Scaling up word sense disambiguation via parallel texts
Yee Seng Chan, Hwee Tou Ng · 2005
A critical problem faced by current supervised WSD systems is the lack of manually annotated training data. Tackling this data acquisition bottleneck is crucial, in order to build high-accuracy and wide-coverage WSD systems. In this paper, we show that the approach of automatically gathering training examples from parallel texts is scalable to a large set of nouns. We conducted evaluation on the nouns of SENSEVAL-2 En-glish all-words task, using fine-grained sense scoring. Our evaluation shows that training on examples gathered from 680MB of parallel texts achieves accuracy comparable to the best system of SENSEVAL-2 English all-words task, and sig-nificantly outperforms the baseline of always choosing sense 1 of WordNet.