Hypertext systems parse documents into content nodes connected by machine s
upported links or relationships. Many hypertext researchers claim that the
node-link relationships of hypertext provide an information organization th
at models the structure of human knowledge and should therefore facilitate
information access (Fiderio 1988). Yet, failures of information access occu
r when users lack an understanding of the overall scope and organization of
a hypertext system (Gay and Mazur 1991). To support this understanding, th
e present research incorporated expert-based domain semantics in the design
of prosthetic devices for hypertext navigation. The task domain was docume
ntation for a word processing system. In the first experiment, the pathfind
er algorithm (Schvaneveldt 1990) and cluster analysis were used to identify
a set of expert-based semantic relationships between word-processing conce
pts. The results from these analyses contributed to the design of two prost
heses to assist hypertext navigation: A hierarchical index and a local sema
ntic browser. These aids were tested in a second experiment, crossing type
of on-line documentation (semantically enhanced hypertext or an alphabetica
lly indexed text) with level of subject expertise (novice or expert). Both
performance and strategy measures suggest that the semantic prostheses impr
oved the accessibility of information for novice users without hampering ex
pert performance.