Knowledge COMmunity for Efficient TrAining through Virtual Technologies


A short description 
Related publications 
Deliverable on Process Minining in Kometa v. 1.0  
Deliverable on Process Minining in Kometa v. 2.0  
The distribution package - PmKOMETA 
The distribution package - VITE 
Authors & Aknowledgement 

A short description

The KDDE lab is the research partner of the project Kometa. In the project, KDDE researchers are aimed at the synthesis of advanced process mining techniques, in order to mine data collected in event logs and produced by tracking the training activities of a group of technicians involved in the aqueduct maintenance. The goal is the discovery of the hidden process models behind the training sessions. Discovered models can be employed to improve the plan of the training activities, as well as to check the conformance of new training sessions.

Project details (in Italian): Azione 1.4 “Promozione di nuovi mercati per l’innovazione” – Avviso pubblico INNOLABS- approvato con A.D. n. 13 del 08/02/2017, A.D. n. 37 del 28/03/2017 e A.D. n. 43 del 10/04/2017, tipologia “Knowledge Community”.

Deliverable on Process Minining in Kometa v. 1.0 (in Italian)

Deliverable on Process Minining in Kometa v. 2.0 (in Italian)

Related publications

Appice, N. Di Mauro, D. Malerba, Leveraging Shallow Machine Learning to Predict Business Process Behavior, IEEE World Congress on Services 2019, IEEE, July 2019.

P. Ardimento, N. Boffoli and C. Mele, A text-based regression approach to predict bug-fix time, Complex Pattern Mining: New Challenges, Methods and Applications Book in Studies in Computational Intelligence, Accepted 2019, to be published

V. Pasquadibisceglie, A. Appice, G. Castellano, D. Malerba, Using Convolutional Neural Networks for Predictive Process Analytics, ICPM 2019, IEEE, pp 129-136

Top of this page


PmKometa implements an accurate data-centric approach to predict business process behaviour. It resorts to shallow machine learning techniques, which are performed as a part of a holistic approach that combines feature construction, local and global learning, classification and regression algorithms. The empirical evaluation performed by Appice ed al (SCC 2019) shows that, despite the emerging attention towards deep learning also in predictive process mining, stacking feature construction and shallow machine learning algorithms can still outperform various process predictor competitors (included deep learning ones). It is implemented in Java and used WEKA toolkit.

jar Description
PmKOMETA This rar bundle contains : pmkometa.jar. and the bat file to run the demo with the benchmark event log helpdesk log.


VITE implements a process mining software to predict the score of an exam based on data collected during the training session of the student. It is implemented in Java and used WEKA toolkit.

VITE (ver 1.0) This rar bundle contains: vite.jar. and the bat file to run a demo with artificial example data.
VITE (ver 2.0) under development

Warning: Both PmKometa and VITE are free for evaluation, research and teaching purposes, but not for commercial purposes.
Please Acknowledge

Top of this page

Project KDDE team

  • Annalisa Appice
  • Pasquale Ardimento
  • Michelangelo Ceci
  • Nicola Di Mauro
  • Stefano Ferilli
  • Antonietta Lanza
  • Donato MALERBA


    Name Email address Tel. number Fax
    Annalisa Appice +39 080 5443262 +39 080 5443262

    Top of this page