Error: A new ambiguity class was found

From Apertium
Jump to: navigation, search
See also: Unsupervised tagger training


(in this example, I use eng as language resp. eng-deu as pair)

If you see an error message like this when executing unsupervised.make:

Error: A new ambiguity class was found. I cannot continue.
Word 'one' not found in the dictionary.
New ambiguity class: {NUM,PRN}

the file ./eng-tagger-data/eng.dic for some reasons is empty (has a file size of 0).

This file contains the words that have multible parts of speech, the error message "Word not found in the dictionary" means, the word was not found in this file.


You can manually create this file by executing the following command:

lt-proc -a eng-deu.automorf.bin < eng-tagger-data/eng.dic.expanded |\
  apertium-filter-ambiguity apertium-eng-deu.eng.tsx > eng-tagger-data/eng.dic

(normally execute by eng-deu-unsupervised.make)


than you can coutinue the generation of the prob file with:

apertium-tagger -t 8 eng-tagger-data/eng.dic eng-tagger-data/eng.crp apertium-eng-deu.eng.tsx  eng-deu.prob
Personal tools