EventSummarizer
Jerry Kiernan, Evimaria Terzi · 2009
We present EventSummarizer - a tool for extracting comprehensive summaries from large event sequences. EventSummarizer takes as input a sequence with events of different types that occur during an observation period, and creates a partitioning of this time period into contiguous non-overlapping intervals such that each interval can be described by a simple model. Within each interval local associations between events of different types are reported. EventSummarizer runs on top of any Relational DataBase Management System (RDBMS), on tables with a timestamp attribute. Our system is parameter free and has a visual interface that provides the user with a global view of the input sequence via the segmentation of the timeline. The easy-to-use interface provides the user with the option to further examine the activity and associations of event types within each segment.