| The design, implementation, and applicability of virtual environments (VEs) have improved significantly in recent times due to advances in technology and interdisciplinary research contributions. However, navigation in large-scale VEs remains problematic for the users. Users often report being lost, disorientated, and lacking the spatial knowledge needed to make appropriate decisions concerning navigation tasks. Therefore, many of the new-sophisticated VE applications cannot reach their full potential in benefiting the user, unless a robust navigation assistance tool is also provided.; The primary goal of this dissertation is to develop a comprehensive navigation assistance system that helps a user navigate the VE space by traveling via routes that closely match the user's unique preferences. In short, a route preview interface is provided which highlights the characteristics of alternative routes to a destination, and assists the user in making informed personalized route selection decisions. This research is inter-disciplinary in nature drawing from diverse fields such as computer graphics, human-computer interactions, urban design, and psychology. In particular, many data mining-based methodologies and techniques are proposed in this research.; The research is divided into four phases. The focus of the first phase is automatic detection of landmarks. Landmarks are unique and prominent objects in an environment, and they play a critical role in navigation by serving as orientation aids and/or marking decision and destination points. In this phase of the research, guidelines are provided on how to automatically identify objects that could serve as visually and semantically salient landmarks in an environment. The guidelines provided for landmark detection are not limited to VE spaces, but are also applicable to real and electronic spaces (e.g. web environments and digital libraries). Experiments demonstrated the benefits gained by incorporating these guidelines in the landmark identification process.; The focus of the second phase of the research is to find alternative preferable routes of travel between the identified landmarks using the Frequent Wayfinding-Sequence (FWS) methodology. In this methodology, a modified sequence mining technique is used to discover a model of frequent sequences representing the routes of travel preferred by experienced VE users. Experiments demonstrated the scalability and efficiency of this approach, as well as the benefits derived by integrating the discovered FWS model in the navigation assistance interface of VEs.; The third phase of the research is the implementation of a route preview interface with both a text-based and an image-based component. The route preview interface is used to inform the VE users about the unique characteristics of alternative routes.{09}For each destination point, the four routes to the destination with the highest FWS-based statistical confidence are previewed. Experiments showed that the route preview interface is informative, capable of highlighting the most salient features of the routes, and provides support to help novice users select an appropriate route.; The fourth phase of the research is geared toward improving and enhancing the quality of the route preview interface. A user profile is automatically generated for each user, and it is used to personalize the route preview interface. Simulation experiments showed that the proposed personalized route preview approach can quickly learn the user's personal preferences, and it can be used to provide route choices in the preview interface that closely match the user's unique preferences. |
