user:zeman:treebanks [ufal wiki]

Original PADT 1.0 as distributed by the LDC
CoNLL 2006
CoNLL 2007

The CoNLL 2007 version reportedly improves over CoNLL 2006 in quality of morphological annotation. Both CoNLL versions miss important parts of the original PADT annotation (see below).

The original PADT 1.0 is distributed by the LDC under the catalogue number LDC2004T23. It is free for LDC members 2004, price for non-members is unknown (contact LDC). The license in short:

non-commercial research usage
no redistribution
cite one publication in publications; also include: “The PADT 1.0 has been developed by the Institute of Formal and Applied Linguistics and the Center for Computational Linguistics (see http://ufal.mff.cuni.cz/).”

The CoNLL 2006 and 2007 versions are obtainable upon request under similar license terms. Their publication in the LDC together with the other CoNLL treebanks is being prepared.

PADT was created by members of the Institute of Formal and Applied Linguistics (Ústav formální a aplikované lingvistiky, ÚFAL), Faculty of Mathematics and Physics (Matematicko-fyzikální fakulta), Charles University in Prague (Univerzita Karlova v Praze), Malostranské náměstí 25, Praha, CZ-11800, Czechia. The CoNLL conversion of the treebank was prepared by Otakar Smrž, one of the key authors of the original PADT.

Website
- http://ufal.mff.cuni.cz/padt/
Data
- Jan Hajič, Otakar Smrž, Petr Zemánek, Petr Pajas, Jan Šnaidauf, Emanuel Beška, Jakub Kráčmar, Kamila Hassanová: Prague Arabic Dependency Treebank 1.0 (LDC2004T23). Linguistic Data Consortium, Philadelphia, USA, 2004. ISBN 1-58563-319-4.
Principal publications
- Jan Hajič, Otakar Smrž, Petr Zemánek, Jan Šnaidauf, Emanuel Beška: Prague Arabic Dependency Treebank: Development in Data and Tools. In: Proceedings of the NEMLAR International Conference on Arabic Language Resources and Tools, pages 110-117, Cairo, Egypt, 2004.

Newswire text from press agencies (Agence France Presse, Ummah, Al Hayat, An Nahar, Xinhua 2001-2003).

According to their website, the original PADT 1.0 contains 113,500 tokens annotated analytically. The CoNLL 2006 version contains 59752 tokens in 1606 sentences, yielding 37.21 tokens per sentence on average (CoNLL 2006 data split: 54379 tokens / 1460 sentences training, 5373 tokens / 146 sentences test). The CoNLL 2007 version contains 116,793 tokens in 3043 sentences, yielding 38.38 tokens per sentence on average (CoNLL 2007 data split: 111,669 tokens / 2912 sentences training, 5124 tokens / 131 sentences test).

As noted in (Nivre et al., 2007), “the parsing units in this treebank are in many cases larger than conventional sentences, which partly explains the high average number of tokens per sentence.”

The original PADT 1.0 is distributed in the FS format. The CoNLL versions are distributed in the CoNLL-X format. The original PADT contains more information than the CoNLL version. There is morphological annotation (tags and lemmas) both manual and by a tagger (while only manual is in CoNLL data), glosses etc. However, the most important piece of information that got lost during the conversion to CoNLL is the FS attribute called parallel. It distinguishes conjuncts from shared modifiers of coordination and thus the syntactic structure is incomplete without it.

Word forms and lemmas are vocalized, i.e. they contain diacritics for short vowels as well as consonant gemination and a few other things. The CoNLL 2006 version includes Buckwalter transliteration of the Arabic script (in the same column as Arabic, attached to the Arabic form/lemma with an underscore character).

Note that tokenization of Arabic typically includes splitting original words (inserting spaces between letters), not just separating punctuation from words. Example: وبالفالوجة = wabiālfālūjah = wa/CONJ + bi/PREP + AlfAlwjp/NOUN_PROP = and in al-Falujah. In PADT, conjunctions and prepositions are separate tokens and nodes.

The original PADT 1.0 uses 10-character positional morphological tags whose documentation is hard to find. The CoNLL 2006 version converts the tags to the three CoNLL columns, CPOS, POS and FEAT, most of the information being encoded as pipe-separated attribute-value assignments in FEAT. There should be a 1-1 mapping between the PADT positional tags and the CoNLL 2006 annotation. The CoNLL 2007 version uses a tag conversion different from CoNLL 2006. Both CoNLL distributions contain a README file with a brief description of the parts of speech and features. Use DZ Interset to inspect the two CoNLL tagsets. Also look at the Elixir FM online interface for a later development of the morphological analyzer created along with PADT.

The guidelines for syntactic annotation are documented in the PADT annotation manual (only peculiarities of Arabic are documented, otherwise it is referenced to the annotation manual for the Czech treebank). The list and brief description of syntactic tags (dependency relation labels) can be found in (Smrž, Šnaidauf and Zemánek, 2002).

The first two sentences of the CoNLL 2006 training data:

1	غِيابُ_giyAbu	غِياب_giyAb	N	N	case=1\|def=R	0	ExD	_	_
2	فُؤاد_fu&Ad	فُؤاد_fu&Ad	Z	Z	_	3	Atr	_	_
3	كَنْعان_kanoEAn	كَنْعان_kanoEAn	Z	Z	_	1	Atr	_	_

1	فُؤاد_fu&Ad	فُؤاد_fu&Ad	Z	Z	_	2	Atr	_	_
2	كَنْعان_kanoEAn	كَنْعان_kanoEAn	Z	Z	_	9	Sb	_	_
3	،_,	،_,	G	G	_	2	AuxG	_	_
4	رائِد_rA}id	رائِد_rA}id	N	N	_	2	Atr	_	_
5	القِصَّة_AlqiS~ap	قِصَّة_qiS~ap	N	N	gen=F\|num=S\|def=D	4	Atr	_	_
6	القَصِيرَةِ_AlqaSiyrapi	قَصِير_qaSiyr	A	A	gen=F\|num=S\|case=2\|def=D	5	Atr	_	_
7	فِي_fiy	فِي_fiy	P	P	_	4	AuxP	_	_
8	لُبْنانِ_lubonAni	لُبْنان_lubonAn	Z	Z	case=2\|def=R	7	Atr	_	_
9	رَحَلَ_raHala	رَحَل-َ_raHal-a	V	VP	pers=3\|gen=M\|num=S	0	Pred	_	_
10	مَساءَ_masA'a	مَساء_masA'	D	D	_	9	Adv	_	_
11	أَمْسِ_>amosi	أَمْسِ_>amosi	D	D	_	10	Atr	_	_
12	عَن_Ean	عَن_Ean	P	P	_	9	AuxP	_	_
13	81_81	81_81	Q	Q	_	12	Adv	_	_
14	عاماً_EAmAF	عام_EAm	N	N	gen=M\|num=S\|case=4\|def=I	13	Atr	_	_
15	._.	._.	G	G	_	0	AuxK	_	_

The first sentence of the CoNLL 2006 test data:

1	اِتِّفاقٌ_Ait~ifAqN	اِتِّفاق_Ait~ifAq	N	N	case=1\|def=I	0	ExD	_	_
2	بَيْنَ_bayona	بَيْنَ_bayona	P	P	_	1	AuxP	_	_
3	لُبْنانِ_lubonAni	لُبْنان_lubonAn	Z	Z	case=2\|def=R	4	Atr	_	_
4	وَ_wa	وَ_wa	C	C	_	2	Coord	_	_
5	سُورِيَّةٍ_suwriy~apK	سُورِيا_suwriyA	Z	Z	gen=F\|num=S\|case=2\|def=I	4	Atr	_	_
6	عَلَى_EalaY	عَلَى_EalaY	P	P	_	1	AuxP	_	_
7	رَفْعِ_rafoEi	رَفْع_rafoE	N	N	case=2\|def=R	6	Atr	_	_
8	مُسْتَوَى_musotawaY	مُسْتَوَى_musotawaY	N	N	_	7	Atr	_	_
9	التَبادُلِ_AltabAduli	تَبادُل_tabAdul	N	N	case=2\|def=D	8	Atr	_	_
10	التِجارِيِّ_AltijAriy~i	تِجارِيّ_tijAriy~	A	A	case=2\|def=D	9	Atr	_	_
11	إِلَى_<ilaY	إِلَى_<ilaY	P	P	_	7	AuxP	_	_
12	500_500	500_500	Q	Q	_	11	Atr	_	_
13	مِلْيُونِ_miloyuwni	مِلْيُون_miloyuwn	N	N	case=2\|def=R	12	Atr	_	_
14	دُولارٍ_duwlArK	دُولار_duwlAr	N	N	case=2\|def=I	13	Atr	_	_

The first sentence of the CoNLL 2007 training data:

1	تَعْدادُ	تَعْداد_1	N	N-	Case=1\|Defin=R	7	Sb	_	_
2	سُكّانِ	ساكِن_1	N	N-	Case=2\|Defin=R	1	Atr	_	_
3	22	[DEFAULT]	Q	Q-	_	2	Atr	_	_
4	دَوْلَةً	دَوْلَة_1	N	N-	Gender=F\|Number=S\|Case=4\|Defin=I	3	Atr	_	_
5	عَرَبِيَّةً	عَرَبِيّ_1	A	A-	Gender=F\|Number=S\|Case=4\|Defin=I	4	Atr	_	_
6	سَ	سَ_FUT	F	F-	_	7	AuxM	_	_
7	يَرْتَفِعُ	اِرْتَفَع_1	V	VI	Mood=I\|Voice=A\|Person=3\|Gender=M\|Number=S	0	Pred	_	_
8	إِلَى	إِلَى_1	P	P-	_	7	AuxP	_	_
9	654	[DEFAULT]	Q	Q-	_	8	Adv	_	_
10	مِلْيُونَ	مِلْيُون_1	N	N-	Case=4\|Defin=R	9	Atr	_	_
11	نَسَمَةٍ	نَسَمَة_1	N	N-	Gender=F\|Number=S\|Case=2\|Defin=I	10	Atr	_	_
12	فِي	فِي_1	P	P-	_	7	AuxP	_	_
13	مُنْتَصَفِ	مُنْتَصَف_1	N	N-	Case=2\|Defin=R	12	Adv	_	_
14	القَرْنِ	قَرْن_1	N	N-	Case=2\|Defin=D	13	Atr	_	_

The first sentence of the CoNLL 2007 test data:

1	مُقاوَمَةُ	مُقاوَمَة_1	N	N-	Gender=F\|Number=S\|Case=1\|Defin=R	0	ExD	_	_
2	زَواجِ	زَواج_1	N	N-	Case=2\|Defin=R	1	Atr	_	_
3	الطُلّابِ	طالِب_1	N	N-	Case=2\|Defin=D	2	Atr	_	_
4	العُرْفِيِّ	عُرْفِيّ_1	A	A-	Case=2\|Defin=D	2	Atr	_	_

Nonprojectivities in PADT are rare. Only 431 of the 116,793 tokens in the CoNLL 2007 version are attached nonprojectively (0.37%).

The results of the CoNLL 2006 shared task are available online. They have been published in (Buchholz and Marsi, 2006). The evaluation procedure was non-standard because it excluded punctuation tokens. These are the best results for Arabic:

Parser (Authors)	LAS	UAS
MST (McDonald et al.)	66.91	79.34
Basis (O'Neil)	66.71	78.54
Malt (Nivre et al.)	66.71	77.52
Edinburgh (Riedel et al.)	66.65	78.62

The results of the CoNLL 2007 shared task are available online. They have been published in (Nivre et al., 2007). The evaluation procedure was changed to include punctuation tokens. These are the best results for Arabic:

Parser (Authors)	LAS	UAS
Malt (Nilsson et al.)	76.52	85.81
Nakagawa	75.08	86.09
Malt (Hall et al.)	74.75	84.21
Sagae	74.71	84.04
Chen	74.65	83.49
Titov et al.	74.12	83.18

The two Malt parser results of 2007 (single malt and blended) are described in (Hall et al., 2007) and the details about the parser configuration are described here.

Original BTB in native format
CoNLL 2006 (BulTreeBank-DP)

The original BTB is based on HPSG (head-driven phrase-structure grammar). The CoNLL version contains only the dependency information encoded in HPSG BulTreeBank.

