Repairing Event Logs to Enhance the Performance of a Process Mining Model

Abstract

Organizations and companies are starving to improve their business processes to stay in competition. As we know that process mining is a young and emerging study that lasts among data mining and machine learning. The main goal of process mining is to obtain accurate information from the data; therefore, in recent years, it attracts the attention of many researchers, practitioners, and vendors. However, the purpose of enhancement is to extend or develop an existing process model by taking information from the actual process recorded in an event log. One type of enhancement of a process mining model is repair. It is common practice that due to logging errors in information systems or the presence of a special behavior process, they have the actual event logs with the noise. Hence, the event logs are traditionally thought to be defined as situation. Actually, when the logging is based on manual logging i.e., entering data in hospitals when patients are admitted for treatment while recording manually, events and timestamps are missing or recorded incorrectly. Our paper is based on theoretical and practical research work. The main purpose of our study is to use the knowledge gather from the process model, and give a technique to repair the missing events in a log. However, this technique gives us the analysis of incomplete logs. Our work is based on time and data perspectives. As our proposed approach allows us to repair the event log by using stochastic Petri net, alignment, and converting them into Bayesian analysis, which improves the performance of the process mining model. In the end, we evaluate our results by using the algorithms described in the alignment and generate synthetic/artificial data that are applied as a plug-in in a process mining framework ProM.

Event Log Data Quality Issues and Solutions

“…The most applied techniques to resolve the event log data quality issues in the literature belong to the artificial intelligence, machine learning, and deep learning category, containing algorithms and approaches applied in different scenarios. Bayesian networks are a class of probabilistic graphical models that can be applied to repair an event log with missing timestamps [14] and missing events [15,16]. Additionally, long short-term memory (LSTM) is an artificial neural network in deep learning, able to predict the missing event and activity labels in event logs [17].…”

“…It should be noted that several previously mentioned approaches from the artificial intelligence, machine learning, and deep learning category are able to partly recover missing data. However, these techniques utilize external reference models, i.e., process models defined based on pre-existing process knowledge, and align the incomplete event log according to the expected behavior [15][16][17].…”

“…Furthermore, a tool was developed to allow manual reparation of the detected problematic timestamps using domain knowledge [40]. Another conformance-checking approach used to discover anomalous behavior is costbased alignment [16]. The cost-based alignment technique involves assigning costs to different types of deviations between the observed events and the process model, such as missing activities, infrequent activities, or ordering violations.…”

Video Process Mining and Model Matching for Intelligent Development: Conformance Checking

“…Synthesizing photo-realistic images from text descriptions is a challenging problem in computer vision and has many practical applications [15]. In [16] they gave a technique to repair the missing events in a log. They used the knowledge gathered from the process model and provided a technique to repair the missing event logs in a trace.…”