Feature Constrained Parallel Data Processing Approach for Spatiotemporal Event Detection

The enormous usage of Online Social Networks (OSN) leads to unleashing usage of the smart technologies in social life which becomes intertwined. Generating meaningful patterns out of various location-based streaming social big data fascinated the techies around the globe to develop innovative approa...

Full description

Saved in:
Bibliographic Details
Published in2017 Ninth International Conference on Advanced Computing (ICoAC) pp. 345 - 351
Main Authors Bhuvaneswari, A., Vallivammai, C., Devakunchari, R.
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.12.2017
Subjects
Big Data
crisis management
crowdsourcing
Event detection
Feature extraction
Logistics
MapReduce
online social networks
Real-time systems
Twitter
Online AccessGet full text

Cover

More Information
Summary:The enormous usage of Online Social Networks (OSN) leads to unleashing usage of the smart technologies in social life which becomes intertwined. Generating meaningful patterns out of various location-based streaming social big data fascinated the techies around the globe to develop innovative approaches to enhance emergency support on-demand. In the specific situation like disasters, nature setbacks down at unpredictable instances. During such circumstances, a scalable and low latency analytics approach is highly required to identify the event and its location at very high speed for recovery. In this paper, the spatiotemporal event detection approach is proposed based on dynamic features emerged from the location of interest in OSN using a map-reduce framework. The experimental results address the advantage of MapReduce paradigm is indeed suitable for scalable and high-speed data streams with minimal latency. In addition, an information theoretic emergency logistic mapper is discussed as a part of disaster recovery phase which can be accomplished via social big data analysis in near future.
DOI:10.1109/ICoAC.2017.8441190