Only the CoNLL version seems to be distributed but you may ask the creators about the HPSG version. For the dependency version, print the license, sign, scan, send to Kiril Simov (kivs (at) bultreebank (dot) org) and wait for the data. The license in short:

research usage
no redistribution
cite one publication in publications

BTB was created by members of the Linguistic Modelling Department (Секция Лингвистично моделиране), Bulgarian Academy of Sciences (Българска академия на науките), Ул. Акад. Г. Бончев, Бл. 25 А, 1113 София, Bulgaria.

Website
- http://www.bultreebank.org/indexBTB.html
Data
- no separate citation
Principal publications
- Kiril Simov, Petya Osenova, Alexander Simov, Milen Kouylekov: Design and Implementation of the Bulgarian HPSG-based Treebank. In: Erhard Hinrichs, Kiril Simov (eds.): Journal of Research on Language and Computation, Special Issue, vol. 2, no. 4, pp. 495–522, Kluwer Academic Publishers, ISSN 1570-7075. 2004.
Documentation
- Kiril Simov, Petya Osenova, Milena Slavcheva: BTB-TR03: BulTreeBank Morphosyntactic Tagset. Technical report, 2004.
- Petya Osenova, Kiril Simov: BTB-TR05: BulTreeBank Stylebook. Technical report, 2004.
- http://www.bultreebank.org/dpbtb/ provides the list of dependency relation labels (s-tags) with brief description.

Unknown (“A set of Bulgarian sentences marked-up with detailed syntactic information. These sentences are mainly extracted from authentic Bulgarian texts. They are chosen with regards two criteria. First, they cover the variety of syntactic structures of Bulgarian. Second, they show the statistical distribution of these phenomena in real texts.”) At least part of it is probably news (Novinar, Sega, Standart).

The CoNLL 2006 version contains 196,151 tokens in 13221 sentences, yielding 14.84 tokens per sentence on average (CoNLL 2006 data split: 190,217 tokens / 12823 sentences training, 5934 tokens / 398 sentences test).

The original morphosyntactic tags have been converted to fit into the three columns (CPOS, POS and FEAT) of the CoNLL format. There should be a 1-1 mapping between the BTB positional tags and the CoNLL 2006 annotation. Use DZ Interset to inspect the CoNLL tagset.

The morphological analysis does not include lemmas. The morphosyntactic tags have been assigned (probably) manually.

The guidelines for syntactic annotation are documented in the other technical report. The CoNLL distribution contains the BulTreeBankReadMe.html file with a brief description of the syntactic tags (dependency relation labels).

The first three sentences of the CoNLL 2006 training data:

1	Глава	_	N	Nc	_	0	ROOT	0	ROOT
2	трета	_	M	Mo	gen=f\|num=s\|def=i	1	mod	1	mod

1	НАРОДНО	_	A	An	gen=n\|num=s\|def=i	2	mod	2	mod
2	СЪБРАНИЕ	_	N	Nc	gen=n\|num=s\|def=i	0	ROOT	0	ROOT

1	Народното	_	A	An	gen=n\|num=s\|def=d	2	mod	2	mod
2	събрание	_	N	Nc	gen=n\|num=s\|def=i	3	subj	3	subj
3	осъществява	_	V	Vpi	trans=t\|mood=i\|tense=r\|pers=3\|num=s	0	ROOT	0	ROOT
4	законодателната	_	A	Af	gen=f\|num=s\|def=d	5	mod	5	mod
5	власт	_	N	Nc	_	3	obj	3	obj
6	и	_	C	Cp	_	3	conj	3	conj
7	упражнява	_	V	Vpi	trans=t\|mood=i\|tense=r\|pers=3\|num=s	3	conjarg	3	conjarg
8	парламентарен	_	A	Am	gen=m\|num=s\|def=i	9	mod	9	mod
9	контрол	_	N	Nc	gen=m\|num=s\|def=i	7	obj	7	obj
10	.	_	Punct	Punct	_	3	punct	3	punct

The first three sentences of the CoNLL 2006 test data:

1	Единственото	_	A	An	gen=n\|num=s\|def=d	2	mod	2	mod
2	решение	_	N	Nc	gen=n\|num=s\|def=i	0	ROOT	0	ROOT

1	Ерик	_	N	Np	gen=m\|num=s\|def=i	0	ROOT	0	ROOT
2	Франк	_	N	Np	gen=m\|num=s\|def=i	1	mod	1	mod
3	Ръсел	_	H	Hm	gen=m\|num=s\|def=i	2	mod	2	mod

1	Пълен	_	A	Am	gen=m\|num=s\|def=i	2	mod	2	mod
2	мрак	_	N	Nc	gen=m\|num=s\|def=i	0	ROOT	0	ROOT
3	и	_	C	Cp	_	2	conj	2	conj
4	пълна	_	A	Af	gen=f\|num=s\|def=i	5	mod	5	mod
5	самота	_	N	Nc	_	2	conjarg	2	conjarg
6	.	_	Punct	Punct	_	2	punct	2	punct

Nonprojectivities in BTB are rare. Only 747 of the 196,151 tokens in the CoNLL 2006 version are attached nonprojectively (0.38%).

The results of the CoNLL 2006 shared task are available online. They have been published in (Buchholz and Marsi, 2006). The evaluation procedure was non-standard because it excluded punctuation tokens. These are the best results for Bulgarian:

Parser (Authors)	LAS	UAS
MST (McDonald et al.)	87.57	92.04
Malt (Nivre et al.)	87.41	91.72
Nara (Yuchang Cheng)	86.34	91.30

ICON 2009
- Shakti Standard Format (SSF; native)
- CoNLL format
ICON 2010
- Shakti Standard Format (SSF; native)
- CoNLL format

There has been no official release of the treebank yet. There have been two as-is sample releases for the purposes of the NLP tools contests in parsing Indian languages, attached to the ICON 2009 and 2010 conferences.

There is no standard distribution channel for the treebank after the ICON 2010 evaluation period. Inquire at the LTRC (ltrc (at) iiit (dot) ac (dot) in) about the possibility of getting the data. The ICON 2010 license in short:

non-commercial research usage
no redistribution

HyDT-Bangla is being created by members of the Language Technologies Research Centre, International Institute of Information Technology, Gachibowli, Hyderabad, 500032, India.

Website
- no website dedicated to the treebank
- ICON 2009 NLP Tools Contest On-Line Proceedings
- ICON 2010 NLP Tools Contest On-Line Proceedings
Data
- no separate citation
Principal publications
- Samar Husain: Dependency Parsers for Indian Languages. In: Proceedings of ICON-2009 NLP Tools Contest, Hyderabad, India, 2009.
- Samar Husain, Prashanth Mannem, Bharat Ambati, Phani Gadde: The ICON-2010 tools contest on Indian language dependency parsing. In: Proceedings of ICON-2010 NLP Tools Contest, pp. 1–8, Kharagpur, India, 2010.
Documentation
- http://ltrc.iiit.ac.in/nlptools2010/documentation.php

Unknown.

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:

Part	Sentences	Chunks	Ratio
Training	980	6449	6.58
Development	150	811	5.41
Test	150	961	6.41
TOTAL	1280	8221	6.42

The ICON 2010 version came with a data split into three parts: training, development and test:

Part	Sentences	Chunks	Ratio	Words	Ratio
Training	979	6440	6.58	10305	10.52
Development	150	812	5.41	1196	7.97
Test	150	961	6.41	1350	9.00
TOTAL	1279	8213	6.42	12851	10.04

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.

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).

The text uses the WX encoding of Indian letters. If we know what the original script is (Bengali in this case) we can map the WX encoding to the original characters in UTF-8. WX uses English letters so if there was embedded English (or other string using Latin letters) it will probably get lost during the conversion.

The CoNLL format contains only the chunk heads. The native SSF format shows the other words in the chunk, too, but it does not capture intra-chunk dependency relations. This is an example of a multi-word chunk:

3       ((      NP      <fs af='rumAla,n,,sg,,d,0,0' head="rumAla" drel=k2:VGF name=NP3>
3.1     ekatA   QC      <fs af='eka,num,,,,,,'>
3.2     ledisa  JJ      <fs af='ledisa,unk,,,,,,'>
3.3     rumAla  NN      <fs af='rumAla,n,,sg,,d,0,0' name="rumAla">
        ))

In the CoNLL format, the CPOS column contains the chunk label (e.g. NP = noun phrase) and the POS column contains the part of speech of the chunk head.

Occasionally there are NULL nodes that do not correspond to any surface chunk or token. They represent ellided participants.

The syntactic tags (dependency relation labels) are karaka relations, i.e. deep syntactic roles according to the Pāṇinian grammar. There are separate versions of the treebank with fine-grained and coarse-grained syntactic tags.

According to (Husain et al., 2010), in the ICON 2010 version, the chunk tags, POS tags and inter-chunk dependencies (topology + tags) were annotated manually. The rest (lemma, morphosyntactic features, headword of chunk) was marked automatically.

Note: There have been cycles in the Hindi part of HyDT but no such problem occurs in the Bengali part.

The first sentence of the ICON 2010 training data (with fine-grained syntactic tags) in the Shakti format:

<document id="">			
<head>
<annotated-resource name="HyDT-Bangla" version="0.5" type="dep-interchunk-only" layers="morph,pos,chunk,dep-interchunk-only" language="ben" date-of-release="20100831">
    <annotation-standard>
        <morph-standard name="Anncorra-morph" version="1.31" date="20080920" />
        <pos-standard name="Anncorra-pos" version="" date="20061215" />
        <chunk-standard name="Anncorra-chunk" version="" date="20061215" />
        <dependency-standard name="Anncorra-dep" version="2.0" date="" dep-tagset-granularity="6" />
    </annotation-standard>
</annotated-resource>		
</head>			
<Sentence id="1">
1	((	NP	<fs af='Age,adv,,,,,,' head="Agei" drel=k7t:VGF name=NP>
1.1	mudZira	NN	<fs af='mudZi,n,,sg,,o,era,era'>
1.2	Agei	NST	<fs af='Age,adv,,,,,,' name="Agei">
	))		
2	((	NP	<fs af='cA,n,,sg,,d,0,0' head="cA" drel=k1:VGF name=NP2>
2.1	praWama	QO	<fs af='praWama,num,,,,,,'>
2.2	kApa	NN	<fs af='kApa,unk,,,,,,'>
2.3	cA	NN	<fs af='cA,n,,sg,,d,0,0' name="cA">
	))		
3	((	VGF	<fs af='As,v,,,5,,A_yA+Ce,A' head="ese" name=VGF>
3.1	ese	VM	<fs af='As,v,,,7,,A,A' name="ese">
3.2	.	SYM	<fs af='.,punc,,,,,,'>
	))		
</Sentence>

And in the CoNLL format:

1	Agei	Age	NP	NST	lex-Age\|cat-adv\|gend-\|num-\|pers-\|case-\|vib-\|tam-\|head-Agei\|name-NP	3	k7t	_	_
2	cA	cA	NP	NN	lex-cA\|cat-n\|gend-\|num-sg\|pers-\|case-d\|vib-0\|tam-0\|head-cA\|name-NP2	3	k1	_	_
3	ese	As	VGF	VM	lex-As\|cat-v\|gend-\|num-\|pers-5\|case-\|vib-A_yA+Ce\|tam-A\|head-ese\|name-VGF	0	main	_	_

And after conversion of the WX encoding to the Bengali script in UTF-8:

1	আগেই	আগে	NP	NST	lex-Age\|cat-adv\|gend-\|num-\|pers-\|case-\|vib-\|tam-\|head-Agei\|name-NP	3	k7t	_	_
2	চা	চা	NP	NN	lex-cA\|cat-n\|gend-\|num-sg\|pers-\|case-d\|vib-0\|tam-0\|head-cA\|name-NP2	3	k1	_	_
3	এসে	আস্	VGF	VM	lex-As\|cat-v\|gend-\|num-\|pers-5\|case-\|vib-A_yA+Ce\|tam-A\|head-ese\|name-VGF	0	main	_	_

The first sentence of the ICON 2010 development data (with fine-grained syntactic tags) in the Shakti format:

