HyperContext: A Model for Adaptive Hypertext

HyperContext is a 3-layer model for adaptive hypertext. The model supports the use of multiple interpretations of objects of information. Whenever an object is accessed, HyperContext determines the context in which the access has taken place and presents the appropriate interpretation to the user. 1 Overview of Adaptive Hypertext Adaptive Hypertext Systems (AHSs) generally provide capabilities for adaptive navigation, adaptive presentation, or both (Brusilovsky, 1996). Adaptive presentation enables the information content of objects (nodes) to adapt to the user,s expectations or requirements. For instance, a student new to user modeling may be unfamiliar with certain terminology in a paper on the subject. An AHS may automatically expand the text to include the meanings of terms. Adaptive navigation, on the other hand, guides a user through a hypertext by recommending those links likely to lead to information relevant to the primary task, or by changing the destinations of links. AHSs usually separate the information base from the adaptive agent (Mayfield, 1994). The agent determines the requirements of the user, and provides a “virtual view” of the underlying information base. The user of an adaptive hypertext system browses, or navigates, through a virtual hyperspace in which the underlying information is represented by a structure which dynamically changes to accommodate the needs and requirements of that user. There are several methods used to elicit the needs and requirements of users. The adaptive agent may interact with a user model (see, e.g., Kobsa et al., to appear). Alternatively, usage patterns may be observed and subsequently generalised or specialised into search strategies (e.g., Armstrong et al., 1995, and Rabinowitz et al., 1995). Matching users with the appropriate strategy is generally performed either by evaluating a user query or by observing user access paths (i.e., recording which links in a hypertext are followed) and then offering the user “short cuts” to the information likely to be of interest (either based on prior knowledge or computed dynamically). 2 An Example of Adaptation in HyperContext Two users, Eric and Pat, are browsing through the same HyperContext system, starting from different locations. Eric is interested in the architecture of cathedrals in South-East England, whereas Pat is searching for information about different organs in Kent. The links they follow eventually lead them to the same node, which contains general information about Canterbury Cathedral. Since the node has been accessed from different locations related to the different tasks, the way the node is linked to other objects is changed, so Eric will see links to other resources on the Cathedral,s architecture, whereas Pat will be presented with links to more information about