Visual Data Mining in Virtual Reality: An Exploratory Study
Authors
Larsen, Anton Christian Mondrup ; Hammer, Jens Lindberg ; Lindhardt, Thomas Krarup
Term
4. term
Education
Publication year
2024
Submitted on
2024-05-29
Pages
55
Abstract
This thesis examines how Virtual Reality (VR) can be combined with Visual Data Mining (VDM), meaning the use of interactive visual displays to explore large, complex datasets and discover patterns. In the era of big data, VR's immersive viewing and interaction may help people see, navigate, and understand complex information more easily. We conducted an exploratory study in a clinical context and built a VR-based VDM prototype to clarify what such a tool should include and to test where VR can support VDM processes. Our results indicate that the integration is feasible: VR can be incorporated at multiple stages of VDM. However, research with data mining experts is needed to determine whether VR actually improves visual data mining in practice. Future work should connect the visual interface to real data mining algorithms running behind the scenes so that data handling is more seamless and requires minimal manual back-end intervention.
Denne afhandling undersøger, hvordan Virtual Reality (VR) kan kombineres med Visual Data Mining (VDM), dvs. brugen af interaktive visuelle visninger til at udforske store, komplekse datasæt og opdage mønstre. I en tid med big data kan VR's omsluttende visning og interaktion hjælpe mennesker med lettere at se, navigere i og forstå kompleks information. Vi gennemførte et eksplorativt studie i en klinisk kontekst og byggede en VR-baseret VDM-prototype for at afklare, hvad et sådant værktøj bør indeholde, og hvor VR kan støtte VDM-processer. Vores resultater peger på, at integrationen er gennemførlig: VR kan indgå i flere faser af VDM. Der er dog behov for forskning sammen med eksperter i datamining for at afgøre, om VR faktisk forbedrer visuel datamining i praksis. Fremtidigt arbejde bør forbinde den visuelle grænseflade med egentlige datamining-algoritmer i baggrunden, så datahåndteringen bliver mere sømløs og kræver minimal manuel back-end-indgriben.
[This apstract has been rewritten with the help of AI based on the project's original abstract]