Summarizing Processes

A large number of processes, of dierent characteristics, are
being created, stored and reused during Services-Oriented Architecture
based implementations. The processes may describe specification of func-
tionality for a to-be built system (e.g., business processes in different
notations), automatically or semi-automatically created compositions/
plans of services to meet some goals (e.g., PDDL plans), or execution
behavior (e.g., web processes). However, a user today can get little in-
sight from such a collection of processes other than to search it with
a keyword based query interface. In this paper, we introduce the prob-
lem of automatically summarizing a collection of processes and propose
a comprehensive solution for it. The solution works on any available
type of process content ranging from metadata, syntactic information
like activities, process features interpreted as semantic annotations, to
multi-dimensional textual content; and lters the analyses based on sig-
nicance. The summary so created provides insights about what the
repository contains at the aggregate level as well as in subsets (clusters)
of processes, built using an extensible set of process distance measures.
We employ the solution on diverse collections containing hundreds of
processes and demonstrate that the technique can shed correct and valu-
able information where none existed before, while being scalable and
general-purpose.