We at Inform believe that data tells a story, across all industries, and every week we round up the most interesting ones. This week: catching lies, fighting the flu, and what the Chinese are buying.
How do you catch a liar? Humans are bad at it, say researchers at the University of Michigan, and perform only slightly better than a coin-flip. That’s why the team is using real-world data to build a better way.
The researchers’ lie-detecting software is based on data from a set of 120 video clips from “high-stakes court cases,” half of which had been deemed to be deceptive. To obtain the data, the audio was transcribed and the frequency and types of words were analyzed, as well as the number and types of gestures.
The software was found to be up to 75 percent accurate in identifying who was lying while humans were right only a little more than half the time. The software also discovered several “tells.” For example, liars were more likely to scowl or grimace; look directly at the questioner (perhaps as a way of overcompensating); gesture with both hands; and use speech fillers such as “um.”
Flu season can be deadly. In Switzerland, the flu virus results in as many as 5,000 hospitalizations and 1,500 deaths every year. So Swiss researchers, along with those from Germany and the U.S., are looking for a way to decrease those numbers.
After analyzing datasets from publications on the host molecules that flu viruses rely upon to replicate, the team “discovered 20 previously unknown host molecules that promote the growth of influenza A viruses.”
One of those host molecules is known as UBR4, which can help a flu virus replicate as many as 20,000 new viruses. The scientists discovered that blocking UBR4 prevents that virus replication and therefore “is feasible as a therapeutic strategy for the treatment of influenza.”
While animal cruelty cases were previously placed in a general category in FBI’s National Incident Based Reporting System, starting in January they will be placed in their own specific categories, including neglect and intentional abuse, and will be classified as “crimes against society.”
Such a change is important not only to prevent cruelty to animals, but to predict escalating acts of violence. Previous research has found links between animal cruelty, domestic violence, and other criminal acts. Most recently, this pattern was found in the case of Robert Lewis Dear, alleged shooter at a Colorado Springs Planned Parenthood clinic, who has been accused of both animal cruelty and domestic violence in the past.
Beijing recently issued its first red alert for pollution, and IBM is trying to use big data to remedy the problem very unhealthy air in China’s capital and other cities.
Using machine learning, data scientists will analyze the quality and accuracy of previous weather forecasts, and build improved forecasting models from there. In the past, when a city knew the source and amount of pollution in the air, the more likely it was to take action, resulting in lowered pollution levels and improved public health.
Ideally, as a result of such number crunching and analyses, cities like Beijing will have issued their first and last red alert.
E-commerce behemoth Alibaba recently released its latest big data report on consumer habits.
Analyzing data based on the behavior of 300 million shoppers from 2011 to this past September, Alibaba came away with a several findings. For instance, they found that consumers were buying healthier, investing much more in purchases such as organic foods, healthcare products, and sports equipment.
They also found that those born in 1980s and ‘90s were the biggest shoppers, and, most surprisingly, that people shop much more during the Magpie Festival, or Qi Xi, a sort of Chinese version of Valentine’s Day that falls in August, than on Western Valentine’s Day, showing perhaps that “young Chinese people have started to value their own tradition.”