10. Process Monitoring

Question 1

What’s the definition of process monitoring?

Answer 1

is about using the data generated by the
execution of a business process (event logs) to extract insights about the actual performance of the process and to verify its conformance with respect to norms, policies, or regulations

Question 2

What are the 2 process monitoring techniques?

Answer 2

Online(on-the-fly) vs offline (post-mortem)
Statistic (process performance dashboards) vs model based (process mining tools) techniques

Question 3

What’s a process performance dashboard and which are the 3 types of categories?

Answer 3

Is a graphical representation of one or more performance measures (or other characteristics) of a business process.

Types:
- Operational dashboards (run-time)
- Tactical dashboards (historic)
- Strategic dashboards (historic)

Question 4

What’s process mining and which are the 4 categories of techniques?

Answer 4

is a family of model-based monitoring techniques to analyze the performance and conformance of business processes based on event logs produced during their execution

Categories:
- Performance mining
- Automated discovery
- Variant analysis
- Conformance checking

Question 5

What’s an event log?
List its minimum requirements and major challenges

Answer 5

An event log is a collection of timestamped event records

Min. req:
- Case ID
- timestamp
- Activity/task

Major challenges:
- Correlation (event -> case)
- Timestamping (e.g., delayed logging, different time zones)
- Longevity (distortion by mixing incomplete with complete cases)
- Granularity

Question 6

What’s automated process discovery?
List:
- input vs output
- Techniques
- Quality measures

Answer 6

• input vs output: event log to discovered process model

Techniques:
- Dependency graph
- Alpha-algorithm
- More robust algorithms

Quality measures:
- Fitness
- Precision
- Generalization
- Complexity

Question 7

In automated process discovery, what are the characteristics and operations of the dependency graph?

Answer 7

Characteristics:
- node represents one event class
- arc represents a “directly follows” relation
- color and thickness of arcs may be used to encode how frequently one pair follows another

Operations:
- Abstraction
- Filtering: Event, event pair, trace

Question 8

What’s Process performance mining?
List:
- input vs output
- Performance dimensions

Answer 8

input vs output
- Event log + Process model-> Enhanced process model (with color-coded elements)

Performance dimensions: devil’s quadrangle

Question 9

What’s Conformance checking?
List:
- input vs output
- Two basic perspectives

Answer 9

input vs output
- Event log + Process model (or set of business rules) à Difference diagnostics

Basic perspectives:
- Process control flow (based on constraints vs. normative process model)
- Process data and/or resources (based on constraints)

Question 10

In conformance checking, which are the Three recurrent types of control-flow constraints?

Answer 10

1. mandatoriness
2. exclusiveness
3. ordering

Question 11

In conformance checking, which are the Two types of log-model discrepancies?

Answer 11

unfitting log behavior vs additional model behavior

Question 12

What’s Variants analysis?
List:
- input vs output
- Two approaches

Answer 12

input vs output:
Two event logs (corresponding to two variants of a process) -> Difference diagnostics

approaches:
- manual approach:
visual comparison; Suitable for relatively small event logs
- automated approach
set of difference statements; set of
(more general) general patterns; Suitable for more complex scenarios

Question 13

In conformance checking, which are the Three families of relevant techniques?

Answer 13

replay
trace alignment
behavioral alignment