Spanish resources for CSTlemma
------------------------------

There are two flex rule files. The biggest one, 'flexrules-no-dict-needed', lemmatises all training words correctly and a bit over 10% of OOV words wrong. The smaller one, 'flexrules-supplement-with-dict', which is the result of pruning the larger one, guesses 4% of the training words wrong, but guesses OOV words wrong in just under 10% of the cases.

So for best results use the 'dict' file and the file called 'flexrules-supplement-with-dict' together with the CSTlemma program. See https://github.com/kuhumcst/cstlemma

The dictionary 'dict' and the flex rules were created on the basis of https://github.com/bumshmyak/lachica using the software found here: https://github.com/kuhumcst/affixtrain

Bart Jongejan 2015.11.12