Log in Help
Print
Homesaleiswc08obi-musingdata 〉 evaluation--thumbs_class.txt
 
Data:

thumbs up/down corpus, manually annotated to add a class feature to
each p annotation.

Experiment:

Train the classifier to mark a "p" annotation with a class, using
n-grams (mostly unigrams) of various features on small, contained
annotations.


All results based on 10-fold cross-validation.


*****  2008-07-21


1-gram on Token.root




1-gram on Token.root + Token.category



1-gram on Token.root + Token.category + Token.orth



1-gram on Token.string + Token.category + Token.orth



1-gram on Token.string + Token.category



1-gram on Token.string



2-gram on Token.string



2-gram on Token.root



2-gram on Token.root + Token.category



2-gram on Token.string + Token.orth



2-gram on Token.root + Token.orth



1-gram on Token.root + Token.orth



1-gram on Token.string + Token.orth



1-gram on Token.category + Token.orth



2-gram on Token.category + Token.orth