User Reputation Analysis for effective Trading on Bitcoin Network

Abstract

Signed network is a refined form of social network where connections or relationships between people are defined with strength. An edge in this network can contain a positive, negative or neutral sign to define friendly, foe or neutral relationship respectively. This paper aims to utilize the properties of signed network and hence builds a Bitcoin alpha (whom-trust-whom) network. This trading network is further scrutinized to estimate user's reputation i.e. predicting ranks of participating actors. Thus, we are able to provide score to individual user to recommend whether he is safe to make transactions with. In addition, novel method for link prediction by calculating strength of path between two unconnected nodes is proposed. This strength value considers all the features of the nodes, edges and neighbors giving results that have higher accuracy over previously defined methods.