When disaster strikes, we are expected to look for a safe haven and wait for the relief to arrive. Avoiding natural disasters as well as being completely prepared for them is not possible, as witnessed in the recent Chennai floods. But can we rely on technology to save our lives during catastrophes? Yes, we can.
A research and development project on big data by B. Muthukumaran, Practice Head – Big Data, HTC Global Service, and his team, following the catastrophic floods in Chennai during December 2015 showed that by analyzing social media feeds, one can anticipate areas stricken with disaster, which can be used for post-disaster measures.
In a situation like the Chennai floods, where more than 500 people were killed and 1.8 million people displaced, it is difficult for anyone to provide all the necessary help. Twitter was flooded with tweets from those stuck in the floods. Relief came from all sides when others understood the gravity of the situation. This was an opportunity that presented a lot of social media data.
“Whenever we get new sets of data, we can experiment with it. So here was an opportunity for us to pick up data and make sense of it for a good purpose,” Muthukumaran said.
The team looked at tweets extensively that used hashtags such as #ChennaiFloods and #ChennaiRainsHelp. Tweets that were geotagged were used to map the locations which were severely affected. Tweets that did not have geotags were used for context analysis.
Analyzing big data is the job of a data scientist, who will constantly be looking at the numbers and figures. But this data has to match with that from the meteorological department.
“For the model we relied on statistics, while for the interpretation we fell back on meteorology. My team members were data scientists and for us it was just numbers. We try to match it back to the weather reports. We check for the probability and, for example, if it is more than 95 percent, we went ahead with the analysis,” he said.
Also read: Big data’s big role in humanitarian aid
It was, however, not possible to use all of the Twitter data because of the restrictions imposed. But Muthukumaran said that even with the available data one could make meaningful analysis through the social media site.
The difficulty that any big data project faces is the variation in the way sentiments are put across. Categorizing the text rightly was the challenge, he said.
Constant supply of electricity was also an issue that many people faced. Phones could not be charged; however, there was constant flow of tweets for analysis. People from other cities who visited Chennai for relief programs also tweeted.
He conveyed that there were other technologies that could have been utilized in times of distress.
“Disaster management is the need of the hour. Provided a disaster reoccurs, how are we prepared? We cannot look around for coordinates at the last minute. If big data can help us, then why not use it?” concluded Muthukumaran.
