Friends, followers and fake news: building trust into online networks
Understanding how members of the public make decisions is of crucial importance to a range of applications, from politics to commerce to healthcare. But accurately modelling sentiment across large groups of people – often relying on fragmented, heterogenous and contradictory data – is challenging. The DeciTrustNET project aims to tackle this issue by developing mathematical models that can identify levels of trust within a network and consensus across crowds. “People tend to define opinions with words rather than numbers,” says project coordinator Francisco Chiclana from De Montfort University, Leicester, United Kingdom. “My research is on how to model agreement between people with fuzzy preferences.”
Online bonds
As social networks have become increasingly central to modern life, understanding how they shape attitudes and decisions is important in order to explain phenomena such as the spread of misinformation online, or the rejection of real facts as ‘fake news’. DeciTrustNET is a framework that takes into consideration current and historical relationships across the network to accurately assess the trustworthiness of information within it. Chiclana says that trust is a core part of how behaviour is influenced through online relationships. “Trust exists between friends – because I know you, and know you are honest, when you say something it’s normally true,” he adds. Modelling this trust computationally can help build networks that resist the spread of misinformation by flagging dubious content.
Testing the waters
The models he has developed can also measure the level of consensus within a group of people, and identify those whose opinions are most distant from the average. “For example, if the government was trying to legislate the minimum wage, it might leak some kind of preliminary info to the media,” explains Chiclana. “People will react on Twitter or Facebook, and the government could analyse the level of acceptance of this idea. If they feel a majority of people are in favour of that change they might go ahead.” Artificial intelligence tools that can analyse the level of trust between people online could also be used by marketing companies to identify influencers most able to drive up sales of the product.
Publication record
This research was undertaken with the support of the Marie Skłodowska-Curie programme, which provided funds for researcher Raquel Ureña to join Chiclana. “The main aim of this project was to develop the mathematical and computational framework for trust-based social choice with imprecise information, and we have managed to get it effective,” he says. “In 2 years we secured seven publications in highly regarded journals.” Chiclana says future work will focus on models that work dynamically and provide feedback such as informing policymakers on how the public will react to new initiatives, or how healthcare organisations can most effectively nudge people away from detrimental behaviours such as smoking.
Keywords
DeciTrustNET, trust, online, social, network, fake, news, decision, social, influencers, misinformation
