The usage of Process Mining
Martin Macák
24. 10. 2019
State of the Art – Process Mining domains
∙ Healthcare
∙ Manufacturing
∙ Finance
∙ Public sector
∙ Usability
∙ Robotics, industry 4.0
∙ Utility
∙ Advisory, audits
∙ Biology
∙ Agriculture
∙ ICT
∙ Education
∙ Logistics
∙ Security
∙ Call center
∙ Entertainment
∙ Garment
∙ Retail
∙ Hotel
More details: [1]
State of the Art – Cybersecurity domains
∙ Used in domains:
∙ Network (IS, DNS, IDS, websites)
∙ Smart grids (anomalous behaviour of energy usage)
∙ Smartphones (social engineering attacks, malwares)
∙ Banking (frauds, security deviations)
∙ Industrial Control Systems (cyberattacks)
∙ Business processes (anomalies, deviations)
State of the Art – Techniques
1. Target period of the analysis
∙ past
∙ present
2. Business process awareness
∙ with business knowledge
∙ without any business knowledge
3. Analysis of a discovered process model
∙ visually
∙ programmatically
4. Detection technique
∙ outlier behavior detection
∙ abnormal behavior detection (only in supervised analysis)
∙ conformance checking
Process discovery: DNS traces
∙ Event log built from DNS traces (caseID, activity,
timestamp)
∙ caseID= client, DNS Server
∙ activity = query/response, type
∙ Detection of spambots
Process discovery: DNS traces
[2]
Model comparison: Smart Grids
∙ Anomaly detection of power consumption
∙ Classification of consumption to levels
∙ Then they discover graphs of consumption per short period
∙ Time-evolving graph approach: comparing consecutive
graphs
∙ They chose Hamming distance and cosine similarity
measure
Model comparison: Smart Grids
[3]
Conformance checking: Smartphones
∙ Attack: user activated a malicious URL, which resulted in
downloading personal user data via known vulnerability
∙ They designed a model of this attack from OS-generated
information about performed actions, browser history, and
network connection log
∙ Token-based replay with this model
Conformance checking: Smartphones
[4]
Demo: Find potential problems in the company
∙ From event logs, we get the model of real people’s
behavior from the past
∙ We then replay their behavior and find potential attacks
or possible security threats
∙ For example, we can detect:
∙ Abnormal behavior (something was done differently)
∙ The absence of the action (e.g., someone forgot to pay)
∙ We only get suspects, so there is still a need for further
investigation
What can WE do?
∙ Generally:
∙ Discover the process from event logs
∙ Replay the log on top of that model and visually detect
deviations
∙ Filter the event log and manually analyse the desired
cases
∙ Find the deviations in an event log from the existing
model using conformance checking
∙ Real-time conformance checking
∙ The biggest challenge might be to find a data to analyze
and to clean them
∙ However, you can create your own prototype event logs
(just like I did in the Demo)
What can WE do?
∙ Specifically:
∙ Your project should be interesting for you
∙ You need to achieve something that can be presented
∙ Real examples of a project:
∙ Process discovery of a group of GitHub repositories (e.g.,
Big Data open source tools)
∙ Process discovery from real datasets, for example:
https://data.gov.cz/datové-sady,
https://data.brno.cz/data/
∙ Process analysis of the behavior of employees / students
∙ Process analysis of the user behavior from a game
∙ Process Mining on publicly available logs, for example:
https://data.4tu.nl/repository/collection:
event_logs
∙ Real-time conformance checking using PM4Py
Resources
[1] C. dos Santos Garcia, A. Meincheim, E. R. F. Junior, M. R.
Dallagassa,D. M. V. Sato, D. R. Carvalho, E. A. P. Santos, and E. E.
Scalabrin,“Process mining techniques and applications - a systematic
mappingstudy,”Expert Systems with Applications, vol. 133, pp. 260 – 295,
2019.doi: https://doi.org/10.1016/j.eswa.2019.05.003. [Online].
[2] J. Bustos-Jiménez, C. Saint-Pierre, and A. Graves, “Applying process
mining techniques to dns traces analysis,” in 2014 33rd International
Conference of the Chilean Computer Science Society (SCCC) , Nov 2014.
doi: 10.1109/SCCC.2014.9. ISSN 1522-4902 pp. 12–16
[3] S. Bernardi, R. Trillo-Lado, and J. Merseguer, “Detection of integrity
attacks to smart grids using process mining and time-evolving graphs,” in
2018 14th European Dependable Computing Conference (EDCC), Sep.
2018. doi: 10.1109/EDCC.2018.00032 pp. 136–139.
[4] L. Hluchý and O. Habala, “Enhancing mobile device security with
process mining,” in 2016 IEEE 14th International Symposium on
Intelligent Systems and Informatics (SISY) , Aug 2016. doi:
10.1109/SISY.2016.7601493. ISSN 1949-0488 pp. 181–184.