Difference between revisions of "Guarani/Disambiguation"
From LING073
| (3 intermediate revisions by the same user not shown) | |||
| Line 2: | Line 2: | ||
The initial ambiguity of the corpus is 1.31922611850060459492. | The initial ambiguity of the corpus is 1.31922611850060459492. | ||
| + | =Finding ambiguity= | ||
| + | One source of ambiguity are the pronouns. The personal pronouns and the possessive pronouns both take the same form. This can cause ambiguity in sentences such as the following: | ||
| + | * <code>Che pyhare porã</code> (My night is good/well), which analyzes to the following: | ||
| + | ** <code>^'''Che/che<prn><nom><poss><p1><sg>'''/che<prn><nom><pers><p1><sg>$ ^pyhare/pyhare<n><sg>$ ^porã/porã<adj>$^./.<sent>$</code> | ||
| + | |||
| + | In this case, we would want the possessive analysis of ''che'' (meaning "my"), not the personal analysis of ''che'' (meaning "I"). | ||
| + | |||
| + | * <code>Che aguata</code> (I walk), which analyzes to the following: | ||
| + | ** <code>^Che/che<prn><nom><poss><p1><sg>/'''che<prn><nom><pers><p1><sg>$''' ^aguata/guata<v><ar><pres><p1><sg>$^./.<sent>$</code> | ||
| + | |||
| + | In this case, we would want the personal analysis of ''che'', not the possessive analysis. | ||
| + | |||
| + | =Final Ambiguity Evaluation= | ||
| + | After adding the constraint rule and adding several words to the lexc file, the ambiguity became 1.32790988735919899875. I think there was only a small decrease in ambiguity because there weren't too many examples of this ambiguous case in the corpus. | ||
[[Category:Sp17_Disambiguation]] | [[Category:Sp17_Disambiguation]] | ||
Latest revision as of 01:19, 3 March 2017
Initial ambiguity evaluation
The initial ambiguity of the corpus is 1.31922611850060459492.
Finding ambiguity
One source of ambiguity are the pronouns. The personal pronouns and the possessive pronouns both take the same form. This can cause ambiguity in sentences such as the following:
-
Che pyhare porã(My night is good/well), which analyzes to the following:-
^Che/che<prn><nom><poss><p1><sg>/che<prn><nom><pers><p1><sg>$ ^pyhare/pyhare<n><sg>$ ^porã/porã<adj>$^./.<sent>$
-
In this case, we would want the possessive analysis of che (meaning "my"), not the personal analysis of che (meaning "I").
-
Che aguata(I walk), which analyzes to the following:-
^Che/che<prn><nom><poss><p1><sg>/che<prn><nom><pers><p1><sg>$ ^aguata/guata<v><ar><pres><p1><sg>$^./.<sent>$
-
In this case, we would want the personal analysis of che, not the possessive analysis.
Final Ambiguity Evaluation
After adding the constraint rule and adding several words to the lexc file, the ambiguity became 1.32790988735919899875. I think there was only a small decrease in ambiguity because there weren't too many examples of this ambiguous case in the corpus.
