background image
- The interaction mode must be able to cope with users arriving at the site with only
a vague idea of what they are looking for.Therefore the structure of guidance pro-
vided by the system must be sensitive enough to let the users keep their vague idea
as the steering element;
- There is no such thing as a precise answer to a vague query. Anything that relates to
it can be useful, and thus any kind of information found by chance or good fortune
should be welcomed.
The characteristics of eStage's interaction mode demonstrate the expectations of a user
community requiring a platform that actively supports their work, i.e. expectations that
determine the rationale of an information platform in the specific context of fostering
creative talents.These users clearly do not want a portal which simply offers a virtual
tour through a puppetry museum.Visitors to a museum (whether real or virtual) expect
an overview of the virtual collections they are encountering, to be guided through these
collections, and to have
learned something by the
end of the tour. Due to dif-
fering user expectations,
eStage did not align its user
interaction mode with the
museum metaphor; instead
a hybrid structure was
tested which links docu-
ments/records in the same way traditional textbooks and guidebooks do. Integrated into
a retrieval environment like eStage, this can be helpful in guiding users towards deeper
explanations or further details on a specific topic. Users can broaden the scope of a cer-
tain topic, or even compensate for missing knowledge by adding explanations to the
pages they visit.Vague concepts give space for a number of development directions that
become concrete in the reflection of the material encountered, and this kind of retrieval
rationale responds to a desire for serendipity in the discovery process.
Semantic mapping the
annotation of documents
and associating documents
to concept hierarchies
depends in eStage on auto-
matic text analysis. Contrib-
utions are accepted in
HTML format only, and
these texts are then anno-
tated automatically. eStage employs a `vector voting' strategy to assign textual documents
to concept hierarchies. Put simply, the more votes a concept receives, the stronger the
link becomes between the text and concept. Classification in eStage generates an XML
header and associates the header to its corresponding document.The header consists of the
annotated concepts of the concept hierarchies (CHs), but the concepts themselves are
represented by pointers to their corresponding CH nodes, not mentioned explicitly.
293
Collaborative Mechanisms
and Technologies
159
The eStage search facility
eStage
The eStage search facility
eStage
293 The use of XML comes into its own here too, with the possibilities it brings for multiple simultaneous links
and richer multidirectional linking.
TWR2004_01_layout#62 14.04.2004 14:07 Uhr Seite 159