HomeNews and blogs hub

Veri.ly – getting the facts straight during humanitarian disasters

Bookmark this page Bookmarked

Veri.ly – getting the facts straight during humanitarian disasters

Author(s)

Victor Naroditskiy

Posted on 13 August 2014

Estimated read time: 5 min
Sections in this article
Share on blog/article:
Twitter LinkedIn

Veri.ly – getting the facts straight during humanitarian disasters

Posted by a.hay on 13 August 2014 - 10:00am

By Victor Naroditskiy, Post-Doctoral researcher on the ORCHID project, University of Southampton.

This article is part of our series: a day in the software life, in which we ask researchers from all disciplines to discuss the tools that make their research possible.

We live in what has been described as the Information Age, but at times a better term would be the Disinformation Age. This is due to the sheer volume of information being propagated by the Internet, and in particular, social media, every day at the click of a mouse. Finding out the truth in this sea of contradictory material has, as a result, become increasingly difficult.

Can a coordinated collective effort be effective in quickly discerning true answers to key questions? Crowdsourcing has been used to solve several seemingly impossible search tasks, but there have been many more failures.

One recent example was Reddit's attempt to identify the Boston Marathon bombers. This not only led to false accusations but was likened to a witch hunt. Reddit turned out to be not well-suited for the task, but it was never designed for it in the first place. Nonetheless, evidence collection requires the involvement of people who have either access to evidence or knowledge of where to look and who to ask. This task is all but impossible for a single person or organisation to undertake, but is well suited for crowdsourcing. It stands to reason that someone somewhere must have evidence: all we need to do is reach them.

Whilst most crowdsourcing projects suffer from a lack of participation or commitment, the search for the truth is a compelling task, which makes it easier to motivate participants. In scenarios where accurate information can save lives, the motivation to keep searching for evidence is even stronger. With that in mind, Veri.ly - a site developed by the University of Southampton, Masdar Institute, and QCRI - is to make people think about the question rather than use gut instinct when answering questions. In particular, a user cannot simply submit a "yes" or a "no" for an answer.

Instead they are encouraged to justify their position by providing evidence either in the form of an image, a video or as text. The platform’s interface makes it clear that opinions and speculations alone are not welcome. In particular, Veri.ly does not make use of “like” buttons or up and down votes (such as the system used by Reddit) to answer questions. While such tools are great for identifying and sharing content, they are not the right tools for verification, which requires searching for evidence rather than just liking or retweeting something.

 

 

 

 

Can a coordinated collective effort be effective in quickly discerning true answers to key questions?​ We designed the Veri.ly Challenge as a test of this question, and asked people whether they could tell if the photo on the left was taken in Jerusalem?  

 

 

 

 

 

 

The question is very difficult to answer for anyone who has not been on the small street in question. It could be anywhere. The goal is to find a person who can both recognise and locate the street. This is exactly what happened: one user recognised the place, pinning down the shop "between Holy Sepulcher and Lutheran church", and another user provided a link to Google StreetView confirming the location


In the above image, we see how Veri.ly works in practice. The user community is presented with questions, be they text only or linked to the displayed images, and they then provide both answers and supporting data accordingly.

While the Veri.ly Challenge asked general fact-checking questions, the central focus of the project is on gathering evidence during natural disasters.  Accurate information can save lives in extreme situations, such as earthquakes and tsunamis. The lack of verifiability of content posted on social media is the main reason preventing humanitarian and news organisations from making wider use of it. The Veri.ly Challenge confirmed the feasibility of rapid evidence collection as its questions and collected evidence demonstrate. The next step is to deploy Veri.ly for evidence collection during a humanitarian disaster, and its progress can be followed on our official Twitter account.

Veri.ly was invented by Carlos Castillo (QCRI), Patrick Meier (QCRI), myself and Iyad Rahwan (Masdar). The technical development of the platform was by Masters students Luis Arenal Mijares, Alex Greenland and Dimitrios Papamiliosin from the University of Southampton’s Web Technology MSc programme, led by Enrico Costanza and I. Justine Mackinnon (QCRI), together with the Southampton team, organised the public trial of the platform. Victor and Enrico research crowdsourcing in the context of ORCHID, a multidisciplinary EPSRC-funded project developing the science and technologies for human-agent collectives, with disaster response as a key application area.

Share on blog/article:
Twitter LinkedIn