A Part-Of-Speech Lexicographic Encoding for an Evolutionary Word Sense Disambiguation Approach

This work proposes a novel distributed scheme based on a part-of-speech tagged lexicographic encoding to represent the context in which a particular word occurs in an evolutionary approach for word sense disambiguation. Tagged dataset for every sense of a polysemous word are considered as inputs to supervised classifiers, Artificial Neural Networks (ANNs), which are evolved by a joint optimization of their structures and weights, together with a similarity based recombination operator. The viability of the approach has been demonstrated through experiments carried out on a representative set of polysemous words. Comparison with the best entries of the Semeval-2007 competition has shown that the proposed approach is competitive with state-of-the-art WSD approaches.