Where: TU Darmstadt / GRIS, Zoom: https://tu-darmstadt.zoom.us/j/86845586436?pwd=RUJiWm1QdWJ4VDg3MU93WUNOWWFTQT09
Who: Shahrukh Badar (Author), Prof. Dr. Arjan Kuijper (Supervisor), Dipl.-Inf. Dirk Burkhardt (Advisor/Co-Supervisor)
What: Master Thesis – “Process Mining for Workflow-Driven Assistance in Visual Trend Analytics”

Abstract:

In today’s data-driven world, a large amount of data is being generated daily. This data is generated by different sources, such as social networking platforms, industrial machinery, daily transactions, etc. The companies or businesses are not only generating data but also utilizing them to improve their processes, business decisions, etc. There are several applications and tools that help users to analyze this big data in-depth, by providing numerous ways to explore it, including different types of visualization, pivoting, filtering, grouping data, etc. The challenge with such applications is that it creates long and heavy learning curves for users, who need to work with such applications. Many systems are often designed for a specific purpose, and therewith to know how a single system works is not enough. To enable a better work entrance with such an analytical system, a kind of adaptive assistance would be helpful. So, the system would hint the users regarding his previous work and interaction, what next action might be useful. The thesis aims to face this challenge with process-driven assistance that is applied to the visual trend analytics domain. The goal is, based on previous users interactions and solved tasks, to assist further users in their work. Therefore, a universal visual assistance model is defined and acts also as the main contribution, based on defined interaction event taxonomy. This concept is applied on the Visual Trend Analytics domain on the SciTic reference system, This “SciTic – Visual Trend Analytics” is connected with different data sources and provides analysis of scientific documents. The interaction model provides assistance in terms of recommendations, where the user has an option either to apply a recommendation or ignore it. The solution provided in this thesis is model-based and utilizes the potential of Process Mining and Discovery techniques. It is started by creating an event taxonomy by identifying all possible ways of user interactions on the “SciTic – Visual Trend Analytics” web application. Next, enable the “SciTic – Visual Trend Analytics” web application to start logging events chronologically based on predefined taxonomy. Later, these events log is converted into Process Mining log format. Next, it applies the Process Discovery algorithm “Heuristics Miner” on these log data to generate a process model, which shows the overall flow of user interaction along with the frequencies. Later, this process model is used to provide users with recommendations.