Discovery of Frequent Episodes in Event Logs

Abstract

Abstract. Lion's share of process mining research focuses on the discovery of end-to-end process models describing the characteristic behavior of observed cases. The notion of a process instance (i.e., the case) plays an important role in process mining. Pattern mining techniques (such as frequent itemset mining, association rule learning, sequence mining, and traditional episode mining) do not consider process instances. An episode is a collection of partially ordered events. In this paper, we present a new technique (and corresponding implementation) that discovers frequently occurring episodes in event logs thereby exploiting the fact that events are associated with cases. Hence, the work can be positioned in-between process mining and pattern mining. Episode discovery has its applications in, amongst others, discovering local patterns in complex processes and conformance checking based on partial orders. We also discover episode rules to predict behavior and discover correlated behaviors in processes. We have developed a ProM plug-in that exploits efficient algorithms for the discovery of frequent episodes and episode rules. Experimental results based on real-life event logs demonstrate the feasibility and usefulness of the approach.

Mining Conditional Partial Order Graphs from Event Logs

“…as reported recently in [42]. Episodes are partially ordered collections of events (not activities), and as such they can also be represented by CPOGs.…”

QRFA: A Data-Driven Model of Information-Seeking Dialogues

“…Imperative workflow mining algorithms, such as the seminal α-algorithm [28] or the more recent Inductive Miner (IM) [13], extract procedural process models that depict the possible process executions, in the form of, e.g., a Petri net [27]. Other approaches, such as frequent episode mining [12] or declarative constraints mining [7,8], extract local patterns and aggregate relations between activities. One such relation is the succession between two activities, denoting that the second one occurs eventually after the first one.…”

“…We used the ProM Episode Miner plug-in [12] and a declarative process mining tool, MINERful 8 [7,8], to discover frequent sequence patterns in the conversation transcripts. Figure 3 illustrates the conversation flows in each of the three datasets used for model discovery (one of the datasets, DSTC2, is held out for model evaluation).…”

Subgraph Mining for Anomalous Pattern Discovery in Event Logs

“…Well-known approaches for subprocess extraction from sequential traces are [8], which detects subprocesses by identifying sequences of events that fit a-priori defined templates; [15], which exploits a sequence pattern mining algorithm to derive frequent sequences of clinical activities from clinical logs; and [17], which introduces an approach to derive "episodes", i.e, directed graphs where nodes correspond to activities and edges to eventually-follow precedence relations, which, given a pair of activities, state which one occurs later. With respect to previous approaches, the one proposed in this work does not require to define any a-priori defined template and extracts the subprocesses that are the most relevant according to the MDL principle, thus taking into account both frequency and size in determining the relevance of each subprocess.…”