El dc 29 de 02 de 2012 a les 17:12 -0500, en/na Hector va escriure: > Hi all, > just ran into the following example: > > $ echo "The priest drinks wine." | lt-proc en-es.automorf.bin | > apertium-tagger -pg en-es.prob > > ^The/The<det><def><sp>$ ^priest/priest<n><sg>$ ^drinks/drink<n><pl>$ > ^wine/wine<n><sg>$^./.<sent>$ > > notice that "drinks" is mislabeled by the part-of-speech tagger. Could > you please point me to how to fix these errors?
Yes, our part-of-speech tagger for English is pretty dire, because it was trained with unsupervised training: http://wiki.apertium.org/wiki/Unsupervised_tagger_training You could try retraining it (probably it won't help), or you could try writing some Constraint Grammar rules to improve the tagging performance: http://wiki.apertium.org/wiki/Apertium_and_Constraint_Grammar In this sentence the rule could be: # If there is an ambiguity between present tense, 3rd person singular # verb and plural noun, then choose the verb if both the preceeding # word and the following word are unambiguously singular nouns. SELECT (vblex pri p3 sg) IF (0 (vblex pri p3 sg)) (0 (n pl)) (-1C (n sg)) (1C (n sg)); ## The priest drinks wine. Regards, Fran ------------------------------------------------------------------------------ Virtualization & Cloud Management Using Capacity Planning Cloud computing makes use of virtualization - but cloud computing also focuses on allowing computing to be delivered as a service. http://www.accelacomm.com/jaw/sfnl/114/51521223/ _______________________________________________ Apertium-stuff mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/apertium-stuff
