Business processes are often monitored by transactional information systems that produce massive dataset called event logs. Such logs contain the process execution traces, typically characterized by heterogeneous and high-dimensional data. Process mining techniques oer a great opportunity to gain valuable knowledge hidden in the data to be used for analysing the multiple characteristics of processes (i.e. perspectives in process mining, like structural aspects, activities, resources, data and time). Therefore, raw data must be encoded into a suitable format that can be more conveniently provided to the mining algorithms. However, most of the existing process encoding techniques focus on the control- ow perspective, i.e. only encode the sequence of activities that characterize a trace, leaving out other process perspectives that are fundamental for describing the process behavior in all its aspects. In this paper we address the problem of computing a concise and informative representation of execution traces that considers the multiple perspectives of the process behavior. We propose a holistic approach that computes trace embedding able to capture patterns of dependencies between the perspectives that are lost in a one-dimensional analysis and, at the same time, it is unsupervised, meaning that no a priori knowledge is needed. The experiments conducted on two real life logs demonstrate that our proposed embedding is appropriate to concisely describe the multiple and various characteristics of the processes, and that the proposed method outperforms existing trace encoding techniques. Furthermore, the embedding includes the elapsed time between events as an additional feature to make us capable to use it as a further dimension of analysis.

A multi-perspective approach for the analysis of complex business processes behavior

Antonella Guzzo
Methodology
;
Antonino Rullo
Methodology
;
2021

Abstract

Business processes are often monitored by transactional information systems that produce massive dataset called event logs. Such logs contain the process execution traces, typically characterized by heterogeneous and high-dimensional data. Process mining techniques oer a great opportunity to gain valuable knowledge hidden in the data to be used for analysing the multiple characteristics of processes (i.e. perspectives in process mining, like structural aspects, activities, resources, data and time). Therefore, raw data must be encoded into a suitable format that can be more conveniently provided to the mining algorithms. However, most of the existing process encoding techniques focus on the control- ow perspective, i.e. only encode the sequence of activities that characterize a trace, leaving out other process perspectives that are fundamental for describing the process behavior in all its aspects. In this paper we address the problem of computing a concise and informative representation of execution traces that considers the multiple perspectives of the process behavior. We propose a holistic approach that computes trace embedding able to capture patterns of dependencies between the perspectives that are lost in a one-dimensional analysis and, at the same time, it is unsupervised, meaning that no a priori knowledge is needed. The experiments conducted on two real life logs demonstrate that our proposed embedding is appropriate to concisely describe the multiple and various characteristics of the processes, and that the proposed method outperforms existing trace encoding techniques. Furthermore, the embedding includes the elapsed time between events as an additional feature to make us capable to use it as a further dimension of analysis.
Process Mining, Trace Embedding, Multi-Perspective Analysis, Deep Learning.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/20.500.11770/323754
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 2
  • ???jsp.display-item.citation.isi??? 2
social impact