<document id="">
<head>
<annotated-resource name="HyDT-Bangla" version="0.5" type="dep-interchunk-only" layers="morph,pos,chunk,dep-interchunk-only" language="ben" date-of-release="20100831">
    <annotation-standard>
        <morph-standard name="Anncorra-morph" version="1.31" date="20080920" />
        <pos-standard name="Anncorra-pos" version="" date="20061215" />
        <chunk-standard name="Anncorra-chunk" version="" date="20061215" />
        <dependency-standard name="Anncorra-dep" version="2.0" date="" dep-tagset-granularity="6" />
    </annotation-standard>
</annotated-resource>
</head>
<Sentence id="1">
1	((	NP	<fs af='parabarwIkAle,adv,,,,,,' head="parabarwIkAle" drel=k7t:VGF name=NP>
1.1	parabarwIkAle	NN	<fs af='parabarwIkAle,adv,,,,,,' name="parabarwIkAle">
	))		
2	((	NP	<fs af='aPisa-biyArAraxera,unk,,,,,,' head="aPisa-biyArAraxera" drel=r6:NP3 name=NP2>
2.1	aPisa-biyArAraxera	NN	<fs af='aPisa-biyArAraxera,unk,,,,,,' name="aPisa-biyArAraxera">
	))		
3	((	NP	<fs af='nAma,n,,sg,,d,0,0' head="nAma" drel=k2:VGNN name=NP3>
3.1	nAma	NN	<fs af='nAma,n,,sg,,d,0,0' name="nAma">
	))		
4	((	NP	<fs af='GoRaNA,unk,,,,,,' head="GoRaNA" drel=pof:VGNN name=NP4>
4.1	GoRaNA	NN	<fs af='GoRaNA,unk,,,,,,' name="GoRaNA">
	))		
5	((	VGNN	<fs af='kar,n,,,any,,,' head="karAra" drel=r6:NP5 name=VGNN>
5.1	karAra	VM	<fs af='kar,n,,,any,,,' name="karAra">
	))		
6	((	NP	<fs af='samay,unk,,,,,,' head="samay" drel=k7t:VGF name=NP5>
6.1	samay	NN	<fs af='samay,unk,,,,,,' name="samay">
	))		
7	((	NP	<fs af='animeRake,unk,,,,,,' head="animeRake" drel=k2:VGF name=NP6>
7.1	animeRake	NNP	<fs af='animeRake,unk,,,,,,' name="animeRake">
	))		
8	((	VGF	<fs af='sariye,unk,,,5,,0_rAKA+ka_ha+la,' head="sariye" name=VGF>
8.1	sariye	VM	<fs af='sariye,unk,,,,,,' name="sariye">
8.2	.	SYM	<fs af='.,punc,,,,,,'>
	))		
</Sentence>

And in the CoNLL format:

1	parabarwIkAle	parabarwIkAle	NP	NN	lex-parabarwIkAle\|cat-adv\|gend-\|num-\|pers-\|case-\|vib-\|tam-\|head-parabarwIkAle\|name-NP	8	k7t	_	_
2	aPisa-biyArAraxera	aPisa-biyArAraxera	NP	NN	lex-aPisa-biyArAraxera\|cat-unk\|gend-\|num-\|pers-\|case-\|vib-\|tam-\|head-aPisa-biyArAraxera\|name-NP2	3	r6	_	_
3	nAma	nAma	NP	NN	lex-nAma\|cat-n\|gend-\|num-sg\|pers-\|case-d\|vib-0\|tam-0\|head-nAma\|name-NP3	5	k2	_	_
4	GoRaNA	GoRaNA	NP	NN	lex-GoRaNA\|cat-unk\|gend-\|num-\|pers-\|case-\|vib-\|tam-\|head-GoRaNA\|name-NP4	5	pof	_	_
5	karAra	kar	VGNN	VM	lex-kar\|cat-n\|gend-\|num-\|pers-any\|case-\|vib-\|tam-\|head-karAra\|name-VGNN	6	r6	_	_
6	samay	samay	NP	NN	lex-samay\|cat-unk\|gend-\|num-\|pers-\|case-\|vib-\|tam-\|head-samay\|name-NP5	8	k7t	_	_
7	animeRake	animeRake	NP	NNP	lex-animeRake\|cat-unk\|gend-\|num-\|pers-\|case-\|vib-\|tam-\|head-animeRake\|name-NP6	8	k2	_	_
8	sariye	sariye	VGF	VM	lex-sariye\|cat-unk\|gend-\|num-\|pers-5\|case-\|vib-0_rAKA+ka_ha+la\|tam-\|head-sariye\|name-VGF	0	main	_	_

And after conversion of the WX encoding to the Bengali script in UTF-8:

1	পরবর্তীকালে	পরবর্তীকালে	NP	NN	lex-parabarwIkAle\|cat-adv\|gend-\|num-\|pers-\|case-\|vib-\|tam-\|head-parabarwIkAle\|name-NP	8	k7t	_	_
2	অফিস-বিযারারদের	অফিস-বিযারারদের	NP	NN	lex-aPisa-biyArAraxera\|cat-unk\|gend-\|num-\|pers-\|case-\|vib-\|tam-\|head-aPisa-biyArAraxera\|name-NP2	3	r6	_	_
3	নাম	নাম	NP	NN	lex-nAma\|cat-n\|gend-\|num-sg\|pers-\|case-d\|vib-0\|tam-0\|head-nAma\|name-NP3	5	k2	_	_
4	ঘোষণা	ঘোষণা	NP	NN	lex-GoRaNA\|cat-unk\|gend-\|num-\|pers-\|case-\|vib-\|tam-\|head-GoRaNA\|name-NP4	5	pof	_	_
5	করার	কর্	VGNN	VM	lex-kar\|cat-n\|gend-\|num-\|pers-any\|case-\|vib-\|tam-\|head-karAra\|name-VGNN	6	r6	_	_
6	সময্	সময্	NP	NN	lex-samay\|cat-unk\|gend-\|num-\|pers-\|case-\|vib-\|tam-\|head-samay\|name-NP5	8	k7t	_	_
7	অনিমেষকে	অনিমেষকে	NP	NNP	lex-animeRake\|cat-unk\|gend-\|num-\|pers-\|case-\|vib-\|tam-\|head-animeRake\|name-NP6	8	k2	_	_
8	সরিযে	সরিযে	VGF	VM	lex-sariye\|cat-unk\|gend-\|num-\|pers-5\|case-\|vib-0_rAKA+ka_ha+la\|tam-\|head-sariye\|name-VGF	0	main	_	_

The first sentence of the ICON 2010 test data (with fine-grained syntactic tags) in the Shakti format:

<document id="">
<head>
<annotated-resource name="HyDT-Bangla" version="0.5" type="dep-interchunk-only" layers="morph,pos,chunk,dep-interchunk-only" language="ben" date-of-release="20101013">
    <annotation-standard>
        <morph-standard name="Anncorra-morph" version="1.31" date="20080920" />
	<pos-standard name="Anncorra-pos" version="" date="20061215" />
	<chunk-standard name="Anncorra-chunk" version="" date="20061215" />
	<dependency-standard name="Anncorra-dep" version="2.0" date="" dep-tagset-granularity="6" />
    </annotation-standard>
<annotated-resource>
</head>
<Sentence id="1">
1	((	NP	<fs af='mAXabIlawA,n,,sg,,d,0,0' head="mAXabIlawA" drel=k1:VGF name=NP>
1.1	mAXabIlawA	NNP	<fs af='mAXabIlawA,n,,sg,,d,0,0' name="mAXabIlawA">
	))		
2	((	NP	<fs af='waKana,pn,,,,d,0,0' head="waKana" drel=k7t:VGF name=NP2>
2.1	waKana	PRP	<fs af='waKana,pn,,,,d,0,0' name="waKana">
	))		
3	((	NP	<fs af='hAwa,n,,sg,,o,era,era' head="hAwera" drel=r6:NP4 name=NP3>
3.1	hAwera	NN	<fs af='hAwa,n,,sg,,o,era,era' name="hAwera">
	))		
4	((	NP	<fs af='GadZi,unk,,,,,,' head="GadZi" drel=k2:VGNF name=NP4>
4.1	GadZi	NN	<fs af='GadZi,unk,,,,,,' name="GadZi">
	))		
5	((	VGNF	<fs af='Kul,v,,,5,,ne,ne' head="Kule" drel=vmod:VGF name=VGNF>
5.1	Kule	VM	<fs af='Kul,v,,,5,,ne,ne' name="Kule">
	))		
6	((	NP	<fs af='tebila,n,,sg,,d,me,me' head="tebile" drel=k7p:VGF name=NP5>
6.1	tebile	NN	<fs af='tebila,n,,sg,,d,me,me' name="tebile">
	))		
7	((	VGF	<fs af='rAK,v,,,5,,Cila,Cila' head="rAKaCila" name=VGF>
7.1	rAKaCila	VM	<fs af='rAK,v,,,5,,Cila,Cila' name="rAKaCila">
7.2	।	SYM	
	))		
</Sentence>

And in the CoNLL format:

1	mAXabIlawA	mAXabIlawA	NP	NNP	lex-mAXabIlawA\|cat-n\|gend-\|num-sg\|pers-\|case-d\|vib-0\|tam-0\|head-mAXabIlawA\|name-NP	7	k1	_	_
2	waKana	waKana	NP	PRP	lex-waKana\|cat-pn\|gend-\|num-\|pers-\|case-d\|vib-0\|tam-0\|head-waKana\|name-NP2	7	k7t	_	_
3	hAwera	hAwa	NP	NN	lex-hAwa\|cat-n\|gend-\|num-sg\|pers-\|case-o\|vib-era\|tam-era\|head-hAwera\|name-NP3	4	r6	_	_
4	GadZi	GadZi	NP	NN	lex-GadZi\|cat-unk\|gend-\|num-\|pers-\|case-\|vib-\|tam-\|head-GadZi\|name-NP4	5	k2	_	_
5	Kule	Kul	VGNF	VM	lex-Kul\|cat-v\|gend-\|num-\|pers-5\|case-\|vib-ne\|tam-ne\|head-Kule\|name-VGNF	7	vmod	_	_
6	tebile	tebila	NP	NN	lex-tebila\|cat-n\|gend-\|num-sg\|pers-\|case-d\|vib-me\|tam-me\|head-tebile\|name-NP5	7	k7p	_	_
7	rAKaCila	rAK	VGF	VM	lex-rAK\|cat-v\|gend-\|num-\|pers-5\|case-\|vib-Cila\|tam-Cila\|head-rAKaCila\|name-VGF	0	main	_	_

And after conversion of the WX encoding to the Bengali script in UTF-8:

1	মাধবীলতা	মাধবীলতা	NP	NNP	lex-mAXabIlawA\|cat-n\|gend-\|num-sg\|pers-\|case-d\|vib-0\|tam-0\|head-mAXabIlawA\|name-NP	7	k1	_	_
2	তখন	তখন	NP	PRP	lex-waKana\|cat-pn\|gend-\|num-\|pers-\|case-d\|vib-0\|tam-0\|head-waKana\|name-NP2	7	k7t	_	_
3	হাতের	হাত	NP	NN	lex-hAwa\|cat-n\|gend-\|num-sg\|pers-\|case-o\|vib-era\|tam-era\|head-hAwera\|name-NP3	4	r6	_	_
4	ঘড়ি	ঘড়ি	NP	NN	lex-GadZi\|cat-unk\|gend-\|num-\|pers-\|case-\|vib-\|tam-\|head-GadZi\|name-NP4	5	k2	_	_
5	খুলে	খুল্	VGNF	VM	lex-Kul\|cat-v\|gend-\|num-\|pers-5\|case-\|vib-ne\|tam-ne\|head-Kule\|name-VGNF	7	vmod	_	_
6	টেবিলে	টেবিল	NP	NN	lex-tebila\|cat-n\|gend-\|num-sg\|pers-\|case-d\|vib-me\|tam-me\|head-tebile\|name-NP5	7	k7p	_	_
7	রাখছিল	রাখ্	VGF	VM	lex-rAK\|cat-v\|gend-\|num-\|pers-5\|case-\|vib-Cila\|tam-Cila\|head-rAKaCila\|name-VGF	0	main	_	_

Nonprojectivities in HyDT-Bangla are not frequent. Only 78 of the 7252 chunks in the training+development ICON 2010 version are attached nonprojectively (1.08%).

