• bot detection
• Steem blockchain
• Steemit social network
• clustering-based bot detection
• behavior analysis
• behavior analysis for bot detection
• textual analysis for bot detection

In this thesis' work, various aspects of the Steem blockchain and of the Steemit social network which relies on it have been considered. This blockchain has a rewarding system that can be exploited by the users of the social network; for this reason, a peculiar figure was born in this context: the bot. It performs various kind of operations for a fee so that the content voted, commented or shared by a bot receives a visibility boost; this results in an increase of the chances, for the author of the curated content, to get more movements linked to that content, so that he earns as much as possible. One of the main tasks of the thesis has been to correlate the rewarding system with the quality of the users' contents. In order to perform a task like that we need to analyse how much the bots are relevant inside the network, and hence a good bot detection method is required in order to find all those unknown bots that could obtain even more rewards, by remaining hidden within the users. The bot figure needs to be identified and to perform this task, we propose a clustering approach in which the known bots gets grouped in a certain number of clusters; each cluster has a different cluster's centre. When the produced clusters are significant, the likelihood to find an unrecognized bot close to the cluster's centre could be high, whereas the bot wrongly belongs in the set of standard users. Besides, other three different kinds of analysis have been performed: 1) one where we compare the operations performed by the known bots with those by the standard users; 2) one where the comparison is over the operations carried out by the various reputation classes among each other; 3) one where we analyse the bots' comments' text by finding the common words used by the bots in their comments, so that, when a user uses some of them, the odds that he is a bot increase.