Changes between Version 7 and Version 8 of sk
- Timestamp:
- Jun 13, 2013, 11:09:55 PM (11 years ago)
Legend:
- Unmodified
- Added
- Removed
- Modified
-
sk
v7 v8 2 2 3 3 4 == Morphological analysis for slovak ==4 == Morphological analysis for Slovak == 5 5 6 6 For Slovak there is only one morphological tagger called '''MORČE'''. '''MORČE''' is Czech morphological tagger based on Averaged Perceptron developed in Prague, Czech Republic in 2007. It was trained on Slovak manually annotated corpus '''r-mak''' and now use for Slovak morphological analysis. … … 53 53 [[BR]] 54 54 55 {{{ 56 Feature Accuracy 57 kind 98.10 % 58 genus 93.87 % 59 number 98.76 % 60 case 93.32 % 61 person 96.67 % 62 mod 99.93 % 63 whole tag 92.31 % 64 }}} 65 [[BR]] 66 67 In this evaluation we don't use any parameters. As a~input for training we use 80% of '''r-mak 3.0''' and we annotate the rest 20% of corpus. Then we determine accuracy between original 20% part of '''r-mak 3.0''' and annotated by '''RFTagger'''. 55 68 56 69 {{{ 57 70 Feature Accuracy 58 kind 98. 16%59 genus 9 4.01 %60 number 9 8.78%61 case 9 3.49%62 person 9 6.85%71 kind 98.02 % 72 genus 95.81 % 73 number 99.24 % 74 case 95.42 % 75 person 98.53 % 63 76 mod 99.92 % 64 whole tag 89.55%77 whole tag 94.10 % 65 78 }}} 66 79 [[BR]] 67 80 68 81 69 We do not use any parameters. As a~input for training we use 90% of '''r-mak 3.0''' and we annotate the rest 10% of corpus. Then we determine accuracy between original 10% part of '''r-mak 3.0''' and annotated by '''RFTagger'''.82 We do use -o POSTag -c 8 -l lexicon parameters. As a~input for training we use 80% of '''r-mak 3.0''' and we annotate the rest 20% of corpus. Then we determine accuracy between original 20% part of '''r-mak 3.0''' and annotated by '''RFTagger'''. 70 83 71 84 … … 78 91 person 98.53 % 79 92 mod 99.898 % 80 whole tag 9 2.42%93 whole tag 94.016 % 81 94 82 95 }}}