Recall of landmarks in information space

Research on navigation and landmarks in physical space, information space and virtual reality environments indicates that landmarks play an important role in all types of navigation. This dissertation tackles the problem of defining and evaluating the characteristics of landmarks in information space. This work validates a recent theory that three types of characteristics, structural, visual and semantic, are important for effective landmarks. This dissertation applies concepts and techniques from the extensive body of research on physical space navigation to the investigation of landmarks on a web site in the World Wide Web. Data was collected in two experiments to examine characteristics of web pages on the University of Pittsburgh web site. In addition, objective measurements were made to examine the characteristics of web pages with relation to the experimental data. The two experiments examined subjects' knowledge, use and evaluation of web pages. This research is unique in research on web navigation in its use of experimental techniques that ask subjects to recall from memory possible navigation paths and URLs. Two measures of landmark quality were used to examine the characteristics of landmarks; one, an algorithm that incorporated objective measures of the structural, visual and semantic characteristics of each web page, and the second, a measure based on the experimental data regarding subjects' knowledge and evaluation of the page. Analysis of this data from a web space confirms the tri-partite theory of characteristics of landmarks. Significant positive correlations were found between the objective and subjective landmark measures, indicating that this work is an important step toward the ability to objectively evaluate web pages and web site design in terms of landmarks. This dissertation further suggests that the characteristics can be utilized to analyze and improve the design of information spaces and lead to more effective navigation by users.