The results of the ICON 2009 NLP tools contest have been published in (Husain, 2009). There were two evaluation rounds, the first with the coarse-grained syntactic tags, the second with the fine-grained syntactic tags. To reward language independence, only systems that parsed all three languages were officially ranked. The following table presents the Bengali/coarse-grained results of the four officially ranked systems, and the best Bengali-only* system.

Parser (Authors)	LAS	UAS
Kolkata (De et al.)*	84.29	90.32
Hyderabad (Ambati et al.)	78.25	90.22
Malt (Nivre)	76.07	88.97
Malt+MST (Zeman)	71.49	86.89
Mannem	70.34	83.56

The results of the ICON 2010 NLP tools contest have been published in (Husain et al., 2010), page 6. These are the best results for Bengali with fine-grained syntactic tags:

Parser (Authors)	LAS	UAS
Attardi et al.	70.66	87.41
Kosaraju et al.	70.55	86.16
Kolachina et al.	70.14	87.10

CoNLL 2007 (CESS-Cat)
CoNLL 2009 (AnCora-CA)

The dependency treebank Cat3LB was extracted automatically from an earlier constituent-based annotation (see Montserrat Civit, Ma. Antònia Martí, Núria Bufí: Cat3LB and Cast3LB: From Constituents to Dependencies. In: T. Salakoski et al. (eds.): FinTAL 2006, LNAI 4139, pp. 141–152, 2006, Springer, Berlin / Heidelberg)

The AnCora-CA corpus ought to be freely downloadable from its website. The download will not work for unregistered and not signed in users. The website offers creating new account but it is not automatic, one has to wait for approval.

Republication of the two CoNLL versions in LDC is planned but it has not happenned yet.

The CoNLL 2007 license in short:

research and demonstrative usage
no redistribution
cite in publications
- The original CoNLL 2007 license required a reference to the CESS-ECE project, not a publication: M. Antònia Martí Antonín, Mariona Taulé Delor, Lluís Màrquez, Manuel Bertran (2007) CESS-ECE: A Multilingual and Multilevel Annotated Corpus.
- Later there was the LREC paper, which is now the required reference for the AnCora corpus.

AnCora-CA was created by members of the Centre de Llenguatge i Computació (CLiC), Universitat de Barcelona, Gran Via de les Corts Catalanes 585, E-08007 Barcelona, Spain.

Website
- http://clic.ub.edu/corpus/
Data
- no separate citation
Principal publications
- Mariona Taulé, M. Antònia Martí, Marta Recasens: AnCora: Multilevel Annotated Corpora for Catalan and Spanish. In: Proceedings of the Sixth International Conference on Language Resources and Evaluation (LREC'08), Marrakech, Morocco, 2008. ISBN 2-9517408-4-0
Documentation
- Maria Antònia Martí, Mariona Taulé, Manu Bertran, Lluís Màrquez: AnCora: Multilingual and Multilevel Annotated Corpora. Draft Technical report, online.
- Morphology
- Syntactic guidelines

Mostly newswire (EFE news, ACN Catalan news, Catalan version of El Periódico, 2000).

The CoNLL 2007 version contains 435,860 tokens in 15125 sentences, yielding 28.82 tokens per sentence on average (CoNLL 2007 data split: 430,844 tokens / 14958 sentences training, 5016 tokens / 167 sentences test).

The CoNLL 2009 version contains 496,672 tokens in 16786 sentences, yielding 29.59 tokens per sentence on average (CoNLL 2009 data split: 390,302 tokens / 13200 sentences training, 53015 tokens / 1724 sentences development, 53355 tokens / 1862 sentences test).

The original morphosyntactic tags (EAGLES?) have been converted to fit into the three columns (CPOS, POS and FEAT) columns of the CoNLL 2006/7 format, resp. the two columns (POS and FEAT) of the CoNLL 2009 format. Note that the missing CPOS column is not the only difference between the two conversion schemes. Feature names and values in the FEAT column are different, too.

The morphosyntactic tags have been disambiguated manually. The CoNLL 2009 version also contains automatically disambiguated tags.

Multi-word expressions have been collapsed into one token, using underscore as the joining character. This includes named entities (e.g. La_Garrotxa, Ajuntament_de_Manresa, dilluns_4_de_juny) and prepositional compounds (pel_que_fa_al, d'_acord_amb, la_seva, a_més_de). Empty (underscore) tokens have been inserted to represent missing subjects (Catalan is a pro-drop language).

The first sentence of the CoNLL 2007 training data:

1	L'	el	d	da	num=s\|gen=c	2	ESPEC	_	_
2	Ajuntament_de_Manresa	Ajuntament_de_Manresa	n	np	_	4	SUJ	_	_
3	ha	haver	v	va	num=s\|per=3\|mod=i\|ten=p	4	AUX	_	_
4	posat_en_funcionament	posar_en_funcionament	v	vm	num=s\|mod=p\|gen=m	0	S	_	_
5	tot	tot	d	di	num=s\|gen=m	7	ESPEC	_	_
6	un_seguit_de	un_seguit_de	d	di	num=p\|gen=c	5	DET	_	_
7	mesures	mesura	n	nc	num=p\|gen=f	4	CD	_	_
8	,	,	F	Fc	_	10	PUNC	_	_
9	la	el	d	da	num=s\|gen=f	10	ESPEC	_	_
10	majoria	majoria	n	nc	num=s\|gen=f	7	_	_	_
11	informatives	informatiu	a	aq	num=p\|gen=f	10	_	_	_
12	,	,	F	Fc	_	10	PUNC	_	_
13	que	que	p	pr	num=n\|gen=c	14	SUJ	_	_
14	tenen	tenir	v	vm	num=p\|per=3\|mod=i\|ten=p	7	SF	_	_
15	com_a	com_a	s	sp	for=s	14	CPRED	_	_
16	finalitat	finalitat	n	nc	num=s\|gen=f	15	SN	_	_
17	minimitzar	minimitzar	v	vm	mod=n	14	CD	_	_
18	els	el	d	da	num=p\|gen=m	19	ESPEC	_	_
19	efectes	efecte	n	nc	num=p\|gen=m	17	SN	_	_
20	de	de	s	sp	for=s	19	SP	_	_
21	la	el	d	da	num=s\|gen=f	22	ESPEC	_	_
22	vaga	vaga	n	nc	num=s\|gen=f	20	SN	_	_
23	.	.	F	Fp	_	4	PUNC	_	_

The first sentence of the CoNLL 2007 test data:

1	Tot_i_que	tot_i_que	c	cs	_	5	SUBORD	_	_
2	ahir	ahir	r	rg	_	5	CC	_	_
3	hi	hi	p	pp	num=n\|per=3\|gen=c	5	MORF	_	_
4	va	anar	v	va	num=s\|per=3\|mod=i\|ten=p	5	AUX	_	_
5	haver	haver	v	va	mod=n	15	AO	_	_
6	una	un	d	di	num=s\|gen=f	7	ESPEC	_	_
7	reunió	reunió	n	nc	num=s\|gen=f	5	CD	_	_
8	de	de	s	sp	for=s	7	SP	_	_
9	darrera	darrer	a	ao	num=s\|gen=f	10	SADJ	_	_
10	hora	hora	n	nc	num=s\|gen=f	8	SN	_	_
11	,	,	F	Fc	_	5	PUNC	_	_
12	no	no	r	rn	_	15	MOD	_	_
13	es	es	p	p0	_	15	PASS	_	_
14	va	anar	v	va	num=s\|per=3\|mod=i\|ten=p	15	AUX	_	_
15	aconseguir	aconseguir	v	vm	mod=n	0	S	_	_
16	acostar	acostar	v	vm	mod=n	15	SUJ	_	_
17	posicions	posició	n	nc	num=p\|gen=f	16	SN	_	_
18	,	,	F	Fc	_	23	PUNC	_	_
19	de_manera_que	de_manera_que	c	cs	_	23	SUBORD	_	_
20	els	el	d	da	num=p\|gen=m	21	ESPEC	_	_
21	treballadors	treballador	n	nc	num=p\|gen=m	23	SUJ	_	_
22	han	haver	v	va	num=p\|per=3\|mod=i\|ten=p	23	AUX	_	_
23	decidit	decidir	v	vm	num=s\|mod=p\|gen=m	15	AO	_	_
24	anar	anar	v	vm	mod=n	23	CD	_	_
25	a	a	s	sp	for=s	24	CREG	_	_
26	la	el	d	da	num=s\|gen=f	27	ESPEC	_	_
27	vaga	vaga	n	nc	num=s\|gen=f	25	SN	_	_
28	.	.	F	Fp	_	15	PUNC	_	_

The first sentence of the CoNLL 2009 training data:

1

El

el

d

postype=article|gen=m|num=s

2

spec

_

2

Tribunal_Suprem

n

postype=proper|gen=c|num=c

7

suj

_

arg0-agt

_

3

(

f

punct=bracket|punctenclose=open

4

f

_

4

TS

n

postype=proper|gen=c|num=c

2

sn

_

5

)

f

punct=bracket|punctenclose=close

4

f

_

6

ha

haver

v

7

v

_

7

confirmat

confirmar

v

postype=main|gen=m|num=s|mood=pastparticiple

0

sentence

Y

confirmar.a32

_

8

la

el

d

postype=article|gen=f|num=s

9

spec

_

9

condemna

n

postype=common|gen=f|num=s

7

cd

_

arg1-pat

_

10

a

s

postype=preposition|gen=c|num=c

9

sp

_

11

quatre

d

postype=numeral|gen=c|num=p

12

spec

_

12

anys

any

n

postype=common|gen=m|num=p

10

sn

_

13

d'

de

s

postype=preposition|gen=c|num=c

12

sp

_

14

inhabilitació

n

postype=common|gen=f|num=s

13

sn

_

15

especial

a

postype=qualificative|gen=c|num=s

14

s.a

_

16

i

c

postype=coordinating

12

9

coord

_

17

una

un

d

postype=indefinite|gen=f|num=s

postype=numeral|gen=f|num=s

18

spec

_

18

multa

n

postype=common|gen=f|num=s

12

9

sn

_

19

de

s

postype=preposition|gen=c|num=c

18

sp

_

20

3,6

3.6

3,6

z

n

_

postype=proper|gen=c|num=c

21

spec

_

21

milions

milió

n

postype=common|gen=m|num=p

19

sn

_

22

de

s

postype=preposition|gen=c|num=c

21

sp

_

23

pessetes

pesseta

z

n

postype=currency

postype=common|gen=f|num=p

22

sn

_

24

per

s

postype=preposition|gen=c|num=c

9

sp

_

25

a

s

postype=preposition|gen=c|num=c

24

sp

_

26

quatre

d

postype=numeral|gen=c|num=p

27

spec

_

27

veterinaris

veterinari

n

postype=common|gen=m|num=p

25

sn

_

28

gironins

gironí

a

postype=qualificative|gen=m|num=p

27

s.a

_

29

,

f

punct=comma

30

f

_

30

per

s

postype=preposition|gen=c|num=c

9

7

sp

cc

_

31

haver

v

n

postype=auxiliary|gen=c|num=c|mood=infinitive

postype=common|gen=m|num=s

33

v

_

32

-se

ell

p

gen=c|num=c|person=3

33

morfema.pronominal

_

33

beneficiat

beneficiar

beneficiat

v

a

postype=main|gen=m|num=s|mood=pastparticiple

postype=qualificative|gen=m|num=s|posfunction=participle

42

30

S

Y

beneficiar.a2

_

34

dels

del

dels

s

postype=preposition|gen=m|num=p|contracted=yes

33

creg

_

arg1-null

_

35

càrrecs

càrrec

n

postype=common|gen=m|num=p

34

sn

_

36

públics

públic

a

postype=qualificative|gen=m|num=p

35

s.a

_

37

que

p

postype=relative|gen=c|num=c

39

cd

_

arg1-pat

_

38

_

p

_

39

suj

_

arg0-agt

_

39

desenvolupaven

desenvolupar

v

35

S

Y

desenvolupar.a2

_

40

i

c

postype=coordinating

42

33

coord

_

41

la_seva

el_seu

d

postype=possessive|gen=f|num=s|person=3

42

spec

_

42

relació

n

postype=common|gen=f|num=s

30

33

sn

cd

_

43

amb

s

postype=preposition|gen=c|num=c

42

sp

_

44

les

el

d

postype=article|gen=f|num=p

45

spec

_

45

empreses

empresa

n

postype=common|gen=f|num=p

43

sn

_

46

càrniques

càrnic

a

postype=qualificative|gen=f|num=p

45

s.a

_

47

de

s

postype=preposition|gen=c|num=c

45

sp

_

48

la

el

d

postype=article|gen=f|num=s

49

spec

_

49

zona

n

postype=common|gen=f|num=s

47

sn

_

50

en

s

postype=preposition|gen=c|num=c

42

sp

_

51

oferir

v

postype=main|gen=c|num=c|mood=infinitive

50

S

Y

oferir.a32

_

52

-los

ell

p

postype=personal|gen=c|num=p|person=3

51

ci

_

arg2-ben

53

serveis

servei

n

postype=common|gen=m|num=p

51

cd

_

arg1-pat

54

particulars

particular

a

postype=qualificative|gen=c|num=p

53

s.a

_

55

.

