We propose a backoff model for phrasebased machine translation that translates unseen word forms in foreign-language text by hierarchical morphological abstractions at the word an...
Slavic languages are characteristic by their relatively high degree of word order freedom. In the process of automatic generation from an underlying representation of the content, ...
It is shown that basic language processes such as the production of free word associations and the generation of synonyms can be simulated using statistical models that analyze th...
Exploiting lexical and semantic relationships in large unstructured text collections can significantly enhance managing, integrating, and querying information locked in unstructur...
We show that unsupervised part of speech tagging performance can be significantly improved using likely substitutes for target words given by a statistical language model. We choo...