Process Activity Ontology Learning From Event Logs Through Gamification

Abstract

The quality of event log data is a constraining factor in achieving reliable insights in process mining. Particular quality problems are posed by activity labels which are meant to be representative of organisational activities, but may take different manifestations (e.g. as a result of manual entry synonyms may be introduced). Ideally, such problems are remedied by domain experts, but they are time-poor and data cleaning is a time-consuming and tedious task. Ontologies provide a means to formalise domain knowledge and their use can provide a scalable solution to fixing activity label similarity problems, as they can be extended and reused over time. Existing approaches to activity label quality improvement use manuallygenerated ontologies or ontologies that are too general (e.g. WordNet). Limited attention has been paid to facilitating the development of purposeful ontologies in the field of process mining. This paper is concerned with the creation of activity ontologies by domain experts. For the first time in the field of process mining, their participation is facilitated and motivated through the application of techniques from crowdsourcing and gamification. Evaluation of our approach to the construction of activity ontologies by 35 participants shows that they found the method engaging and that its application results in high-quality ontologies.

A Multi-View Framework to Detect Redundant Activity Labels for More Representative Event Logs in Process Mining

“…In order to detect redundant activity labels, relevant works [11,13,14] suggest ways to address this issue at the event log level. Sadeghianasl et al [11] proposed a contextual approach that takes control-flow relations, resources, time, and data attributes into consideration.…”

“…It relies on a weighted clustering method to combine the final results, which requires domain knowledge or ground truth to determine the best weight setting. The other method [13,14] collaboratively and interactively detects problematic activity labels by adopting a gamified crowdsourcing approach, which utilises gamification elements (e.g., badges) to encourage a large group of domain experts to identify and repair redundant activity labels.…”

“…Though previous studies have made efforts to address redundant activity labels [11,13,14], many of them have difficulties identifying activity labels with low-occurrence frequencies, or if invalid labels have been used. However, redundant labels usually occupy a small portion of event logs.…”

“…Nevertheless, many activities have such attributes, especially in healthcare logs, such as laboratory tests and observations. Other approaches [13,14] require domain knowledge to improve the data quality. Hence, we propose a multi-view framework to consider contextual information from event logs and aim to accurately detect redundant activity labels without domain knowledge even when the redundant activity labels are infrequent and contain numerical values as attributes.…”

“…Event logs with such redundant activity labels can suffer a significant quality loss in the discovered process model, because most existing process discovery algorithms assume that every activity label in the event log is meaningful and unique. Several methods are proposed in the field of process mining to detect redundant activity labels in event logs [11,13,14]. However, none of these methods can accurately detect redundant activity labels in event logs without domain knowledge, especially when the redundant activity labels are less frequent and contain numerical values as attributes.…”