Monthly Archives: December 2015

Congratulations to Cyril Beyney!

Très bien Cyril!

Congratulations to Cyril Beyney upon a successful completion of his Masters of Science in Industrial and Systems Engineering from the University of Oklahoma.

Post defense party for Cyril

Cyril’s friends throw a surprise party for him in the Analytics Lab conference room

On December 10, 2015, Cyril presented and successfully defended his thesis entitled, “Quantitative analysis of social media sensitivity to natural disasters”.   The committee members, Dr. Charles Nicholson, Dr. Hank Grant, and Dr. Kash Barker, unanimously agree that the work and novel contribution meet or exceed the standards for Master-level research. His research is a first step in a nearly untouched area of social media analysis, namely how and under what circumstances can social media be used to inform decision makers during and after disastrous events to improve society’s resilience.

His work offers a quantitative analysis of the sensitivity of Twitter as a sensor to detect and measure social effects under different natural hazards including tornadoes (an all too common event in Oklahoma!), major winter storms, and forest fires.

Quantitative analysis of social media sensitivity to natural disasters

Social media platforms such as Facebook or Twitter have become a major tool for people to communicate, and for news media to relay their content. In parallel with this growth, researchers have started to use social media in various domains of study, including the analysis of people’s behavior during natural disasters. Diverse aspects of this behavior have been studied in the literature, including the detection of a disaster, the variations of the sentiment expressed, the differences regarding the distance to a disaster, or even the improvements of relevant content labeling. To the best of our knowledge, no study has been conducted that considers Twitter as a sensor with different sensitivity levels to different types of natural disasters under various circumstances.

We select 5 natural disasters of different types occurring in recent years, and for each of them we detect the shifts in behavior of the Twitter data in order to compare and contrast before, during and after the disaster the variations of tweet frequency, the proximity to the center of the disaster, the variations of sentiment expressed and the variations of tweet frequency by level of social vulnerability.

The results obtained in the empirical analysis demonstrate that Twitter is indeed a social sensor with different sensitivity levels to natural disasters. As a matter of fact we observe, depending on the type of disaster, different patterns of tweet frequency and proximity-to-disaster; negative sentiment tweets also tend to cluster closer to the disaster during the disrupted period, and finally areas with high level of social vulnerability are generally more sensitive compared to the others. The lack of available data can sometimes be an issue, but this work is an important finding to define Twitter as a social sensor to natural disasters.

Besides the tremendous work Cyril completed to be able to capture, process, clean, and filter the vast amount of available data, his contribution also includes a novel lexicon building approach to help isolate disaster related informative content.

Cyril has now completed dual Master’s degree program — one from OU and the other from ISIMA.

Congratulations again to Cyril on a job well done.

Probabilistic Mechanics & Reliability Conference 2016

Vanderbilt University

Vanderbilt University is the site for the 2016 Probabilistic Mechanics and Reliability Conference

Two abstracts have been accepted for presentation at the 2016 Probabilistic Mechanics & Reliability Conference to be held at Vanderbilt University May 22-25, 2016 in Nashville. This work is a collaborative effort of Weili Zhang, Naiyu Wang, and Charles Nicholson.

Resilience-Based Risk Mitigation and Recovery for Highway Transportation Networks

The resilience of robust, large-scale, interdependent civil infrastructure networks plays a major role in determining the resilience of a community as a whole. The performance of transportation networks, in particular, is critical since post-disaster emergency response, recovery and restoration of virtually all other facilities and lifeline systems are dependent on people and equipment being able to move to the sites where damage has occurred. Highway bridges are points of vulnerability in transportation networks exposed to extreme natural hazards. To enhance community resilience, risk mitigation strategies and decision frameworks for transportation networks should take a system perspective at the community or regional scale and should be aimed at maximizing the resilience of the network as a whole.

In this study, we utilize modern network theory to introduce a novel, comprehensive indicator to measure resilience of a transportation network, which permits risk mitigation alternatives for improving transportation network resilience to be compared on a common basis. This metric integrates the topology, redundancy, traffic patterns, and functionality as well as structural reliability (failure probability) of individual components for network resilience quantification. A project ranking mechanism is proposed, based on the newly developed metric, for identifying and prioritizing retrofit projects that are critical for effective pre-disaster risk mitigation for bridge networks. We further propose a restoration scheduling method for optimal post-disaster recovery planning using a two-dimensional network recovery metric defined for the first time in this study to capture the characteristics of the recovery trajectory that relate to the efficacy of the restoration strategies. An illustration of this resilience-based risk mitigation and recovery framework is given using a hypothetical bridge network susceptible to seismic hazards. A sensitivity study using this network illustrates the impact of the resourcefulness of a community and its time-dependent commitment of resources on the network recovery characteristics.

Cascading Failures in Interdependent Networks: A Network Flow Approach

Modern societies depend on large-scale, interdependent networks, including transportation networks, utility networks, and telecommunication networks. Each individual network plays its own critical role in community resilience. However, these networks are not independent, but rather coupled and interdependent in a variety of ways. The fundamental characteristic of which is that the disruption of a single component could result in wide-reaching cascading failures of components in other networks. Recent extreme hazard events have shown that the initial failures of small fraction of nodes in one network may propagate to the entire interdependent system

in a catastrophic manner. Consequently, in order to realistically quantify the impact of a disaster on infrastructure systems, we need an efficient and practical model to simulate and predict the failure cascading mechanism in interdependent networks.

This study introduces a newly developed multi-level interdependent network (MLIN) modeling approach. The MLIN is a binary mixed integer programming problem and can be solved exactly by branch and bound algorithm. The output of the MLIN includes post-disaster serviceability of each network, and failure status of nodes and arcs of all networks. Furthermore, community building inventory can be integrated in the MLIN as a special layer. The MLIN supports physics-based community resilience modeling, enables strategy optimization for risk mitigation and recovery at network component level, and facilitates community resilience modeling of different resolutions.


Vanderbilt University

Vanderbilt University