TDT-2002 Topic Tracking at Maryland: First Experiments with the Lemur Toolkit

摘要

The University of Maryland submitted six topic tracking runs for the2002 Topic Detection and Tracking evaluation. Two runs were producedusing the Lemur language modeling toolkit, the remaining four wereproduced using an separate system coded in Perl. The Lemur runsoutperformed the Perl runs on the required condition because termfrequency information was better handled. Two of the Perl runs usednative Arabic orthography with two-best translation based on astatistical lexicon, obtaining similar results to those obtained withthe Arabic-to-English translations provided with the collection.UMIACS-TR-2003-24LAMP-TR-099