f

punct=period

7

f

_

The first sentence of the CoNLL 2009 development data:

1

Fundació_Privada_Fira_de_Manresa

n

postype=proper|gen=c|num=c

3

suj

_

arg0-agt

2

ha

haver

v

3

v

_

3

fet

fer

v

postype=main|gen=m|num=s|mood=pastparticiple

0

sentence

Y

fer.a2

_

4

un

d

postype=numeral|gen=m|num=s

5

spec

_

5

balanç

n

postype=common|gen=m|num=s

3

cd

_

arg1-pat

6

de

s

postype=preposition|gen=c|num=c

5

sp

_

7

l'

el

d

postype=article|gen=c|num=s

8

spec

_

8

activitat

n

postype=common|gen=f|num=s

6

sn

_

9

del

s

postype=preposition|gen=m|num=s|contracted=yes

8

sp

_

10

Palau_Firal

n

postype=proper|gen=c|num=c

9

sn

_

11

durant

s

postype=preposition|gen=c|num=c

8

3

sp

cc

_

12

els

el

d

postype=article|gen=m|num=p

15

spec

_

13

primers

primer

a

postype=ordinal|gen=m|num=p

12

a

_

14

cinc

d

postype=numeral|gen=c|num=p

12

d

_

15

mesos

mes

n

postype=common|gen=m|num=p

11

sn

_

16

de

s

postype=preposition|gen=c|num=c

15

sp

_

17

l'

el

d

postype=article|gen=c|num=s

18

spec

_

18

any

n

postype=common|gen=m|num=s

16

sn

_

19

.

f

punct=period

3

f

_

The first sentence of the CoNLL 2009 test data:

1

El

el

d

postype=article|gen=m|num=s

_

2

darrer

a

postype=ordinal|gen=m|num=s

_

3

número

n

postype=common|gen=m|num=s

_

4

de

s

postype=preposition|gen=c|num=c

_

5

l'

el

d

postype=article|gen=c|num=s

_

6

Observatori_del_Mercat_de_Treball_d'_Osona

n

postype=proper|gen=c|num=c

_

7

inclou

incloure

v

_

Y

8

un

d

postype=numeral|gen=m|num=s

_

9

informe

n

postype=common|gen=m|num=s

_

10

especial

a

postype=qualificative|gen=c|num=s

_

11

sobre

s

postype=preposition|gen=c|num=c

_

12

la

el

d

postype=article|gen=f|num=s

_

13

contractació

n

postype=common|gen=f|num=s

_

14

a_través_de

s

postype=preposition|gen=c|num=c

_

15

les

el

d

postype=article|gen=f|num=p

_

16

empreses

empresa

n

postype=common|gen=f|num=p

_

17

de

s

postype=preposition|gen=c|num=c

_

18

treball

n

postype=common|gen=m|num=s

_

19

temporal

a

postype=qualificative|gen=c|num=s

_

20

,

f

punct=comma

_

21

les

el

d

postype=article|gen=f|num=p

_

22

ETT

n

postype=proper|gen=c|num=c

_

23

.

f

punct=period

_

Nonprojectivities in AnCora-CA are very rare. Only 487 of the 435,860 tokens in the CoNLL 2007 version are attached nonprojectively (0.11%). In the CoNLL 2009 version, there are no nonprojectivities at all.

The results of the CoNLL 2007 shared task are available online. They have been published in (Nivre et al., 2007). The evaluation procedure was changed to include punctuation tokens. These are the best results for Catalan:

Parser (Authors)	LAS	UAS
Titov et al.	87.40	93.40
Sagae	88.16	93.34
Malt (Nilsson et al.)	88.70	93.12
Nakagawa	87.90	92.86
Carreras	87.60	92.46
Malt (Hall et al.)	87.74	92.20

The two Malt parser results of 2007 (single malt and blended) are described in (Hall et al., 2007) and the details about the parser configuration are described here.

The results of the CoNLL 2009 shared task are available online. They have been published in (Hajič et al., 2009). Unlabeled attachment score was not published. These are the best results for Catalan:

Parser (Authors)	LAS
Merlo	87.86
Che	86.56
Bohnet	86.35
Chen	85.88

PDT 0.5 (1998)
PDT 1.0 (2001)
PDT 2.0 (2006)
CoNLL 2006
CoNLL 2007
CoNLL 2009

The CoNLL 2006 version is based on PDT 1.0. The CoNLL 2007 and 2009 versions are based on PDT 2.0.

The original PDT 1.0 and 2.0 is distributed by the LDC under the catalogue numbers LDC2001T10 and LDC2006T01. It is free for LDC members 2001 and 2006, price for non-members is unknown (contact LDC). The license in short:

non-commercial research usage
no redistribution
include in publications: “The Prague Dependency Treebank, version 2.0 has been developed by the Institute of Formal and Applied Linguistics, http://ufal.mff.cuni.cz/.”

The CoNLL 2006, 2007 and 2009 versions are obtainable upon request under similar license terms. Their publication in the LDC together with the other CoNLL treebanks is being prepared.

PDT was created by members of the Institute of Formal and Applied Linguistics (Ústav formální a aplikované lingvistiky, ÚFAL), Faculty of Mathematics and Physics (Matematicko-fyzikální fakulta), Charles University in Prague (Univerzita Karlova v Praze), Malostranské náměstí 25, Praha, CZ-11800, Czechia. The CoNLL 2006 conversion of the treebank was prepared by Yuval Krymolowski; the CoNLL 2007 and 2009 conversions were prepared by ÚFAL (Zdeněk Žabokrtský and Jan Štěpánek).

Website
- http://ufal.mff.cuni.cz/pdt2.0/
Data
- Jan Hajič, Eva Hajičová, Petr Pajas, Jarmila Panevová, Petr Sgall: Prague Dependency Treebank 1.0 (LDC2001T10). Linguistic Data Consortium, Philadelphia, USA, 2001. ISBN 1-58563-212-0.
- Jan Hajič, Eva Hajičová, Jarmila Panevová, Petr Sgall, Petr Pajas, Jan Štěpánek, Jiří Havelka, Marie Mikulová: Prague Dependency Treebank 2.0 (LDC2006T01). Linguistic Data Consortium, Philadelphia, USA, 2006. ISBN 1-58563-370-4.
Principal publications
- Jan Hajič, Alena Böhmová, Eva Hajičová, Barbora Hladká: The Prague Dependency Treebank: A Three-Level Annotation Scenario. In: Anne Abeillé (ed.): Treebanks: Building and Using Parsed Corpora, pages 103-127, Kluwer, Amsterdam, The Netherlands, 2000.
Documentation
- Jiří Hana, Daniel Zeman: Manual for Morphological Annotation, Revision for the Prague Dependency Treebank 2.0, ÚFAL Technical Report No. 2005-27, Praha, Czechia, 2005.
- Jan Hajič, Jarmila Panevová, Eva Buráňová, Zdeňka Urešová, Alla Bémová: Annotations at Analytical Level, Instructions for annotators, ÚFAL MFF UK, Praha, Czechia, 1999.

Newswire text (Lidové noviny, Mladá fronta Dnes), business weekly (Českomoravský Profit) and a scientific magazine (Vesmír).

All distributions of PDT are officially split to training, development (d-test) and test (e-test) data sets. PDT 2.0 contains data that are annotated only morphologically (M-layer), those that are annotated both morphologically and analytically (surface syntax; M+A layers), and the smallest subset is also annotated tectogrammatically (M+A+T layers). The statistics in this section cover the M+A subset, which is relevant for surface dependency parsing.

Size of CoNLL 2007 data was limited because some teams of CoNLL 2006 complained that they did not have enough time and resources to train the larger models. For CoNLL 2009, only that part of PDT was selected that contained also tectogrammatical annotation, because the 2009 task included semantic learning.

Parts of the following table have been taken from (Zeman 2004, page 21). Only non-empty sentences counted (e.g. PDT 1.0 had 81614 sentence tags but only 73088 non-empty ones).

Version	Train Sentences	Train Tokens	D-test Sentences	D-test Tokens	E-test Sentences	E-test Tokens	Total Sentences	Total Tokens	Sentence Length
PDT 0.5	19126	327,597	3697	63718	3787	65390	26610	456,705	17.16
PDT 1.0	73088	1,255,590	7319	126,030	7507	125,713	87914	1,489,748	16.95
PDT 2.0	68562	1,172,299	9270	158,962	10148	173,586	87980	1,504,847	17.10
CoNLL 2006	72703	1,249,408	365	5853			73068	1,255,261	17.18
CoNLL 2007	25364	432,296	286	4724			25650	437,020	17.04
CoNLL 2009	38727	652,544	5228	87988	4213	70348	48168	810,880	16.83

PDT 1.0 is distributed in the CSTS format. PDT 2.0 uses the PML format. CoNLL 2006 and 2007 uses the CoNLL-X format; CoNLL 2009 format is slightly different (number and meaning of columns). Unlike the other formats, the CSTS format used the ISO-8859-2 character encoding.

The CSTS format (PDT 0.5 and 1.0) contains morphological annotation (lemmas and tags) both manual and by two taggers. The CoNLL 2009 version contains manual and one automatic disambiguation. The official distribution of PDT 2.0 and the CoNLL 2006 and 2007 versions contain only manual morphology.

The original PDT uses 15-character positional morphological tags. The CoNLL versions convert the tags to the two/three CoNLL columns, CPOS, POS and FEAT. In addition, the CoNLL versions contain the Sem feature, which is derived from the tags attached to lemma in PDT (see Hana and Zeman, 2005).

See above for documentation of the morphological tags. All CoNLL distributions contain a README file with a brief description of the parts of speech and features. Use DZ Interset to inspect the PDT and the CoNLL tagsets.

The guidelines for syntactic annotation are documented in the PDT annotation manual.

