The rise of the Sharing Economy has enabled new types of transactions between strangers: from car sharing to micro-lending or short-term house renting. These peer-to-peer transactions come with inherent risks, so for they to happen both participants have to trust each other. To facilitate building trust, Sharing Economy platforms display feedback from past transactions forming the users reputation.
While this solution is very effective, it makes harder for new users to make their first
transactions as they are not trusted by others users. We propose a solution to know more about how these new users will behave by analysing their Online Social Networks. By identifying key traits from social network accounts we can infer beliefs about the user future reputation on the platform and act accordingly to this information to avoid future problems (as fraud) or allow an easier on-boarding process for good willing users.
The goal of this project is to mine Online Social Networks to predict users behaviour
at Sharing Economy Platforms. To make it possible it was necessary to extract users information from both platforms and match them using advanced user matching techniques.
This data was used to train machine learning algorithms for the task of classifying Wallapop users according to their reputation just by looking at their Twitter accounts. To finish, the features that were important for the machine learning classifier to make the predictions are presented and analyzed.