Follow Datanami:
April 28, 2016

A Big Data App to Help You Travel Safer


Let’s face it: the world is a dangerous place. From terrorist attacks and earthquakes to outbreaks of hemorrhagic fever and pickpockets, there are a thousand ways for an American to stumble into trouble when travelling abroad. Now a new app from Prescient aims to help travelers stay safe by tapping into real-time data feeds, analyzing threats, and sending alerts when users wander too close to danger.

Prescient Traveler is a mobile app that functions as a digital concierge for businesspeople, tourists, and students as they travel around the U.S. and the world. Conceptually similar to an application Prescient developed for the Defense Intelligence Agency, Prescient Traveler helps people stay aware of emerging and persistent threats as they venture to strange locales, both foreign and domestic.

Michael Bishop, the chief systems architect for Prescient, says the company developed the application at the request of a large global law firm that was concerned about the safety of its lawyers as they travelled to foreign cities—specifically, Bogota, Columbia, which has a relatively high crime rate, according to the U.S. Overseas Security Advisory Council (OSAC).


Prescient Traveler alerts you when you travel to a dangerous area

“They had problems with kidnappings and outright abductions that happen down there, and they wanted to understand what they could do to mitigate that risk and keep their travelers safer,” Bishop tells Datanami. “I began to look at available solutions, and they basically send somebody a 20-page report about where they’re going, and nobody reads it.”

As Bishop scoped out the new app’s requirements, he realized there was a potential to take some of the lessons learned from the DIA program and apply them to a broader class of citizen. What resulted is an app that can quickly analyze huge amounts of structured and unstructured data from the Internet, compare it to known baselines, and deliver alerts targeted to a person’s individual location, characteristics, and needs.

Prescient Tech

Prescient Traveler is a veritable case study in how to use some of the latest and greatest big data technologies. It uses:

  • Apache NiFi stream data processing engine to ingest and perform entity extraction on massive volumes of data from more than 41,000 sources, including social media sites, news sites, weather feeds, and geological alerts;
  • Hortonworks (NASDAQ: HDP) Hadoop distribution to create a 1PB data lake where the unstructured data can be stored and prepped for analysis;
  • Esri‘s ArcGIS and the open source PostGIS geospatial analytics package to process location data;
  • SAP (NYSE: SAP) HANA column-store to run the geospatial and linguistic analytics;
  • MongoDB‘s NoSQL database to power the mobile apps, the dashboards, and the internal applications used by Prescient analysts.

“We use SAP HANA for a lot of the real-time geospatial and linguistic analytics,” Bishop says. “We process in excess of 350,000 tweets per minute for sentiment analysts. A big part of our rationalizing to invest in HANA was it can perform that in 33 languages.”


Prescient Traveler uses multiple big data technologies

The company has built its own dictionaries in HANA that allow it to automatically extract relevant information in regards the physical, environment, and health threats. “Each one of those is decomposed into about 20 threat factors, like terrorism, conflict, and violent crime, so we have very extensive lexicons that have been built out, and that’s what informs the national language processing in HANA.”
Prescient has also written dozens of NiFi processors to parse incoming data streams. “We’re looking for geocode-able terms that allow us to say, oh [something happened in] Madrid, Spain. Do we have anybody in Spain? All the geo-spatial products are queryable by the in-memory columnar HANA data store, and we’re constantly balancing the location of our travelers against those threats.”

The company persists more than half terabyte of fresh data every day to its Hadoop data lake, where it’s combined with data from sources like OSAC and, as well as data from non-government organizations (NGOs), which provide statistics about the state of regional healthcare infrastructure, communicable diseases, and law enforcement capabilities–and even the relative level of official corruption. (However, some of the demographic data from government-run sites must be taken with a grain of salt.)

Human in the Loop

Information about more than 700 locations around the world has been collected and coded into the system, and more are being added all the time. “In some cases where there is no data we use inference algorithms to fill the gaps in sparse data sets, where there’s no crime but maybe demographics and property value” data is available, Bishop says. “We also do overhead imagery analyses, and conclude, for instance, than an area looks like the same kind of slum that we see elsewhere.”


MongoDB powers Prescient Traveler user interfaces, while NiFi, HDP, Esri, and HANA do the heavy analytics lifting behind the scenes

Most of the alerts stemming from weather or geological events—such as severe storms, earthquakes, or volcanic eruptions—are very automated, but some human-caused events require a human in the loop.

“For instance,” Bishop says, “if there’s an active-shooter situation being reported in a shopping mall, within a matter of seconds or minutes, the analyst can geo-fence an area, send out alert, and everybody in that region can check in to say they’re okay. They can also be advised, and told to head east or shelter in place. That’s part of the workflow.”

Users can set their own safety parameters for the software. Bishop has set his to alert him whenever he comes within 500 meters of a high-crime area, and the app regularly pings him when travelling from his home in Indiana to Prescient’s headquarters in Chicago, which takes him through Gary, Indiana—a well-known crime hotspot.

“We believe anybody would be interested in that. That’s why we offer it on a subscription basis,” he says. “Not only do you get real-time alerts when there’s an active shooter nearby, but also literally as you’re driving down the street or walking down the sidewalk, it will tell you to take a left not a right. Those are the types of actionable recommendation we can make with this.”

Norms and Taboos

In addition to the real-time notifications, the Prescient Traveler includes a “cultural norms” section that’s designed to keep travelers from offending the locals. This section includes pictographic representations on what you should not do in certain countries. For instance, it’s rude to point with a finger in some Asian countries (use your whole hand instead), while shaking hands while standing in a doorway in Russia is considered bad luck.

“We’ve integrated that into what we believe is a very coherent and easy to understand user interface. It’s the antithesis of the 20-page report about Beirut before you get to Beirut,” he says. “We know from operational experience [from defense and intelligence work] that you do not want to alienate other people….I might at a minimum disrespect someone. But I can also make myself a target.”prescient_logo

For instance, a blond-hair, blue-eyed female going to Saudi Araba would do well to prep for the trip. “In the cultural norms part of the app, it’s advised she packs a scarf, that she shouldn’t expect to drive or dine alone, and to stay with a male companion. Likewise if she’s walking around town if there’s a likelihood of problems, she gets a proximity alert as she gets closer to that part of town.”

Government and military personnel regularly rely on this sort of next-gen tool to boost their situational awareness, and to avoid standing out to the locals, which would boost the chance of mission failure. But Prescient Traveler is the first app of this class to be offered to the public, Bishop says.

“There’s no one in the private sector, for the business traveler, who’s offering down to the street level that kind of geospatial threat assessment,” he says. “I often say, there’s no way a comparable solution could be out there, because this technology just didn’t exist when they competitive offerings were built.”

The system is just coming out of beta and is available for $20 per user per month, with volume discounts available for large customers. For more info, see the video below:


Related Items:

Unleashing Artificial Intelligence with Human-Assisted Machine Learning

This Information Concierge Is Powered by Big Data