The first sentence of the PDT 1.0 training data:

<csts lang=cs>
<h>
<source>Českomoravský profit</source>
<markup>
<mauth>js
<mdate>1996-2000
<mdesc>Manual analytical annotation
</markup>
<markup>
<mauth>kk,lk
<mdate>1996-2000
<mdesc>Manual morphological annotation
</markup>
</h>
<doc file="s/inf/j/1994/cmpr9406" id="001">
<a>
<mod>s
<txtype>inf
<genre>mix
<med>j
<temp>1994
<authname>y
<opus>cmpr9406
<id>001
</a>
<c>
<p n=1>
<s id="cmpr9406:001-p1s1">
<p n=2>
<s id="cmpr9406:001-p2s1">
<f cap>Třikrát<l>třikrát`3<t>Cv-------------<MDl src="a">třikrát`3<MDt src="a">Cv-------------<MDl src="b">třikrát`3<MDt src="b">Cv-------------<A>Adv<r>1<g>2
<f>rychlejší<l>rychlý<t>AAFS1----2A----<MDl src="a">rychlý<MDt src="a">AANS1----2A----<MDl src="b">rychlý<MDt src="b">AAFS1----2A----<A>ExD<r>2<g>0
<f>než<l>než-2<t>J,-------------<MDl src="a">než-2<MDt src="a">J,-------------<MDl src="b">než-2<MDt src="b">J,-------------<A>AuxC<r>3<g>2
<f>slovo<l>slovo<t>NNNS1-----A----<MDl src="a">slovo<MDt src="a">NNNS4-----A----<MDl src="b">slovo<MDt src="b">NNNS1-----A----<A>ExD<r>4<g>3

The first two sentences of the PDT 1.0 d-test data:

<csts lang=cs>
<h>
<source>Lidové noviny</source>
<markup>
<mauth>zu
<mdate>1996-2000
<mdesc>Manual analytical annotation
</markup>
</h>
<doc file="s/pub/nws/1994/ln94206" id="1">
<a>
<mod>s
<txtype>pub
<genre>mix
<med>nws
<temp>1994
<authname>y
<opus>ln94206
<id>1
</a>
<c>
<p n=1>
<s id="ln94206:1-p1s1">
<i>ti
<f cap>Lidé<MDl src="a">člověk<MDt src="a">NNMP1-----A---1<MDl src="b">člověk<MDt src="b">NNMP1-----A---1<A>ExD<r>1<g>0
<p n=2>
<s id="ln94206:1-p2s1">
<f upper.abbr>ING<MDl src="a">Ing-1_:B_^(inženýr)<MDt src="a">NNMXX-----A---8<MDl src="b">Ing-1_:B_^(inženýr)<MDt src="b">NNMXX-----A---8<A>Atr<r>1<g>4
<D>
<d>.<MDl src="a">.<MDt src="a">Z:-------------<MDl src="b">.<MDt src="b">Z:-------------<A>AuxG<r>2<g>1
<f upper>PETR<MDl src="a">Petr_;Y<MDt src="a">NNMS1-----A----<MDl src="b">Petr_;Y<MDt src="b">NNMS1-----A----<A>Atr<r>3<g>4
<f upper>KARAS<MDl src="a">karas<MDt src="a">NNMS1-----A----<MDl src="b">karas<MDt src="b">NNMS1-----A----<A>Sb_Ap<r>4<g>11
<D>
<d>,<MDl src="a">,<MDt src="a">Z:-------------<MDl src="b">,<MDt src="b">Z:-------------<A>AuxX<r>5<g>6
<f mixed>CSc<MDl src="a">CSc-1_:B_^(kandidát_věd)<MDt src="a">NNMXX-----A---8<MDl src="b">CSc-1_:B_^(kandidát_věd)<MDt src="b">NNMXX-----A---8<A>Atr<r>6<g>4
<D>
<d>.<MDl src="a">.<MDt src="a">Z:-------------<MDl src="b">.<MDt src="b">Z:-------------<A>AuxG<r>7<g>6
<d>(<MDl src="a">(<MDt src="a">Z:-------------<MDl src="b">(<MDt src="b">Z:-------------<A>ExD<r>8<g>4
<D>
<f num>53<MDl src="a">53<MDt src="a">C=-------------<MDl src="b">53<MDt src="b">C=-------------<A>ExD_Pa<r>9<g>4
<D>
<d>)<MDl src="a">)<MDt src="a">Z:-------------<MDl src="b">)<MDt src="b">Z:-------------<A>ExD<r>10<g>4
<D>
<d>,<MDl src="a">,<MDt src="a">Z:-------------<MDl src="b">,<MDt src="b">Z:-------------<A>Apos<r>11<g>20
<f>generální<MDl src="a">generální<MDt src="a">AAMS1----1A----<MDl src="b">generální<MDt src="b">AAMS1----1A----<A>Atr<r>12<g>13
<f>ředitel<MDl src="a">ředitel<MDt src="a">NNMS1-----A----<MDl src="b">ředitel<MDt src="b">NNMS1-----A----<A>Sb_Co<r>13<g>15
<f upper>ČEZ<MDl src="a">ČEZ-1_:B_;K_^(České_energetické_závody)<MDt src="a">NNIPX-----A---8<MDl src="b">ČEZ-1_:B_;K_^(České_energetické_závody)<MDt src="b">NNIPX-----A---8<A>Atr<r>14<g>13
<f>a<MDl src="a">a-1<MDt src="a">J^-------------<MDl src="b">a-1<MDt src="b">J^-------------<A>Coord_Ap<r>15<g>11
<f>předseda<MDl src="a">předseda<MDt src="a">NNMS1-----A----<MDl src="b">předseda<MDt src="b">NNMS1-----A----<A>Sb_Co<r>16<g>15
<f>jeho<MDl src="a">jeho_^(přivlast.)<MDt src="a">PSXXXZS3-------<MDl src="b">jeho_^(přivlast.)<MDt src="b">PSXXXZS3-------<A>Atr<r>17<g>18
<f>představenstva<MDl src="a">představenstvo<MDt src="a">NNNS2-----A----<MDl src="b">představenstvo<MDt src="b">NNNS2-----A----<A>Atr<r>18<g>16
<D>
<d>,<MDl src="a">,<MDt src="a">Z:-------------<MDl src="b">,<MDt src="b">Z:-------------<A>AuxX<r>19<g>11
<f>je<MDl src="a">být<MDt src="a">VB-S---3P-AA---<MDl src="b">být<MDt src="b">VB-S---3P-AA---<A>Pred<r>20<g>0
<f>absolventem<MDl src="a">absolvent<MDt src="a">NNMS7-----A----<MDl src="b">absolvent<MDt src="b">NNMS7-----A----<A>Pnom<r>21<g>20
<f>elektrotechnické<MDl src="a">elektrotechnický<MDt src="a">AAFS2----1A----<MDl src="b">elektrotechnický<MDt src="b">AAFS2----1A----<A>Atr<r>22<g>23
<f>fakulty<MDl src="a">fakulta<MDt src="a">NNFS2-----A----<MDl src="b">fakulta<MDt src="b">NNFS2-----A----<A>Atr_Co<r>23<g>25
<f upper>ČVUT<MDl src="a">ČVUT-1_:B_;K_^(České_vysoké_učení_technické)<MDt src="a">NNNXX-----A---8<MDl src="b">ČVUT-1_:B_;K_^(České_vysoké_učení_technické)<MDt src="b">NNNXX-----A---8<A>Atr<r>24<g>23
<f>a<MDl src="a">a-1<MDt src="a">J^-------------<MDl src="b">a-1<MDt src="b">J^-------------<A>Coord<r>25<g>21
<f>postgraduálního<MDl src="a">postgraduální<MDt src="a">AANS2----1A----<MDl src="b">postgraduální<MDt src="b">AANS2----1A----<A>Atr<r>26<g>27
<f>studia<MDl src="a">studium<MDt src="a">NNNS2-----A----<MDl src="b">studium<MDt src="b">NNNS2-----A----<A>Atr_Co<r>27<g>25
<f>v<MDl src="a">v-1<MDt src="a">RR--6----------<MDl src="b">v-1<MDt src="b">RR--6----------<A>AuxP<r>28<g>29
<f>oboru<MDl src="a">obor_^(lidské_činnosti)<MDt src="a">NNIS6-----A----<MDl src="b">obor_^(lidské_činnosti)<MDt src="b">NNIS6-----A----<A>AuxP<r>29<g>27
<f>metod<MDl src="a">metoda<MDt src="a">NNFP2-----A----<MDl src="b">metoda<MDt src="b">NNFP2-----A----<A>Atr<r>30<g>29
<f>operační<MDl src="a">operační<MDt src="a">AAFS2----1A----<MDl src="b">operační<MDt src="b">AAFS2----1A----<A>Atr<r>31<g>32
<f>analýzy<MDl src="a">analýza<MDt src="a">NNFS2-----A----<MDl src="b">analýza<MDt src="b">NNFS2-----A----<A>Atr<r>32<g>30
<D>
<d>.<MDl src="a">.<MDt src="a">Z:-------------<MDl src="b">.<MDt src="b">Z:-------------<A>AuxK<r>33<g>0

The first sentence of the PDT 1.0 e-test data:

<csts lang=cs>
<h>
<source>Lidové noviny</source>
<markup>
<mauth>zu
<mdate>1996-2000
<mdesc>Manual analytical annotation
</markup>
</h>
<doc file="s/pub/nws/1994/ln94209" id="1">
<a>
<mod>s
<txtype>pub
<genre>mix
<med>nws
<temp>1994
<authname>y
<opus>ln94209
<id>1
</a>
<c>
<p n=1>
<s id="ln94209:1-p1s1">
<f cap>Přádelny<MDl src="a">přádelna<MDt src="a">NNFP1-----A----<MDl src="b">přádelna<MDt src="b">NNFP1-----A----<A>Sb<r>1<g>2
<f>mají<MDl src="a">mít<MDt src="a">VB-P---3P-AA---<MDl src="b">mít<MDt src="b">VB-P---3P-AA---<A>Pred<r>2<g>0
<f>dvojnásob<MDl src="a">dvojnásob<MDt src="a">Db-------------<MDl src="b">dvojnásob<MDt src="b">Db-------------<A>Obj<r>3<g>2
<f>vad<MDl src="a">vada<MDt src="a">NNFP2-----A----<MDl src="b">vada<MDt src="b">NNFP2-----A----<A>Atr<r>4<g>3

Morphological annotation of the first amw training file of the PDT 2.0:

<mdata xmlns="http://ufal.mff.cuni.cz/pdt/pml/">
 <head>
  <schema href="mdata_schema.xml" />
  <references>
   <reffile id="w" name="wdata" href="cmpr9406_001.w.gz" />
  </references>
 </head>
 <meta>
  <lang>cs</lang>
  <annotation_info id="manual">
   <desc>Manual annotation</desc>
  </annotation_info>
 </meta>
 <s id="m-cmpr9406-001-p2s1">
  <m id="m-cmpr9406-001-p2s1w1">
   <src.rf>manual</src.rf>
   <w.rf>w#w-cmpr9406-001-p2s1w1</w.rf>
   <form>Třikrát</form>
   <lemma>třikrát`3</lemma>
   <tag>Cv-------------</tag>
  </m>
  <m id="m-cmpr9406-001-p2s1w2">
   <src.rf>manual</src.rf>
   <w.rf>w#w-cmpr9406-001-p2s1w2</w.rf>
   <form>rychlejší</form>
   <lemma>rychlý</lemma>
   <tag>AAFS1----2A----</tag>
  </m>
  <m id="m-cmpr9406-001-p2s1w3">
   <src.rf>manual</src.rf>
   <w.rf>w#w-cmpr9406-001-p2s1w3</w.rf>
   <form>než</form>
   <lemma>než-2</lemma>
   <tag>J,-------------</tag>
  </m>
  <m id="m-cmpr9406-001-p2s1w4">
   <src.rf>manual</src.rf>
   <w.rf>w#w-cmpr9406-001-p2s1w4</w.rf>
   <form>slovo</form>
   <lemma>slovo</lemma>
   <tag>NNNS1-----A----</tag>
  </m>
 </s>

Analytical (surface-syntactic) annotation of the first amw training file of the PDT 2.0:

<adata xmlns="http://ufal.mff.cuni.cz/pdt/pml/">
 <head>
  <schema href="adata_schema.xml" />
  <references>
   <reffile id="m" name="mdata" href="cmpr9406_001.m.gz" />
   <reffile id="w" name="wdata" href="cmpr9406_001.w.gz" />
  </references>
 </head>
 <meta>
  <annotation_info>
   <desc>Manual annotation</desc>
  </annotation_info>
 </meta>
 <trees>
  <LM id="a-cmpr9406-001-p2s1">
   <s.rf>m#m-cmpr9406-001-p2s1</s.rf>
   <ord>0</ord>
   <children>
    <LM id="a-cmpr9406-001-p2s1w2">
     <m.rf>m#m-cmpr9406-001-p2s1w2</m.rf>
     <afun>ExD</afun>
     <ord>2</ord>
     <children>
      <LM id="a-cmpr9406-001-p2s1w1">
       <m.rf>m#m-cmpr9406-001-p2s1w1</m.rf>
       <afun>Adv</afun>
       <ord>1</ord>
      </LM>
      <LM id="a-cmpr9406-001-p2s1w3">
       <m.rf>m#m-cmpr9406-001-p2s1w3</m.rf>
       <afun>AuxC</afun>
       <ord>3</ord>
       <children>
        <LM id="a-cmpr9406-001-p2s1w4">
         <m.rf>m#m-cmpr9406-001-p2s1w4</m.rf>
         <afun>ExD</afun>
         <ord>4</ord>
        </LM>
       </children>
      </LM>
     </children>
    </LM>
   </children>
  </LM>

