Detecting artificial behaviours in the Bitcoin users graph

A unique feature of cryptocurrencies such as Bitcoin is that the blockchain containing all the economic transactions is publicly available. This makes it possible to obtain insights in the behaviour of the users through an analysis of the topological properties of the users graph which is derived from the Bitcoin transaction graph through clustering heuristics. In a previous work, we have analysed the users graph and discovered that the graph is not a small world, due to the presence of outliers in the in-degree frequency distribution of the nodes and of a high diameter, in spite of a small average distance between the nodes of the graph. In this paper, we explain our findings, showing that these structural properties of the network are due to peculiar unusual patterns in the users graph. As a further remark, we argue that these patterns are probably due to artificial users behaviours and not strictly related to normal economic interactions.