Discovering Target-Branched Declare Constraints

Discovering Target-Branched Declare Constraints
(Claudio Di Ciccio , Fabrizio Maria Maggi , and Jan Mendling)
Student Name
Institution Affiliation
1

DISCOVERING TARGET-BRANCHED DECLARE CONSTRAINTS
Introduction
Process discovery can be described as a significant first step in business process
management. Through it, one is able to arrive at the as-is model of a research process. The step is
however highly complex and time-consuming giving the necessity to employ techniques capable
of automatically discovering a process scheme from the logs event. The data log is usually
derived from information systems supporting sections or the whole process. The results are
displayed in the representation of a flow chart (Petri net). The immediate discovery process can
also be denoted as process mining. The assignment will typically dwell on some of the essential
Declare concepts providing formal basis for Target-Branched constraint mining. Other sections
to be featured in the assignment include the performance evaluation and investigation
contribution sections.
Body
Since process mining is a complex and useful approach when it comes to standardized
and structured processes, a lot of arguments have been raised on how effective mining can be
realized from processes with high variable rates (Feshland et al, 2009). In such a case, the
declarative process model can serve as a viable approach to processes with high variability
degrees. This model prioritizes on showing behavioral constraints rather than the execution
sequences. The results are outlaid in the Declare language. In numerous cases, the models
provide a simple way of representing unstructured and complex behavior which through the Petri
net representation would seem highly complex,(Di Ciccio et al., 2012). It should however be
noted that in Declare models, it’s quite complex to mine simple relative statements, for instance,
“if you carry out a, you eventually will have to do b or c. To address the Declare mining issue,
it’s paramount to begin by defining the Target-Branched Declare class and create effective
2

DISCOVERING TARGET-BRANCHED DECLARE CONSTRAINTS
mining algorithms. The idea behind this approach is to take advantage of the dominance
relationships which will aid shape the search space. Formal evidences are presented in order to
outlay the merits. Prototypical implementations are utilized to analyze performances, assess
feasibility as well as the efficiency to an approach,( Di Cicci0 et al., 2012).
Presentation of mined models serves as one of the process mining challenges. Models
with procedural complexities, for instance, Petri nets tend to initiate difficulties and rigidity in
simple processes. In such scenarios, it’s advisable to go for declarative models as they serve to
avail better enlightenment on the mined processes by individuals. Declare is the mostly used and
utilized languages for declarative. Declare comprises of a group of constraints employed on
activities. Constraints, on the other hand, have a template basis which serves to define a given
class of properties. It should be noted that templates can be graphically represented with their
semantics being formalized through the use of presentable logics such as the Liner Semi
permanent Logic on endless traces (LTLf),(Van der Aelst et al., 2009). Through this, analysts
are able to operate with the template representation while hiding the underlying formulas.
Template parameters are indicated through x and y variables. For the real actions during
instantiation, variables such as a, b or c are used. For the Responded Existence template,
template variables are highly relative, that is, when x occurs, y is bound to occur. It should be
noted that in the Responded Existence case, y is bound to occur pre or post x. in the Response
template, x and y are directly relative, that is, x occurrence should lead to y’s occurrence. It
should however be noted that y only comes after x occurs,(Dumas et al., 2013). The Precedence
template, on the other hand, indicates that y’s occurrence depends on x’s occurrence, that is, y
can only occur if x occurs. The Precedence and Response templates are reaffirmed by the
Alternate Precedence and Alternate Response templates respectively. Other templates with
3

DISCOVERING TARGET-BRANCHED DECLARE CONSTRAINTS
strong order relations include the Chain Precedence and the Chain Response. In these templates,
activity occurrences are bound together, that is x and y. in semantic illustration, consider (a, b) as
the response constraint. This means taking into consideration a occurs, followed by b is bound to
happen. The following traces can therefore satisfy the constraint;
T1= [a, a, b, c], T2= [b, b, c, d] and T3= [a, b, c, d]. For T4, the case would be different as
activity a is not followed by an ab. It’s paramount to note that an activation is an event affects
targets. For the prior illustration given, it can be said that a in the trace stands as an activation
while b represents the object in the feedback limitations (a, b). This is highly due to a’s
compilation will eventually lead to b’s execution. A fulfillment is fully realized once there is
constraint compliancy in the trace,(Burratin et al, 2012). Going back to the earlier mentioned
traces, in T1 , there is constraint activation and twice fulfillment. The same case is featured for T3
but the constraint is only fulfilled once. In cases where there is no compliancy in the trace,
activation can create a fulfillment. It should be however noted that it can also initiate at least one
activation violation. In T4, activation for the response constraint can be initiated two times. The
initial activation leads to the eventual occurrence of b while the second stands as a violation as b
fails to occur. In order to reaffirm the significance of constraints, confidence and support are
adopted from data mining. Declare constraint’s confidence lies in the degree of traces holding
activation. Despite its numerous benefits, the Declare constraint fails to accommodate branching,
( Smirnov et al., 2012).
Branch as outlaid in the alpha algorithm and behavioral profile synthesis approach
explicitly mines for statements such as ‘if we carry out a, we shall resort to doing one of b or c.
the exclusive results are normally employed on process model comprehension due to their
practical significance. A Target-Branched Declare refers to a Declare extension where the target
4