The Tanl Tagger for Named Entity Recognition on Transcribed Broadcast News at Evalita 2011

摘要

The Tanl tagger is a flexible sequence labeller based on Conditional Markov Model that can be configured to use different classifiers and to extract features according to feature templates expressed through patterns provided in a configuration file. The Tanl Tagger was applied to the task of Named Entity Recognition (NER) on Transcribed Broadcast News of Evalita 2011. The goal of the task was to identify named entities within texts produced by an Automatic Speech Recognition (ASR) system. Since such texts do not provide capitalization, punctuation or even sentence segmentation and transcription is often noisy, this represents a challenge for state of the art NER tools. We report on the results of our experiments using the Tanl Tagger as well as another widely available tagger in both the closed and open modalities.