[ Skip to the content ]

Institute of Formal and Applied Linguistics Wiki

• ÚFAL Wiki Home
• ÚFAL Home
• Sitemap
• RSS
• ÚFAL calendar
• Recent changes
• Login
• Hledat:

[ Back to the navigation ]

Differences

This shows you the differences between two versions of the page.

View differences: Side by SideInline Go

Link to this comparison view

Both sides previous revision Previous revision 2012/12/15 13:15 zeman 2012/10/02 21:10 zeman Zarovnání textu v tabulce.2012/10/02 16:45 zeman HPST 2012 sample.2012/10/02 16:38 zeman HPST 2012 sample.2012/10/02 16:33 zeman HPST 2012 sample.2012/10/02 16:25 zeman HPST 2012 sample.2012/10/02 16:13 zeman HTB 0.5 for the 2012 shared task.2011/12/08 08:38 zeman Zarovnání čísel v tabulkách.2011/12/06 22:46 zeman Parsing results.2011/12/06 22:35 zeman Inside.2011/12/06 22:23 zeman Size.2011/12/06 17:58 zeman Hindi test data sample.2011/12/06 17:39 zeman Hindi development data sample.2011/12/06 17:12 zeman CoNLL sample corrected.2011/12/06 16:51 zeman Sample training CoNLL.2011/12/06 16:32 zeman Sample training Shakti.2011/12/06 16:24 zeman vytvořeno Go		2012/12/15 13:15 zeman 2012/10/02 21:10 zeman Zarovnání textu v tabulce.2012/10/02 16:45 zeman HPST 2012 sample.2012/10/02 16:38 zeman HPST 2012 sample.2012/10/02 16:33 zeman HPST 2012 sample.2012/10/02 16:25 zeman HPST 2012 sample.2012/10/02 16:13 zeman HTB 0.5 for the 2012 shared task.2011/12/08 08:38 zeman Zarovnání čísel v tabulkách.2011/12/06 22:46 zeman Parsing results.2011/12/06 22:35 zeman Inside.2011/12/06 22:23 zeman Size.2011/12/06 17:58 zeman Hindi test data sample.2011/12/06 17:39 zeman Hindi development data sample.2011/12/06 17:12 zeman CoNLL sample corrected.2011/12/06 16:51 zeman Sample training CoNLL.2011/12/06 16:32 zeman Sample training Shakti.2011/12/06 16:24 zeman vytvořeno Go Next revision Both sides next revision
user:zeman:treebanks:hi [2011/12/06 17:58] zeman Hindi test data sample.		user:zeman:treebanks:hi [2011/12/06 22:23] zeman Size.
Line 43:		Line 43:
	==== Size ====		==== Size ====
-	HyDT-Bangla shows dependencies between chunks, not words. The node/tree ratio is thus much lower than in other treebanks. The ICON 2009 version came with a data split into three parts: training, development and test:	+	HyDT-Hindi contains dependencies on two levels: between chunks and inside chunks. The ICON 2009 CoNLL-formatted version contained only dependencies between chunks, thus the node/tree ratio was much lower than in other treebanks. The ICON 2009 version came with a data split into three parts: training, development and test:
	^ Part ^ Sentences ^ Chunks ^ Ratio ^		^ Part ^ Sentences ^ Chunks ^ Ratio ^
-	| Training | 980 | 6449 | 6.58 |	+	| Training | 1501 | 13779 | 9.18 |
-	| Development | 150 | 811 | 5.41 |	+	| Development | 150 | 1250 | 8.33 |
-	| Test | 150 | 961 | 6.41 |	+	| Test | 150 | 1156 | 7.71 |
-	| TOTAL | 1280 | 8221 | 6.42 |	+	| TOTAL | 1801 | 16185 | 8.99 |
-	The ICON 2010 version came with a data split into three parts: training, development and test:	+	The ICON 2010 version came with a data split into three parts: training, development and test. The intra-chunk dependencies have been added:
	^ Part ^ Sentences ^ Chunks ^ Ratio ^ Words ^ Ratio ^		^ Part ^ Sentences ^ Chunks ^ Ratio ^ Words ^ Ratio ^
-	| Training | 979 | 6440 | 6.58 | 10305 | 10.52 |	+	| Training | 2972 | | | 64452 | 21.69 |
-	| Development | 150 | 812 | 5.41 | 1196 | 7.97 |	+	| Development | 543 | | | 12616 | 23.23 |
-	| Test | 150 | 961 | 6.41 | 1350 | 9.00 |	+	| Test | 321 | | | 6588 | 20.52 |
-	| TOTAL | 1279 | 8213 | 6.42 | 12851 | 10.04 |	+	| TOTAL | 3836 | | | 83656 | 21.81 |
-	I have counted the sentences and chunks. The number of words comes from (Husain et al., 2010). Note that the paper gives the number of training sentences as 980 (instead of 979), which is a mistake. The last training sentence has the id 980 but there is no sentence with id 418.	+	I have counted the sentences and tokens (words) on the ''.conll'' files; there are slight differences from the statistics presented in (Husain et al., 2010).
-		+
-	Apparently the training-development-test data split was more or less identical in both years, except for the minor discrepancies (number of training sentences and development chunks).	+
	==== Inside ====		==== Inside ====

---

[ Back to the navigation ] [ Back to the content ]