Editor’s note: This month we are exploring the theme “using technology to extract knowledge from data.” Elsevier colleagues are using innovative tools and techniques in data analytics to prepare a report about the burgeoning field of disaster science.
Hurricane Harvey was one of the most damaging storms in US history, providing a stark warning of the far-reaching effects of natural disasters. At the same time, South Asia experienced some of the worst flooding in years, claiming more than 1,000 lives. Days later, Hurricane Irma ripped through the Caribbean and Florida, followed by José and Maria, devastating cities and island nations and leaving tens of thousands homeless without power, food and water.
As climate change contributes to an increase in floods, droughts and other extreme weather events, it puts further pressure on a world already failing to meet humanitarian needs.
Science and technology can help us address these issues.
Elsevier has been preparing a report on the current state of disaster science. The analysis uses innovative natural language processing techniques to provide an evidence base from which governments, industry and humanitarian organizations can plot a course of action to mitigate the effects of disasters. The full report – A Global Outlook on Disaster Science – will be released in November. Meanwhile, here are three key findings:
1. Research needs to cross disaster management stages
The disaster management cycle encompasses several overlapping stages, such as preparedness, response, recovery and mitigation. As Sarah Huggett, Analytical Services Product Manager at Elsevier, explained:
Our analysis shows that disaster management is a continuum and research on the stages converge. It means that researchers work on these stages in an integrated fashion rather than in isolation. This interlinkage also reflects how the research community’s work addresses several aspects of the cycle at once or reviews specific stages in the context of disaster management overall.
Dr. Rajib Shaw, a professor at Keio University in Japan, member of the UNISDR Scientific and Technical Advisory Group, and contributor to the report, agreed, calling for a new approach to disaster science research:
It’s only in the last few years that we’ve seen more emphasis placed on transdisciplinary research. Before that, research in disaster science was much more compartmentalized. Addressing disaster science as a whole in our education system has required a mindset change nationally and globally. The report gives an evidence base to guide that change.
2. The disaster science community is responsive – recent disasters appear quickly in research
A positive trend is the speed at which research is being done. “Actual life occurences appear into research fast,” Huggett said. “That provides opportunity to learn quickly from each event.”
As Prof. Shaw pointed out, science has a significant role to play in disaster reduction. He gave one example from flood-prone Bangladesh, where 45 percent of the country is underwater even in an average year. In a bad year, that can rise to 70 percent. This flooding causes major problems with drinking water, in addition to groundwater contamination with arsenic in the southwestern part of the country.
“We’ve seen good examples from the private sector in providing a tank system that can store rain water for a longer duration,” Prof. Shaw said. “That is being widely popularized, and it’s an innovation that has benefitted the economy as well as the people. The responsiveness of disaster science research gives us opportunities to save lives, as long as we understand what we can learn from it.”
On the other hand, if the world fails to learn from previous disaster experiences, the humantarian crises they cause will worsen – a factor central to Prof. Shaw’s forthcoming book: Science and Technology in Disaster Risk Reduction in Asia (Elsevier, 2017). “We have plenty of examples, but there’s a danger that we’re not learning from those examples,” he said.
3. In the last 5 years, over 27,000 disaster science papers were published globally
Prof. Shaw said the report is coming at a crucial time in this burgeoning field.
“When you talk about disaster reduction as an integrated field, it’s still very new – you can trace it back to 1990 the start of the International Decade for Natural Disaster Reduction,” he explained. “It’s vital that we understand what the trends have been in recent years so we can understand where we should be going. Understanding where we are helps identify where the gaps are and where the needs are and makes the discipline of disaster science more effective.”
In the past, Prof. Shaw said, the trend has been for governments to use disaster research to understand how to help people respond and how to make systems more robust. However, the development of new programs and partnerships with the private sector can place a greater emphasis on disaster reduction. As more and more papers are published, understanding where to focus will ensure that priority areas are covered.
How the analysis was done
Extracting insights from such a complex, interdisciplinary field is no easy task. “To address all the implications of disasters, you need to harness expertise from across scientific disciplines,” Huggett explained. Her Analytical Services team drew on international experts – including Prof. Shaw.
Then they set out to define disaster science and search for the relevant research.
Huggett explained the approach they took:
We adopted a keyword-search approach to avoid restrictions to any particular subject areas, as that could ignore a significant portion of the relevant corpus. We focused our search on those publications that explicitly adopt a disaster science perspective, thereby also ensuring consistency across different corpora.
The team also examined disaster types defined by the Sendai Framework for Disaster Risk Reduction and disaster management cycle stages as defined by the experts.
Having defined the field, they used the Elsevier Fingerprint Engine to extract the most frequent phrases and concepts in each of these documents. The engine can search for context-specific words and phrases and give them a weighting depending on factors such as how common they are, whether they appear in an abstract, and how frequently a paper is cited. As Dr. Jeroen Geertzen, Team Lead for Natural Language Processing (NLP) at Elsevier, explained, this process is essential to making connections and creating insights from this data.
“To make the data meaningful, we didn’t limit ourselves to the keywords,” he said. “If you only take that approach, you run the risk that the data will be inaccurate. In an area such as disaster science, a lot of disambiguation is required. If you search on a word like “solution,” you don’t want results involving a chemical solution or a mathematical solution.”
To make the data more more precise, the team deployed NLP using complex integrated algorithms to match words to concepts. “That way, you don’t get frequently used words like “human” and “article” showing up in the data, just relevant concepts from the research itself,” Dr. Geertzen explained.
Using this technology, we can create a far deeper understanding of where we are with disaster science and where the gaps are. As governments and industry plot a course forward, they have a map that has been created with the most effective tools available.
Elsevier’s research reports
Elsevier partners with research institutions, funders and policymakers worldwide to address research management challenges. Through our Research Intelligence and Research Networks groups, extensive data sources such as Scopus, sophisticated tools like the Elsevier Fingerprint Engine, and our technological and bibliometric expertise, Elsevier helps organizations engage in data-driven strategic planning to improve research performance; share data across systems; and collaborate to address major societal challenges.
Our reports related to sustainability science include:
- Sustainability Science in a Global Research Landscape (2015)
- Mapping Gender in the German Research Arena (2015)
- Gender in a Global Research Landscape (2017)
Read more about Elsevier’s reports and research initiatives.
Using technology to extract knowledge from data
To prepare for disasters and respond and rebuild effectively, government and industry must draw on the collective knowledge of the scientific community. Elsevier has worked with experts in disaster science and used cutting-edge techniques in data analytics to inform scientists and policymakers. Their upcoming report draws on high-quality global data to support policy development and implementation. This research will help government and industry reduce the damage wrought by natural disasters and develop resilience for the future.
comments powered by Disqus