=Paper=
{{Paper
|id=Vol-3098/dc_218
|storemode=property
|title=Online conformance checking to support human behavior study (Extended Abstract)
|pdfUrl=https://ceur-ws.org/Vol-3098/dc_218.pdf
|volume=Vol-3098
|authors=Gemma Di Federico
|dblpUrl=https://dblp.org/rec/conf/icpm/Federico21
}}
==Online conformance checking to support human behavior study (Extended Abstract)==
https://ceur-ws.org/Vol-3098/dc_218.pdf
Online Conformance Checking to Support Human
Behavior Study (Extended Abstract)
Gemma Di Federico
Department of Applied Mathematics and Computer Science
Technical University of Denmark
Kgs. Lyngby, Denmark
gdfe@dtu.dk
Abstract—Human behavior could be represented in form of a schema. In order to succeed in such goal, in the following
process model. Internet of Things systems allow the collection of section I elaborate the research questions.
the behavior in the reality, and the data are then processed by
Process Mining algorithms. To achieve this integration, several II. R ESEARCH QUESTIONS
challenges must be faced. The objective of this PhD project is
The overall process could be divided into three main topics
to adapt online conformance checking to be able to deal with
human-related processes. which are the data collection, the process modeling and the
mining. Firstly, there is a simulated environment, equipped
I. P ROBLEM C ONTEXT AND M OTIVATION with IoT sensors, that captures the behavior of a patient in
The project aims to contribute into the integration between form of sensors measurements. Secondly, data collected must
Internet of Things (IoT) and Business Process Management, be organized in an event log that is suitable for discovery
combining the potentiality of sensor data toward the imple- algorithms which are used to derive a reference model. This
mentation of dynamic and enriched models with the objective model represents the typical behavior of the patient. The
to support conformance checking techniques. In particular, last is the conformance, in which the behavior in the reality
the project focuses on the modeling of human behavior [8] is compared with the reference models. The derivation and
by means of process mining techniques, obtaining data from replication of processes is what process mining is intended to
IoT devices. There is a desire to dwell on aspects relating to do. Complications emerge as the nature of human processes
the use of conformance checking techniques in the context of differs from the structured one that are typically dealt with.
behavioral study. Observing human behavior for a long time, RQ1 - What is the most appropriate way to abstract
can lead to the collection of various information that can be activities from a stream of sensor measurements?
used in a plenty of application fields [10]. The use case of Are semi-supervised approaches the proper manner
my PhD project is the dementia disease, under the healthcare to organize sensor data into an event log?
field. Patients effected by dementia usually fit into a very tight Data produced by IoT systems are unstructured and of a
routine. The repetition of routines helps people to control their different level of granularity compared to the event log ex-
world in a predictable manner, adding a sense of order to the pected by Process Mining algorithms. There is a need to
days. Repetition of activities is a comfort zone. Establishing bring sensors measurements and model to the same level
a routine contributes to a sense of security and peace, which of abstraction. Actually, there is not a standard technique
helps to reduce agitation and troublesome behavior [6]. In the to organize sensors data into a structured event log. In the
middle to later stages of most types of dementia, a person may literature there are several approaches of event abstraction[15],
start to behave differently. Given that, the breaking of a routine both supervised or unsupervised. The unsupervised techniques
could be symptom of a worsening of the disease [12]. The idea group fine-granular events on the basis of strong re-occurrence
is to observe the behavior of people living with dementia to of patterns [9]. In the supervised event abstraction techniques,
identify significant variations in the behavior. To do so, an authors propose a set of dimensions based on their understand-
habits model is derived for each patient and it is continuously ing of the field and classifying them accordingly [7]. After a
compared with the actual behavior to find discrepancies. IoT first investigation of the available techniques, we would like
systems are used to capture the behavior in the reality, while to focus in the between, i.e., on semi-supervised approaches.
online conformance checking techniques to compare the actual A fully supervised abstraction could be influenced by human
behavior with the modeled (and expected) one. errors, since it is an expert who must identify every single
The work in this PhD project also aims to be generalized, activity starting from a series of proposed configurations. On
so that it can be reused in other contexts. My plan is to the other hand, an unsupervised technique does not allow
extend Process Mining techniques to be applicable in IoT, to verify the way in which the labeling was carried out, or
with the aim of contribute into verifying the compliance of how the sequence of actions for each activity was identified.
the actual IoT system execution with the expected process Therefore, a semi-supervised approach could provide both
Copyright © 2021 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0
International (CC BY 4.0).
the right level of automation and allow us to oversee and there is no certainty that a Petri net has enough expressiveness
contribute with knowledge to the activity recognition. Once to represent these processes. For this reason, the conformance
the data has been structured in an event log, we need to focus algorithm must be able to deal with the process modeling
on the process modeling language that is able to represent language selected, perhaps, also in this case, the algorithm
human-related processes. should be adapted to respond to the process requirements. To
RQ2 - Are current modeling languages able to conclude, given the large amount of data produced by IoT
represent human behavior and their context? It may systems, it is desirable that the conformance is carried out
be valid to consider stochastic models? online [2, 16].
