FOA Home | UP: Inference beyond the \Index

Deep interfaces

Probably because interface technology did not support rich graphical presentations and pointing devices until late in the development of computer science, the human-computer interface (HCI) is often an after-thought in software engineering. This bias remains today in search engine designs which assume the indexing and match algorithms can be separated from the presentation of result [Harman92b] [REF654] .

Sometimes interface design is approached as a data visualization task [Veerasamy96] . Korfage VIBE presentation [Korfhage95] highlights a user and author's perspectives on topical areas. Rather than assuming that there is any absolute, preferred perspective on keywords, Korfage considers what the words look like from the perspective of users and authors, respectively.

In terms of the vector space model, we can think of these as projections. The huge dimensional space of keywords, or even the still large reduced dimension representation, is still far more than we can visualize on the two dimensional computer screen. We can try to impose other dimensions (e.g. color, size) but still we must pick some projected subspace of the larger data set. Norman and Schneiderman have written extensively on the design of interfaces which are deeply connected to the user's underlying task [Norman88] [Shneiderman92] ; Marchianoni has focused particularly on interfaces for the ``information seeking'' task [Marchionini95] . As part of the Xerox PARC group, Hearst has explored a number of visualization techniques (cf. Figure (FOAref) ); she has also recently provided an extensive survey of interface technologies [Hearst99] .


Top of Page | UP: Inference beyond the \Index | ,FOA Home

FOA © R. K. Belew - 00-09-21