Discovering and clustering hidden time patterns in blockchain ledger

Abstract

Currently, immutable blockchain-based ledgers become important tools for cryptocurrency transactions, auditing, smart contracts, copyright registration and many other applications. In this regard, there is a need to analyze the typical, repetitive actions written to the ledger, for example, to identify suspicious cryptocurrency transactions, a chain of events that led to information security incident, or to predict recurrence of some situation in the future. We propose to use for these purposes the algorithms for T-patterns discovering and to cluster the identified behavioral patterns subsequently. In case of having labeled patterns, clustering might be replaced by classification.