Analyzing user interactions with hypermedia systems

Abstract

The unprecedented freedom for users to control the scope and sequence of their interactions with hypermedia systems presents many challenges to those who design and study these systems in educational settings. Early efforts to develop hypermedia systems revealed that the node-link structure of such systems is both advantageous and problematic (Conklin, 1987). When users have the freedom to follow any of a multitude of link permutations, disorientation often results. Further, without appropriate training, novice users do not possess the strategies necessary for effective "browsing" of large hypermedia documents (Duffy & Knuth, 1991). It has also been noted that the purpose for using the system or the task that the users are engaged in can influence patterns of interaction with hypermedia systems (Nelson, 1991). Many designers, therefore, advocate that features such as visual maps, database search facilities and guided tours be included in hypermedia systems to alleviate some of these problems (e.g. Hammond, 1989; Laurel, 1990, 1991).With the emergence of hypermedia systems as a major architecture for educational and other information-oriented software comes the related problem of how to document and analyze user interactions with such systems for the purposes of research and evaluation. There are a variety of interface design strategies that impact on how a system performs and should be evaluated. Many hypermedia systems to date have employed a "browsing" interface, but alternative approaches are also emerging (Nelson & Palumbo, 1992). Regardless of the type of interface, many questions can be generated when studying the interactions of users with hypermedia systems. For example, how many users chose to follow a particular link, and why was one link chosen over another? How does the choice of one link affect choices of subsequent links? When are graphic images, animations and video segments accessed? What user tasks are appropriate for guiding interaction with the system? What kinds of strategies do users develop while working with hypermedia systems? These and many other questions need answers when designing, developing and evaluating hypermedia applications for education and other settings, and provide the focus for this short article.A wealth of user interaction data can be easily collected within many hypermedia development environments in order to study aspects of the interface, including the nature of user navigation patterns, the time spent at each node and the use of help and orienting facilities. The data can represent the paths a user follows through the system, and the choices made at each node in the system. The problem is that because of the nature of this data, traditional methods of analysis such as surveys or pretest-posttest designs, are not particularly effective for determining usability or comparing alternative interface designs. Researchers have had to develop new techniques for analyzing patterns of user interaction in order to evaluate the design and effectiveness of hypermedia systems (Misanchuk & Schwier, 1992). There is a need to categorize and compare groups of users in order to compare the effectiveness of alternative system features, as well as describing characteristics of interaction by individual users within the same system.