Supporting Data Interaction and Hybrid Asymmetric Collaboration Using Virtual Reality Within the Context of Immersive Analytics

Abstract: Immersive display and interaction technologies have rapidly evolved in recent years, offering advanced techniques compared to traditional Human-Computer Interaction. Computer-generated Virtual Environments viewed with stereoscopic depth perception and explored using 3D spatial interaction can represent more accurately how humans naturally interact in the real world. Data analysis is a promising area of application for such technologies, holding potential to promote intuitive interaction, user engagement, collaboration, and data curiosity, as well as to foster appropriate contextual visualization. Even when techniques such as Machine Learning and Data Mining assist with the analysis of data, human interpretation, contextualization, and meaning making are still needed. The design of immersive data visualization and interaction is challenging due to the complexity of the involved technologies and human factors, which calls for an interdisciplinary research effort.The focus of this thesis is to investigate means of exploration, interaction, and collaboration using Virtual Reality and 3D gestural input in immersive environments within the context of spatio-temporal data analysis. Based on existing literature as well as following an applied and interdisciplinary research approach, a design space for this type of Immersive Analytics is defined. The emphasis on spatio-temporal data is relevant across various real-world contexts and scenarios, such as sociolinguistics and climate analysis, given that data collected nowadays commonly feature descriptors of where and when they were captured. An immersive data analysis system has been implemented and evaluated across three virtual environment iterations. Two core themes from a user-centered perspective are interaction and collaboration. The design of useful and engaging 3D gestural interaction techniques support the conduction of typical analytical tasks that aid the data exploration and thus the discovery of insights. Furthermore, data analysis is seldom a solitary activity, but can be conducted in collaboration with multiple analysts, who combine their knowledge to interpret and discuss the discoveries. For this purpose, the concept of Hybrid Asymmetric Collaboration is defined, aiming to facilitate an envisioned broader analytical workflow that assumes a mixture of immersive and non-immersive interfaces (hybrid) as well as distinct user roles (asymmetric). To bridge data analysis across heterogeneous interface types, the design of visual information cues is investigated to support foundational aspects of collaboration, such as awareness, common ground, reference, and deixis.The conducted research has been empirically evaluated using a combination of standardized and custom methods in a total of six main studies. The outcomes of these studies allow for reflections and the proposal of design guidelines for collaborative data interaction in immersive spaces.