Information Technology Reference
In-Depth Information
2
nd
Generalization
Reference points
1
st
Generalization
Fig. 3.
Reference points among operations
Whereas the lecturers, module coordinators and lecturer team coordinators request for
lecture halls, the administrative cooperator and the secretary receive the requests and
topic the halls.
Afterwards, each single To-do of the 399 To-do's was coded at the lowest possible
level of abstraction already included in the To-do by paraphrasing. 195 labels reflect-
ing the key essence of the To-do's were inductively derived. 77 hints to connections
among process participants, 39 hints to operations with participation of several proc-
ess participants (e.g. meetings), 49 hints to decisions (each operation of the alternative
path was marked as a decision hint), 40 hints to events (e.g. eMail, phone calls, or
messages), 12 hints to delegation, and 43 time notes could be identified. 8 tools could
be extracted out of the To-do's as well as 30 different data material (e.g. “teaching
scheduling sheet”, “numbers of exams of the last semester”, documented course
schedule”) could be identified.
In a next step we elaborated the process view logs per process participant (per
role). The main challenges were (a) to define the artificial time stamps and (b) the
inclusion of all possible combinations of process paths as these issues were the critical
points for the quality of the mining result. Based on the number and conditions of the
decisions we elaborated possible process paths per role and captured each possible
path as a case (process instance) of each process view. The alpha mining algorithm
was used for the transformation of the individual process view logs into process view
models. We elaborated process logs in which each case illustrates an entire path of the
entire process, and considered the course (with unique course ID) as the “common
thread” throughout the process. To organize all the possible paths of the process per
case the selected paths of the process views were documented. Log segments were
compound to a possible way of the process. The heuristics mining algorithm was used
to analyze the entire process logs. Fig. 4 illustrates a comparison of the result of ap-
plying the heuristic mining algorithm in ProM [18] of the collected, analyzed, and
merged To-do's to a process model (left and middle part) and a manually modeled
process segment in which BPMN notation was used (right part). The BPMN model
segment was designed by the process-aware designer before the To-do's of the proc-
ess participants were analyzed and mined. Both segment models gather the same
activities (prepare, conduct, and post process unit).
