Morphological prediction for Polish by a statistical a tergo index
Maciej Piasecki, Adam Radziszewski · Systems Science · 2008
We present a direct method to construct a morpho-syntactic guesser for Polish. Such guesser produces morpho-syntactic descriptions for word forms unknown to the morphological analyser. The method relies upon a statistical a tergo index, in which pseudo-suffixes (endings) extracted from a statistical tree define morphosyntactic properties of corresponding word forms. The secondary aim is to investigate to what extent it is possible to develop the morphological analysis exclusively on the basis of endings. A statistically extracted a tergo index of Polish word forms was created. Various experiments giving insights into the properties of the index are presented. The method seems to be easily applicable to any other inflectional language with only minor technical changes.