The first two sentences of the CoNLL 2006 and 2007 training data:

1	Třikrát	třikrát`3	C	v	_	2	Adv	_	_
2	rychlejší	rychlý	A	A	Gen=F\|Num=S\|Cas=1\|Gra=2\|Neg=A	0	ExD	_	_
3	než	než-2	J	,	_	2	AuxC	_	_
4	slovo	slovo	N	N	Gen=N\|Num=S\|Cas=1\|Neg=A	3	ExD	_	_

1	Faxu	fax	N	N	Gen=I\|Num=S\|Cas=3\|Neg=A	2	Obj	_	_
2	škodí	škodit	V	B	Num=P\|Per=3\|Ten=P\|Neg=A\|Voi=A	0	Pred	_	_
3	především	především	D	b	_	6	AuxZ	_	_
4	přetížené	přetížený	A	A	Gen=F\|Num=P\|Cas=1\|Gra=1\|Neg=A	6	Atr	_	_
5	telefonní	telefonní	A	A	Gen=F\|Num=P\|Cas=1\|Gra=1\|Neg=A	6	Atr	_	_
6	linky	linka	N	N	Gen=F\|Num=P\|Cas=1\|Neg=A	2	Sb	_	_
7	*	*	Z	:	_	2	AuxG	_	_

The first sentence of the CoNLL 2006 test data:

1	Podobně	podobně	D	g	Gra=1\|Neg=A	5	Adv	_	_
2	,	,	Z	:	_	3	AuxX	_	_
3	myslím	myslit	V	B	Num=S\|Per=1\|Ten=P\|Neg=A\|Voi=A	5	Pred_Pa	_	_
4	,	,	Z	:	_	3	AuxX	_	_
5	postupuje	postupovat	V	B	Num=S\|Per=3\|Ten=P\|Neg=A\|Voi=A	0	Pred	_	_
6	většina	většina	N	N	Gen=F\|Num=S\|Cas=1\|Neg=A	5	Sb	_	_
7	českých	český	A	A	Gen=F\|Num=P\|Cas=2\|Gra=1\|Neg=A	8	Atr	_	_
8	bank	banka	N	N	Gen=F\|Num=P\|Cas=2\|Neg=A	6	Atr	_	_
9	,	,	Z	:	_	11	AuxX	_	_
10	zejména	zejména	D	b	_	12	AuxZ	_	_
11	v	v-1	R	R	Cas=6	5	AuxP	_	_
12	případech	případ	N	N	Gen=I\|Num=P\|Cas=6\|Neg=A	11	Adv	_	_
13	,	,	Z	:	_	17	AuxX	_	_
14	kdy	kdy	D	b	_	17	Adv	_	_
15	by	být	V	c	Num=X\|Per=3	17	AuxV	_	_
16	se	se	P	7	Num=X\|Cas=4	18	AuxT	_	_
17	mělo	mít	V	p	Gen=N\|Num=S\|Per=X\|Ten=R\|Neg=A\|Voi=A	12	Atr	_	_
18	jednat	jednat	V	f	Neg=A	17	Obj	_	_
19	o	o-1	R	R	Cas=4	18	AuxP	_	_
20	větší	velký	A	A	Gen=F\|Num=P\|Cas=4\|Gra=2\|Neg=A	21	Atr	_	_
21	částky	částka	N	N	Gen=F\|Num=P\|Cas=4\|Neg=A	19	Obj	_	_
22	.	.	Z	:	_	0	AuxK	_	_

The first sentence of the CoNLL 2007 test data:

1	Proč	proč	D	b	_	2	Adv	_	_
2	mají	mít	V	B	Num=P\|Per=3\|Ten=P\|Neg=A\|Voi=A	0	Pred	_	_
3	každý	každý	A	A	Gen=I\|Num=S\|Cas=4\|Gra=1\|Neg=A	4	Atr	_	_
4	rok	rok	N	N	Gen=I\|Num=S\|Cas=4\|Neg=A	5	Adv	_	_
5	fasovat	fasovat	V	f	Neg=A	2	Obj	_	_
6	speciální	speciální	A	A	Gen=F\|Num=S\|Cas=4\|Gra=1\|Neg=A	7	Atr	_	_
7	taxu	taxa	N	N	Gen=F\|Num=S\|Cas=4\|Neg=A	5	Obj	_	_
8	na	na	R	R	Cas=4	7	AuxP	_	_
9	oblečení	oblečení	N	N	Gen=N\|Num=S\|Cas=4\|Neg=A	8	AtrAdv	_	_
10	?	?	Z	:	_	0	AuxK	_	_

The first sentence of the CoNLL 2009 training data:

1

Celní

celní

A

2

Atr

Y

celní

_

RSTR

_

2

unie

N

SubPOS=N|Gen=F|Num=S|Cas=1|Neg=A

0

ExD

Y

unie

_

3

v

R

SubPOS=R|Cas=6

2

AuxP

_

4

ohrožení

N

SubPOS=N|Gen=N|Num=S|Cas=6|Neg=A

3

Atr

Y

v-w3017f1

_

The first sentence of the CoNLL 2009 development data:

1

|

Z

SubPOS=:

0

3

ExD

AuxG

_

2

Daňový

daňový

A

3

Atr

Y

daňový

_

RSTR

3

poradce

N

SubPOS=N|Gen=M|Num=S|Cas=1|Neg=A

0

ExD

Y

poradce

_

4

|

Z

SubPOS=:

0

3

AuxK

AuxG

_

The first sentence of the CoNLL 2009 test data:

1

Názor

názor

N

SubPOS=N|Gen=I|Num=S|Cas=1|Neg=A

_

Y

2

experta

expert

N

SubPOS=N|Gen=M|Num=S|Cas=2|Neg=A

_

Y

PDT is a mildly nonprojective treebank. 8351 of the 437,020 tokens in the CoNLL 2007 version are attached nonprojectively (1.91%).

There is an online summary of known results in Czech parsing.

The results of the CoNLL 2006 shared task are available online. They have been published in (Buchholz and Marsi, 2006). The evaluation procedure was non-standard because it excluded punctuation tokens. These are the best results for Czech:

Parser (Authors)	LAS	UAS
MST (McDonald et al.)	80.18	87.30
Basis (O'Neil)	76.60	85.58
Malt (Nivre et al.)	78.42	84.80
Nara (Yuchang Cheng)	76.24	83.40

The results of the CoNLL 2007 shared task are available online. They have been published in (Nivre et al., 2007). The evaluation procedure was changed to include punctuation tokens. These are the best results for Czech:

Parser (Authors)	LAS	UAS
Nakagawa	80.19	86.28
Carreras	78.60	85.16
Titov et al.	77.94	84.19
Malt (Nilsson et al.)	77.98	83.59
Attardi et al.	77.37	83.40
Malt (Hall et al.)	77.22	82.35

The two Malt parser results of 2007 (single malt and blended) are described in (Hall et al., 2007) and the details about the parser configuration are described here.

The results of the CoNLL 2009 shared task are available online. They have been published in (Hajič et al., 2009). Unlabeled attachment score was not published. These are the best results for Czech:

Parser (Authors)	LAS
Merlo (Gesmundo et al.)	80.38
Bohnet	80.11
Che et al.	80.01

Original DDT 1.0 in the TEI-based DTAG or Tiger-XML format
CoNLL 2006

The original DDT is based on Discontinuous Grammar. It natively encodes dependencies and other relations such as anaphora. The CoNLL version contains only the dependency relations.

DDT is available under the GNU General Public License version 2. Download the original distribution (DTAG + TIGER-XML formats) from http://www.buch-kromann.dk/matthias/treebank/. Download the CoNLL 2006 conversion from http://ilk.uvt.nl/conll/free_data.html. The license in short:

any usage, commercial or not
modification and redistribution under same license permitted
citation in publications not required (but it is common decency)

DDT was created by members of the Department of International Language Studies and Computational Linguistics, Copenhagen Business School (Handelshøjskolen i København), Dalgas Have 15, DK-2000 Frederiksberg, Denmark. The underlying PAROLE corpus (morphologically annotated) was created by the Society for Danish Language and Literature (Det Danske Sprog- og Litteraturselskab), Christians Brygge 1, DK-1219 København K, Denmark.

Website
- http://www.buch-kromann.dk/matthias/treebank/ (the old and no longer accessible website from http://www.id.cbs.dk/~mtk/ has been moved here)
Data
- no separate citation
Principal publications
- Matthias Trautner Kromann: The Danish Dependency Treebank and the DTAG Treebank Tool. In: Proceedings of Treebanks and Linguistic Theories, Växjö, Sweden, 2003.
Documentation
- see the left-hand-side links at the treebank website, eg.:
- Dependency theory and list of dependency relation labels
- Britt Keson: Vejledning til det danske morfosyntaktisk taggede PAROLE-korpus (morphosyntactic tags). Det Danske Sprog- og Litteraturselskab (DSL)

Unknown (the underlying PAROLE corpus “consists of quotations of 150-250 words from a wide range of randomly selected linguistically representative Danish texts from 1983-1992.”)

The CoNLL 2006 version contains 100,238 tokens in 5512 sentences, yielding 18.19 tokens per sentence on average (CoNLL 2006 data split: 94386 tokens / 5190 sentences training, 5852 tokens / 322 sentences test).

The original morphosyntactic tags have been converted to fit into the three columns (CPOS, POS and FEAT) of the CoNLL format. There should be a 1-1 mapping between the DDT positional tags and the CoNLL 2006 annotation. Use DZ Interset to inspect the CoNLL tagset.

The morphological analysis in the CoNLL 2006 version does not include lemmas (the original DTAG version does contain them). The morphosyntactic tags have been assigned (probably) manually.

Some multi-word expressions have been collapsed into one token, using underscore as the joining character. This includes adverbially used prepositional phrases (e.g. i_lørdags = on Saturdays) but not named entities.

The first sentence of DDT 1.0 in the DTAG format:

<tei.2>
  <teiHeader type=text>
    <fileDesc>
      <titleStmt>
        <title>Tagged sample of: 'Jeltsins skæbnetime'</title>
      </titleStmt>
      <extent words=158>158 running words</extent>
      <publicationStmt>
         <distributor>PAROLE-DK</distributor>
         <address><addrline>Christians Brygge 1,1., DK-1219 Copenhagen K.</address>
         <date>1998-06-02</date>
         <availability status=restricted><p>by agreement with distributor</availability>
      </publicationStmt>
      <sourceDesc>
        <biblStruct>
          <analytic>
            <title>Jeltsins skæbnetime</title>
            <author gender=m born=1925>Nikulin, Leon</author>
          </analytic>
          <monogr>
            <imprint><pubPlace>Denmark</pubPlace>
              <publisher>Det Fri Aktuelt</publisher>
              <date>1992-12-01</date>
            </imprint>
          </monogr>
        </biblStruct>
      </sourceDesc>
    </fileDesc>
    <profileDesc>
      <creation>1992-12-01</creation>
      <langUsage><language>Danish</langUsage>
      <textClass>
        <catRef target="P.M2">
        <catRef target="P.G4.8">
        <catRef target="P.T9.3">
      </textClass>
    </profileDesc>
  </teiHeader>
<text id=AJK>
<body>
<div1 type=main>
<p>
<s>
<W lemma="to" msd="AC---U=--" in="9:subj" out="1:mod|2:mod|3:nobj|5:appr">To</W>
<W lemma="kendt" msd="ANP[CN]PU=[DI]U" in="-1:mod" out="">kendte</W>
<W lemma="russisk" msd="ANP[CN]PU=[DI]U" in="-2:mod" out="">russiske</W>
<W lemma="historiker" msd="NCCPU==I" in="-3:nobj" out="">historikere</W>
<W lemma="Andronik" msd="NP--U==-" in="1:namef" out="">Andronik</W>
<W lemma="Mirganjan" msd="NP--U==-" in="-5:appr" out="-1:namef|1:coord">Mirganjan</W>
<W lemma="og" msd="CC" in="-1:coord" out="2:conj">og</W>
<W lemma="Igor" msd="NP--U==-" in="1:namef" out="">Igor</W>
<W lemma="Klamkin" msd="NP--U==-" in="-2:conj" out="-1:namef">Klamkin</W>
<W lemma="tro" msd="VADR=----A-" in="" out="-9:subj|1:mod|2:pnct|3:dobj|12:pnct">tror</W>
<W lemma="ikke" msd="RGU" in="-1:mod" out="">ikke</W>
<W lemma="," msd="XP" in="-2:pnct" out="">,</W>
<W lemma="at" msd="CS" in="-3:dobj" out="2:vobj">at</W>
<W lemma="Rusland" msd="NP--U==-" in="1:subj|2:[subj]" out="">Rusland</W>
<W lemma="kunne" msd="VADR=----A-" in="-2:vobj" out="-1:subj|1:vobj|2:mod">kan</W>
<W lemma="udvikle" msd="VAF-=----P-" in="-1:vobj" out="-2:[subj]">udvikles</W>
<W lemma="uden" msd="SP" in="-2:mod" out="1:nobj">uden</W>
<W lemma="en" msd="PI-CSU--U" in="-1:nobj" out="2:nobj">en</W>
<W lemma="&quot;" msd="XP" in="1:pnct" out="">"</W>
<W lemma="jernnæve" msd="NCCSU==I" in="-2:nobj" out="-1:pnct|1:pnct">jernnæve</W>
<W lemma="&quot;" msd="XP" in="-1:pnct" out="">"</W>
<W lemma="." msd="XP" in="-12:pnct" out="">.</W>
</s>

