Criteria of Credibility for Collaborative Virtual Environments
Jolanda G. Tromp
Communications Research Group,
Computer Science Department
University of Nottingham, NG7 2RD
ABSTRACT
Mythesis is aimed at developing guidelines for usability evaluation and design of collaborative virtual environments; creating 3D interfaces for distributed participants.
KEYWORDS:
multi-user virtual reality, methodology of usability testing, interaction design
INTRODUCTION
Collaborative Virtual Environments (CVEs) are a novel application area of computing technology, demanding an understanding of computer mediated human collaboration and human-computer interaction in 3D virtual spaces. A general tendency to ignore or minimise VE evaluation has been observed (Durlach & Mavor, 1995). To date, there are no systematic guidelines for the design and evaluation of CVEs, although work is on progress for single user VEs, to find design principles (Kaur, 1997) and a user requirements analysis method (Parent, 1998). The hypotheses for the thesis work reported here are based on the work of Kaur (1997), in order to extend her pioneering work on single user human-computer interaction in VEs.
THE HYPOTHESES
Hypothesis 1: Interface design guidelines are needed specifically for
collaboration in CVEs.
Hypothesis 2: General patterns of collaboration can be predicted.
Hypothesis 3: Collaboration design properties can be predicted.
Hypothesis 4: New design properties support CVE collaboration. To test H1, the need for CVE guidelines, two studies were made, looking at CVE usability problems and CVE design problems. First, the exploration of CVE usability problems involved an Inspection (e.g. Cognitive Walkthrough and a Heuristic Evaluation) based on Nielsen (1994. This Inspection method was redesigned to inquire into the collaborative aspects of the CVEs and applied to the COVEN Platform (a CVE being build on Division's dVS_, a leading VR product). The results of this inspection are applicable to many CVE systems and applications, not just those developed by COVEN, and identified a significant number of usability issues at system, interaction and application levels (Steed and Tromp, 1998), such as tensions between 3D object representation and 3D object interaction, and tensions between physically remote user collaboration and network latency. Second, the exploration of CVE design problems involved interviewing CVE designers. To date four CVE designers have been interviewed. The preliminary results show that these designers show a tendency to forget about end-user usability issues, do not make use of HCI design guidelines, and feel they lack the skills to create artistically satisfying CVEs. To test H2 a model of collaboration in CVEs has been developed, and a method to analyze group collaboration in CVEs has been created using this model. The method
identifies atomic actions for users involved in focused and unfocused collaboration, and is tested on video recordings of CVE user interaction. The analysis will reveal sequences of collaborative activity (Bales, 1951). H3 is tested by creating a Hierarchical Task Analysis (HTA) for CVE collaboration. Predictions for usability problems will be made from the HTA, by describing the difficulties likely to occur if the interface properties needed to perform those collaborative actions are missing or inadequate. Subsequently, user behaviour during representative collaborative tasks is observed using the group collaboration analysis method. These experiments act as the control condition for experiments to test H4. Implementing the design properties in a CVE is hypothesized to improve multi-user collaboration by supporting the users during each stage of their collaboration, and so avoiding usability problems. To test the impact of the design properties a controlled studies during network trials of the application have been planned. The test condition is the third iteration COVEN platform design, improved by implementation of the missing design properties. Task performance is assessed for both groups with a post-study test. The observations of breakdown in collaboration will be checked against the predictions made for usability problems. A good match will support the desirable design properties (H3). H4 will be tested by comparing results for the 2 conditions. Improvements in collaboration are defined as fewer usability problems, better task performance or lower task completion times, and general improved satisfaction.
CONCLUSIONS
Results from this research will be used to define and refine a set of design properties and evaluation methods for CVEs. The results can be used to help develop guidelines for CVE designers and evaluators.
ACKNOWLEDGEMENTS
Thanks go to Dr. Steve Benford, Kulwinder Kaur, Prof. Alistair Sutcliffe, Prof. John Wilson. Jolanda Tromp is employed on the COVEN Project (ACTS N. AC040).
REFERENCES
Bales, R.F., (1951). Interaction Process Analysis; A Method for the Study of Small
Groups, Addison-Wesley Press, Cambridge.
Durlach, N.I., Mavor, A.S., (1995). Virtual Reality: Scientific and Technological
Challenges, National Research Council, National Academy Press, USA. Kaur, K. (1997). Designing Virtual Environments for Usability. Doctoral Consortium
Paper, in: IFIP Proc. Of INTERACT'97 Conference on Human-Computer
Interaction, Sydney, Australia.
Nielsen, J, Mack, R.L, (1994). Usability inspection methods, John Wiley and Sons,
New York, NY.
Parent, A., (1998). Designing life-like virtual environments, submitted to Presence:
Teleoperators and Virtual Environments, MIT Press.
Steed, A., Tromp, J.G., (1998). Experiences with the Evaluation of CVE Applications,
to appear in: Proc. of Collaborative Virtual Environments, 2ndCVE98
Conference, Manchester, June 17-19.
