In this page we present some projects at DFKI, in which
Automatic Hyperlinking was developed or has been applied (see the
"applications" menu point in the left frame).
Within the further development of this
strategy we focus on information and knowledge management applications, and we are looking for a substantial contribution to the Semantic Web. Important in this context is the exploration of different strategies for the
visualisation of complex information structures.
Automatic Hyperlinking is performed by computer programs that insert
independently references into hyperdocuments. Hyperdocuments are
documents, which are usually read and visualized with the help of a
browser, whereby navigating in larger documents (also in the Web) is
supported by the links contained in the document.
A computer program, which is able to automatically add to a document
links/references, is so fast, compared to the manual hyperlinking, that one can afford an essentially higher processing volume (larger
data amount and/or higher reference density) and a higher technical
complexity (structured links bundles instead of single
It has already been detected, that language technologies
can play a substantial role in the automation of hyperlinking, since they can help in reaching a better quality and a high consistency of the hyperlinking procedure.
This approach is particularly investigated and developed at the DFKI:
Language technology can contribute with the help of well-know
statistic and rule-based procedures to the automatic
recognition of the application-relevant terms and expressions in
the documents. Language technology is characterised by robustness in relation with variations at morphologic and syntactic level. This ensures
an increased consistency of the automatic hyperlinking procedure, since all
possible variants of a term to be linked can be recognized and associated with the corresponding references. Language
technology, including semantic analysis, can for example treat on an
equal basis synonyms, whereby thesaurus entries,
with their conceptual relations, can enrich the documents (vvia links). And finally language
technology can detect certain syntactic functions (subject, object
etc..) in sentences and those can play a role for establishing preferences and compactness of hyperlinking.
The automation concerns not only the recognition in
the text of application-relevant expressions, but also the adding of the
responsible for the visualization of the link structure. This
contributes also substantially to the consistency of the linking, as is
ensured that all equivalent expressions receives a uniform hyperlinking