Even/Disambiguation

From LING073
Revision as of 21:24, 8 April 2023 by Ptrifon1 (talk | contribs)

Jump to: navigation, search

Initial Evaluation of Ambiguity

Number of Tokens in Corpus: 11186

Total Number of Analyses in Corpus: 11186

Ambiguity: 1

Final Evaluation of Ambiguity

Total Number of Analyses in Corpus: 11186

Ambiguity: 1

Example

cayenne[ling073-eve]$ echo "TESTVERB ду" | apertium -d . eve-disam Template:Break "<TESTVERB>" "TESTVERB" v acc "<ду>" "ду" n sg acc

"ду" n sg nom REMOVE
15
"<.>" "." sent

Notes

We ran into substantial problems with building our transducer. Specifically, due to the limited grammar resources and the complexity of the Even language, despite having over 50 morph tests our transducer only actually supports noun cases for two nouns. Thus, in order to disambiguate we added a test verb element to disambiguate between nominative and accusative cases. While this shows that our rule works, it doesn't actually help to disambiguate our corpus. This is particularly the case since our corpus only has text from the Bible which is entirely written in past tense. Due to all of these complicating factors, we have decided that we do not have the time/resources to effectively continue working on Even and will be switching to a different language for the duration of the class.