Over the past few years, millions of refugees have fled into Europe with little more than the clothes on their backs. They often struggle with finding basic items such as food, clothes, shelter, and baby supplies. While a major effort has been made to help those based in camps, when refugees are on the move, they often have no where to turn when times are hardest. We set out to solve one basic question: How can we connect these refugees with local individuals willing to help?

What it does

RefuTweet aims to provide refugees on the move with the help they need by connecting with local individuals sympathetic to their cause. It aims to solve two important problems: 1) provide real-time help to refugees in need and 2) identifying the individuals in the area who are most likely to provide help.

First, the refugee anonymously contacts RefuTweet by messaging their need and general location such as "I need two blankets near Trafalgar Square". The messages are parsed through IBM Watson to extract both the location and request. RefuTweet then searches Twitter in a radius of up to 16 miles around the area and identifies all handles which have tweeted about the refugee crisis in the past week. Using IBM Watson, a personality insight analysis is run on all relevant Twitter profiles to identify specific users in the area who are sympathetic the refugee cause and exhibit personality traits such as love, harmony, idealism, sympathy, and altruism. Based on the personality analysis, the handles are ranked on their likelihood to provide help and RefuTweet sends out friendly tweets and a link, asking the top three users if they would like to help a refugee in their area who is in need. If the link is clicked and confirmed a message is sent back to the refugee alerting them that someone is willing to help and how to arrange a meeting.

How we built it

The core of RefuTweet is based on Twilio, IBM Watson, and the Twitter API. Twilio is used to connect refugees with the service via SMS and to contact the refugee by the same means once help has been offered. The Twitter API is used to identify individuals in the area whom have recently discussed issues relating to the refugee crisis and to contact those individuals likely to provide help. IBM Watson is used in two unique ways: 1) to parse the natural language and extract both the location and need of the refugee and 2) to run a personality analysis of the tweets of individual Twitter users to identify those traits which correlate with sympathy to the refugee cause.

Our backend is written in Python and a frontend is created to visualize the entire process using the Esri API. The frontend is hosted on a Radix domain (

Share this project: