Tagging des Stellingen-Korpus mit striktem Mapping


Diese Mappingtabelle erlaubt keine Tags die nicht im Stellingen-Korpus vorkommen

single tagging mit eindeutigem mapping:

accuracy : 95.822029
ambiguity : 1.174173


multi tagging (z=1000):
accuracy : 98.752035
ambiguity : 1.921867

Kommandos:
single:
cat nf-n-200.annotation.tt| ~/dawai/cdg/utils/mapper.pl  -m ~/dawai/cdg/utils/stts_stellingen_map -l /data/linux/opt/tnt/models/negra.tnt -c | ~/Tagger/Tools/negra-statistics.pl -f nf-n-200.annotation.tt

multi:
cat nf-n-200.annotation.tt| ~/dawai/cdg/utils/mapper.pl -z1000 -m ~/dawai/cdg/utils/stts_stellingen_map -l /data/linux/opt/tnt/models/negra.tnt -c | ~/Tagger/Tools/negra-statistics.pl -f nf-n-200.annotation.tt

Interessant:

Der Mapper reduziert die ambiguity um 0.5! Wird das Stellingen-Korpus ohne mapper getaggt (was natürlich wg falschem tagset miserable accuracy ergibt, beträgt die ambiguity 2.56)
Kommando:
cat Stellingen.nf-n-200.tt | /data/linux/opt/tnt/tnt -m -v0 -z1000 /data/linux/opt/tnt/models/negra.tnt - | ~/Tagger/Tools/negra-statistics.pl -f Stellingen.nf-n-200.tt

Definition von accuracy, recall, f-measure, ambiguity

Page Preferences

-- JochenHagenstroem - 20 Mar 2002

Topic attachments
I Attachment Action Size Date Who Comment
mapper.plpl mapper.pl manage 7.7 K 20 Mar 2002 - 15:59 UnknownUser  
nf-n-200.annotation.tttt nf-n-200.annotation.tt manage 22.3 K 20 Mar 2002 - 14:59 UnknownUser  
statistics.plpl statistics.pl manage 3.7 K 17 Nov 2002 - 14:50 UnknownUser  
stts_stellingen_mapEXT stts_stellingen_map manage 2.9 K 20 Mar 2002 - 12:50 UnknownUser  
This topic: User > JochenHagenstroem > DiplomJochenHagenstroem > TAnfn200_2
Topic revision: 17 Oct 2012, UnknownUser
 
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback