Lexicon induction and part-of-speech tagging of non-resourced languages without any bilingual resources

摘要

We introduce a generic approach for transferring part-of-speech annotations from a resourced language to a non-resourced but etymologically close language. We first infer a bilingual lexicon between the two languages with methods based on character similarity, frequency similarity and context similarity. We then assign part-of-speech tags to these bilingual lexicon entries and annotate the remaining words on the basis of suffix analogy. We evaluate our approach on five language pairs of the Iberic peninsula, reaching up to 95% of precision on the lexicon induction task and up to 85% of tagging accuracy.