disaster messages by repeatedly learning the retrieved results using machine learning techniques. To do this, we designed and implemented a system based on Hadoop and Spark, a distributed parallel processing platform, to handle Twitter stream messages that open SNS. In order to develop the technology to notify the information of forest disaster risk, a linkage of technology such as CBS (Cell Broadcasting System) based on mobile communication, internet-based civil defense siren, SNS and the legal and institutional issues for applying these technologies are examined. And the protocol of the forest disaster warning information service system that can deliver the SNS analysis result was developed. As a result, it was possible to grasp real-time forest disaster situation by real-time big data analysis of SNS that occurred during forest disasters. In addition, we confirmed that it is possible to rapidly propagate alarm or warning according to the disaster situation by using the function of the forest disaster warning information notification service. However, the limitation of system application due to the restriction of opening and sharing of SNS data currently in service and the disclosure of personal information remains a problem to be solved in the future. Keyword : SNS stream, Big data, Machine learning techniques, CBS, Forest disaster warning information service system Acknowledgement : This research was supported by the Forestry Technology 2015 Forestry Technology Research and Development Project (Planning project).
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2017
- Bibcode:
- 2017AGUFMNH33A0241O
- Keywords:
-
- 1916 Data and information discovery;
- INFORMATICS;
- 1976 Software tools and services;
- INFORMATICS;
- 4332 Disaster resilience;
- NATURAL HAZARDS;
- 4333 Disaster risk analysis and assessment;
- NATURAL HAZARDS