Human related process differs from business one. Human R EFERENCES
behavior evolves over the time [5] and it is influenced by [1] N. Banovic et al. “Modeling and understanding human
the participant and by the context. The process is completely routine behavior”. In: Proc. of CHI. 2016, pp. 248–260.
guided by the participants, who decide how and when to [2] A. Burattin and J. Carmona. “A framework for on-
perform activities [3]. The concept means that there are a series line conformance checking”. In: BPM. Springer. 2017,
of variables, external or internal to the process execution, pp. 165–177.
which influence the process. These variables could also change [3] C. Di Ciccio, A. Marrella, and A. Russo. “Knowledge-
over the time [14]. For example, the factors that influence intensive processes: characteristics, requirements and
the behavior could be the seasonality, the weather, the day of analysis of contemporary approaches”. In: J. Data Se-
the week, but also a dependency between executed activities. mant. 4.1 (2015).
Based on each of them, the behavior adapts [13]. Often the [4] G. Di Federico, A. Burattin, and M. Montali. “Human
cause of change is hidden, not known a priori, making the Behavior as a Process Model: Which Language to
learning task more complicated. Hidden contexts may be ex- Use?” In: ITBPM (2021).
pected to recur, generating cycling phenomena. In this domain, [5] M. S. Feldman and B. T. Pentland. “Reconceptualizing
is a key element to keep track of these changes in order to organizational routines as a source of flexibility and
use them to adapt models. Another important fact to consider change”. In: Adm. Sci. Q. 48.1 (2003), pp. 94–118.
is the non-determinism of human behaviors. Human behavior [6] H. C. Kales, L. N. Gitlin, and C. G. Lyketsos. “Assess-
is usually characterized by the repetition of activities [1] ment and management of behavioral and psychological
and, each time an activity is executed it follows a different symptoms of dementia”. In: BMJ (2015).
execution pattern. In addition, human beings could perform [7] M. de Leoni and S. Dündar. “Event-log abstraction
multiple actions at the same time, mixing the execution of using batch session identification and clustering”. In:
activities, i.e. concurrency. Since the objective of the use case Proc. SAC20. 2020, pp. 36–44.
is to identify discrepancies in the behaviors, also the frequency [8] F. Leotta, M. Mecella, and J. Mendling. “Applying pro-
concept must be taken into account [11]. Human behavior is cess mining to smart spaces: Perspectives and research
governed by uncertainty and variability, therefore it is not challenges”. In: Lect. Notes Bus. Inf. Process. 2015.
possible to establish ex ante which is a frequent or a not [9] F. Leotta, M. Mecella, and D. Sora. “Visual process
frequent behavior. As a starting point I identified a list of maps: A visualization tool for discovering habits in
requirements that the model must fulfill [4], hence now the smart homes”. In: J. Ambient Intell. Humaniz. (2019),
objective is to position the work toward one direction. Once pp. 1–29.
the modeling language has been identified (or extended) I can [10] S. Mandal et al. “A classification framework for IoT
focus on the process conformance. scenarios”. In: BPM. Springer. 2018, pp. 458–469.
RQ3 - Are the actual mining algorithms able to deal [11] A. Rebuge and D. R. Ferreira. “Business process anal-
with unstructured scenario? Which one is the most ysis in healthcare environments: A methodology based
appropriate? Is there the need of extending these on process mining”. In: Inf. Syst. (2012).
algorithms? [12] K. SantaCruz and Jr. Daniel L. Swagerty. Early Diag-
In conformance checking, the behavior of a process model and nosis of Dementia. Tech. rep. 2001.
the behavior recorded in an event log are compared to find [13] A. Stefanini et al. “A process mining methodology for
commonalities and discrepancies. This model is the reference modeling unstructured processes”. In: Knowl. Process.
model, on which all the replay work is based, so it is very Manag. 27.4 (2020), pp. 294–310.
important that it is accurate and representative of the behavior [14] A. Tsymbal. The problem of concept drift: definitions
recorded in the log. The discovery of a process related to and related work. Tech. rep. 2004.
human behavior, unstructured by nature, is an aspect not to [15] S. J. van Zelst et al. “Event abstraction in process
be underestimated. Considering the above, there is a close mining: literature review and taxonomy”. In: Granul.
dependence between mining algorithms and modeling lan- (2020).
guages. Most of the current process mining algorithms, assume [16] S. J. van Zelst et al. “Online conformance checking:
the derivation of Petri nets and consequently conformance relating event streams to process models using prefix-
checking techniques have been developed along this line. But alignments”. In: Int. J. Data Sci. Anal. (2019).
Copyright © 2021 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0
International (CC BY 4.0).