Clustering and Automatic Labelling Within Time Series of Categorical Observations—With an Application to Marine Log Messages

Abstract System logs or log files containing textual messages with associated time stamps are generated by many technologies and systems. The clustering technique proposed in this paper provides a tool to discover identify patterns macrolevel events data. motivating application is frequency converters the propulsion system on ship, while general setting fault identification classification complex industrial introduces an offline approach for dividing series of into discrete segments random lengths. These clustered limited set states. A state assumed correspond specific operation condition system, can be mode normal operation. Each states specific, messages, where appear semi-structured order within segments. structures not defined priori. We propose Bayesian hierarchical model characterised both temporal type each segment. An algorithm inference based reversible jump MCMC proposed. performance method assessed simulations operational