Exemplar-based word-space model for compositionality detection: shared task system description

In this paper, we highlight the problems of polysemy in word space models of compositionality detection. Most models represent each word as a single prototype-based vector without addressing polysemy. We propose an exemplar-based model which is designed to handle polysemy. This model is tested for compositionality detection and it is found to outperform existing prototype-based models. We have participated in the shared task (Biemann and Giesbrecht, 2011) and our best performing exemplar-model is ranked first in two types of evaluations and second in two other evaluations.