Process mining provides an important bridge between data mining and business process analysis. This technique allows for the extraction of information from event logs. In general, there are two steps in process mining: correlation definition or discovery and then process inference or composition. Firstly, the authors mine small patterns from log traces of two applications; those patterns are the representation of the execution traces of a business process. In this step, the authors use existing techniques. The patterns are represented by finite state automaton or their regular expression. The final model is the combination of only two types of small patterns that are represented by the regular expressions (ab)* and (ab*c)*. Secondly, the authors compute these patterns in parallel and then combine those small patterns using the composition rules. They have two parties. The first is the mine, where the authors discover patterns from execution traces, and the second is the combination of these small patterns. The pattern mining and the composition is illustrated by the automaton existing techniques.
TopMany techniques are suggested in the domain of process mining, we quote:
M.Gabel and al (Gabel, M., & Su, Z, 2008) present a new general technique for mining temporal specification, they realized their work in two steps; firstly they discovered the simple patterns using existing techniques, then combine these patterns using the composition and some rules like Branching and Sequencing rules.
Temporal specification expresses formal correctness requirement of an application’s ordering of specific actions and events during execution, they discovered patterns from traces of execution or program source code; The simples patterns are represented using regular expression (ab)* or (ab*c)* and their representation using finite state automaton, after they combine simple patterns to construct a temporal specification using a finite state automaton.
G.Greco and al (Greco, G and al., 2006) discovered several clusters by using a clustering technique, and then they calculate the pattern from each cluster, they combine these patterns to construct a final model, they discovered a workflow scheme from, and then they mine a workflow using a Mine Workflow Algorithm, after they define many clusters from a log traces by using clustering technique and Process Discover Algorithm and some rules cluster.
Then they use a Find Features Algorithm to find a patterns of each cluster, finally they combine these patterns to construct a completely hierarchical workflow model.
In their clustering algorithm, clusters reflect only structural similarities among traces; they say that in future works extending their techniques to take care of the environment so that clusters may reflect not only structural similarities among traces, but also information about, e.g., users and data values.