Proceedings of the Morpho Challenge 2010 Workshop

Abstract

In natural language processing many practical tasks, such as speech recognition, information retrieval and machine translation depend on a large vocabulary and statistical language models. For morphologically rich languages, such as Finnish and Turkish, the construction of a vocabulary and language models that have a sufficient coverage is particularly difficult, because of the huge amount of different word forms. In Morpho Challenge 2010 unsupervised and semi-supervised algorithms are suggested to provide morpheme analyses for words in different languages and evaluated in various practical applications. As a research theme, unsupervised morphological analysis has received wide attention in conferences and scientific journals focused on computational linguistic and its applications. This is the proceedings of the Morpho Challenge 2010 Workshop that contains one introduction article with a description of the tasks, evaluation and results and six articles describing the participating unsupervised and supervised learning algorithms. The Morpho Challenge 2010 Workshop was held at Espoo, Finland in 2-3 September, 2010.