SUMMaR: Automatic Text Summarization
Funded by the German Ministry for Education and Research (BMBF)
project management: Professor
Dr. Manfred Stede
staff: Dr. Stefanie Dipper, Heike Bieler, Arthit Suriyawongkul
student assistants: Annika Neumann, Johannes
Schröder, Nikolaus Werner, Lena Wittke
duration: november 2004 - october 2007
In the SUMMaR project, we developed a prototypical text summarization
system. Our goal was to produce high quality summaries at a high level
of robustness. To this end, we combined symoblic, linguistic methods
with statistic approaches.
The
overall architecture of our system is modular, the
interface between the module is realised via the xml-based interchange
format PAULA.
The system was developed for German and extended for
English.
It can be extended easily for other languages as well.
The characteristics of selected text genres (news, political speeches,
film- and hotel reviews, press releases) have some influence on
the system, but in general it is genre-independent.
SUMMaR is part of the BMBF project PINK ("Plattform fuer INtelligente Kollaborationsportale"), a consortium of companies and universities from Berlin/Brandenburg, funded in the framework "Innovative regionale Wachstumskerne".
More information
Components of the summarization system
- Layout-Recognition (xdoc)
- Textsortspecific Structure Recognition (idoc)
- Term Weighting (Calculation of Sentence Relevance)
- Theme Recognition
- Discourse Marker Recognition
- Sentence Selection
- Visualization
 |
 |
 |