Ενίσχυση της εξόρυξης διεργασιών με μηχανική μάθηση για τον εντοπισμό προτύπων στην ανθρώπινη συμπεριφορά

Abstract

The study of human behavior is a broad and interdisciplinary field. A variety of research methods has been employed in order to gain insights into the complexities of human actions. The process of identifying and classifying specific activities or actions that individuals perform based on data collected from sensors is a rapidly evolving field that benefits from several technologies which provide the necessary tools for collecting, processing, and analyzing data related to human movements and behaviors. The combination of sensor data and Machine Learning algorithms, enable analyzing data using Human Activity Recognition (HAR) techniques, which yield valuable insights into behavior, preferences, and daily routines as well as identify human behavior disorders, anomalies and patterns. This is particularly crucial for promoting healthier and more independent living, aiding those with disabilities, and providing support to caregivers and medical professionals. Α field that has contributed significantly to the effort to achieve these goals is the Process Mining field. Process mining relies on event logs, enables the analysis of them, offering insights into the workflow and sequence of activities. As human behavior is not structured enough to be represented using a process model, more robust techniques need to be incorporated within process mining. Change Point Detection emerges as a potential tool for identifying disorders in human behavior and along with Clustering methods which are effectively applied to categorize and group diverse human activities seem to be a solution to the aforementioned challenge. This research work aims to develop an advanced framework that combines Process Mining techniques, Change Point Detection, and Clustering methods to achieve a holistic analysis of human behavior patterns. The study explores the integration of these diverse techniques to enhance the understanding of complex human activities, with a specific focus on identifying patterns and deviations indicative of potential disorders. Human activities are represented in a sequential manner. By using also unsupervised Machine Learning techniques, which enable handling and analyzing the inherent variability in human behavior, makes it possible to identify and detect changes throughout the dynamic process. The experiments reveal that the proposed framework effectively identifies anomalies or disorders in human behavior and offers insights into the workflow and sequence of activities. Furthermore, it facilitates the discovery and modeling of behavioral patterns, along with the identification of deviations in human activities, which is crucial for analyzing unstructured human behaviors.