The first sentence of the CoNLL 2006 training data:

1	Samme	_	A	AN	degree=pos\|gender=common/neuter\|number=sing/plur\|case=unmarked\|def=def/indef\|transcat=unmarked	0	ROOT	_	_
2	cifre	_	N	NC	gender=neuter\|number=plur\|case=unmarked\|def=indef	1	nobj	_	_
3	,	_	X	XP	_	1	pnct	_	_
4	de	_	P	PD	gender=common/neuter\|number=plur\|case=unmarked\|register=unmarked	7	subj	_	_
5	norske	_	A	AN	degree=pos\|gender=common/neuter\|number=plur\|case=unmarked\|def=def/indef\|transcat=unmarked	4	mod	_	_
6	piger	_	N	NC	gender=common\|number=plur\|case=unmarked\|def=indef	4	nobj	_	_
7	tabte	_	V	VA	mood=indic\|tense=past\|voice=active	1	rel	_	_
8	med	_	SP	SP	_	7	pobj	_	_
9	i_lørdags	_	RG	RG	degree=unmarked	7	mod	_	_
10	mod	_	SP	SP	_	7	pobj	_	_
11	VMs	_	N	NP	case=gen	10	nobj	_	_
12	værtsnation	_	N	NC	gender=common\|number=sing\|case=unmarked\|def=indef	11	possd	_	_
13	.	_	X	XP	_	1	pnct	_	_

The first sentence of the CoNLL 2006 test data:

1	To	_	A	AC	case=unmarked	10	subj	_	_
2	kendte	_	A	AN	degree=pos\|gender=common/neuter\|number=plur\|case=unmarked\|def=def/indef\|transcat=unmarked	1	mod	_	_
3	russiske	_	A	AN	degree=pos\|gender=common/neuter\|number=plur\|case=unmarked\|def=def/indef\|transcat=unmarked	1	mod	_	_
4	historikere	_	N	NC	gender=common\|number=plur\|case=unmarked\|def=indef	1	nobj	_	_
5	Andronik	_	N	NP	case=unmarked	6	namef	_	_
6	Mirganjan	_	N	NP	case=unmarked	1	appr	_	_
7	og	_	C	CC	_	6	coord	_	_
8	Igor	_	N	NP	case=unmarked	9	namef	_	_
9	Klamkin	_	N	NP	case=unmarked	7	conj	_	_
10	tror	_	V	VA	mood=indic\|tense=present\|voice=active	0	ROOT	_	_
11	ikke	_	RG	RG	degree=unmarked	10	mod	_	_
12	,	_	X	XP	_	10	pnct	_	_
13	at	_	C	CS	_	10	dobj	_	_
14	Rusland	_	N	NP	case=unmarked	15	subj	_	_
15	kan	_	V	VA	mood=indic\|tense=present\|voice=active	13	vobj	_	_
16	udvikles	_	V	VA	mood=infin\|voice=passive	15	vobj	_	_
17	uden	_	SP	SP	_	15	mod	_	_
18	en	_	P	PI	gender=common\|number=sing\|case=unmarked\|register=unmarked	17	nobj	_	_
19	“	_	X	XP	_	20	pnct	_	_
20	jernnæve	_	N	NC	gender=common\|number=sing\|case=unmarked\|def=indef	18	nobj	_	_
21	”	_	X	XP	_	20	pnct	_	_
22	.	_	X	XP	_	10	pnct	_	_

Nonprojectivities in DDT are not frequent. Only 988 of the 100,238 tokens in the CoNLL 2006 version are attached nonprojectively (0.99%).

The results of the CoNLL 2006 shared task are available online. They have been published in (Buchholz and Marsi, 2006). The evaluation procedure was non-standard because it excluded punctuation tokens. These are the best results for Danish:

Parser (Authors)	LAS	UAS
MST (McDonald et al.)	84.79	90.58
Malt (Nivre et al.)	84.77	89.80
Riedel et al.	83.63	89.66

TIGER Treebank 1 (2003)
TIGER Treebank 2 (2005)
TIGER Treebank 2.1 (2007) in TIGER-XML or Negra export (text) format
CoNLL 2006
CoNLL 2009

The TIGER Treebank is freely downloadable after you accept the license terms by pressing a button.

Republication of the two CoNLL versions in LDC is planned but it has not happenned yet.

The license in short:

non-commercial research and evaluation usage by academic or educational institutions
no redistribution
acknowledge the use of the corpus in publications

The TIGER Treebank was created by members of three institutes:

Department of Computational Linguistics and Phonetics (Computerlinguistik, CoLi), Saarland University (Universität des Saarlandes), Postfach 151150, D-66041 Saarbrücken, Germany.
Institute for Natural Language Processing (Institut für Maschinelle Sprachverarbeitung, IMS), University of Stuttgart (Universität Stuttgart), Azenbergstraße 12, D-70174 Stuttgart, Germany.
German Department (Institut für Germanistik), Philosophische Fakultät, Universität Potsdam, Am Neuen Palais 10, Haus 05, D-14469 Potsdam, Germany.

Website
- http://www.ims.uni-stuttgart.de/projekte/TIGER/TIGERCorpus/
Data
- no separate citation
Principal publications
- Sabine Brants, Stefanie Dipper, Silvia Hansen, Wolfgang Lezius, George Smith: The TIGER Treebank. In: Proceedings of the Workshop on Treebanks and Linguistic Theories (TLT), Sozopol, Bulgaria, 2002.
- List of publications
Documentation
- Stuttgart-Tübingen Tagset (part of speech)
- Berthold Crysmann, Silvia Hansen-Schirra, George Smith, Dorothea Ziegler-Eisele: TIGER Morphologie-Annotationsschema, 2005.
- Stefanie Albert, Jan Anderssen, Regine Bader, Stephanie Becker, Tobias Bracht, Sabine Brants, Thorsten Brants, Vera Demberg, Stefanie Dipper, Peter Eisenberg, Silvia Hansen, Hagen Hirschmann, Juliane Janitzek, Carolin Kirstein, Robert Langner, Lukas Michelbacher, Oliver Plaehn, Cordula Preis, Marcus Pußel, Marco Rower, Bettina Schrader, Anne Schwartz, George Smith, Hans Uszkoreit: TIGER Annotationsschema (syntax), 2003.
- The header of the XML version of the TIGER Treebank contains lists of various sorts of tags with brief explanation.

Mostly newswire (Frankfurter Rundschau).

According to their website, the TIGER Treebank version 1 contains approximately 700,000 tokens in 40,000 sentences. Version 2.1 contains approximately 900,000 tokens in 50,000 sentences.

The CoNLL 2006 version contains 705,304 tokens in 39573 sentences, yielding 17.82 tokens per sentence on average (CoNLL 2006 data split: 699,610 tokens / 39216 sentences training, 5694 tokens / 357 sentences test).

The CoNLL 2009 version contains 712,332 tokens in 40020 sentences, yielding 17.80 tokens per sentence on average (CoNLL 2009 data split: 648,677 tokens / 36020 sentences training, 32033 tokens / 2000 sentences development, 31622 tokens / 2000 sentences test).

All versions contain semi-automatic part of speech tags (Stuttgart-Tübingen Tagset, STTS) and syntactic structure. Lemmas and morphosyntactic features are available only for newer versions (TIGER Treebank version 2 and onwards, and CoNLL 2009). The parts of speech are heavily context-dependent, e.g. many words can be used both substantively (pronouns) and attributively (determiners), which is distinguished by different POS tags.

It is not clear what the semi-automatic annotation means (probably first auto-tagging, then manual correction?) and whether it also applies to the morphosyntactic annotation. The CoNLL 2009 version also contains automatically disambiguated lemmas, tags and features.

The original treebank is phrase-based. The dependencies in the CoNLL versions must have thus been drawn using a head-selection procedure. Besides CoNLL data, the TIGER project also provides a subset of the TIGER Treebank in a dependency format.

The first sentence of TIGER Treebank 2.1 in the TIGER-XML format:

<s id="s1">
  <graph root="s1_VROOT">
    <terminals>
      <t id="s1_1" word="``" lemma="--" pos="$(" morph="--" case="--" number="--" gender="--" person="--" degree="--" tense="--" mood="--" />
      <t id="s1_2" word="Ross" lemma="Ross" pos="NE" morph="Nom.Sg.Masc" case="Nom" number="Sg" gender="Masc" person="--" degree="--" tense="--" mood="--" />
      <t id="s1_3" word="Perot" lemma="Perot" pos="NE" morph="Nom.Sg.Masc" case="Nom" number="Sg" gender="Masc" person="--" degree="--" tense="--" mood="--" />
      <t id="s1_4" word="wäre" lemma="sein" pos="VAFIN" morph="3.Sg.Past.Subj" case="--" number="Sg" gender="--" person="3" degree="--" tense="Past" mood="Subj" />
      <t id="s1_5" word="vielleicht" lemma="vielleicht" pos="ADV" morph="--" case="--" number="--" gender="--" person="--" degree="--" tense="--" mood="--" />
      <t id="s1_6" word="ein" lemma="ein" pos="ART" morph="Nom.Sg.Masc" case="Nom" number="Sg" gender="Masc" person="--" degree="--" tense="--" mood="--" />
      <t id="s1_7" word="prächtiger" lemma="prächtig" pos="ADJA" morph="Pos.Nom.Sg.Masc" case="Nom" number="Sg" gender="Masc" person="--" degree="Pos" tense="--" mood="--" />
      <t id="s1_8" word="Diktator" lemma="Diktator" pos="NN" morph="Nom.Sg.Masc" case="Nom" number="Sg" gender="Masc" person="--" degree="--" tense="--" mood="--" />
      <t id="s1_9" word="''" lemma="--" pos="$(" morph="--" case="--" number="--" gender="--" person="--" degree="--" tense="--" mood="--" />
    </terminals>
    <nonterminals>
      <nt id="s1_500" cat="PN">
        <edge label="PNC" idref="s1_2" />
        <edge label="PNC" idref="s1_3" />
      </nt>
      <nt id="s1_501" cat="NP">
        <edge label="NK" idref="s1_6" />
        <edge label="NK" idref="s1_7" />
        <edge label="NK" idref="s1_8" />
      </nt>
      <nt id="s1_502" cat="S">
        <edge label="SB" idref="s1_500" />
        <edge label="HD" idref="s1_4" />
        <edge label="MO" idref="s1_5" />
        <edge label="PD" idref="s1_501" />
      </nt>
      <nt id="s1_VROOT" cat="VROOT">
        <edge label="--" idref="s1_1" />
        <edge label="--" idref="s1_502" />
        <edge label="--" idref="s1_9" />
      </nt>
    </nonterminals>
  </graph>
</s>

The first sentence of the CoNLL 2006 training data:

1	``	_	$(	$(	_	4	PUNC	4	PUNC
2	Ross	_	NE	NE	_	4	SB	4	SB
3	Perot	_	NE	NE	_	2	PNC	2	PNC
4	wäre	_	VAFIN	VAFIN	_	0	ROOT	0	ROOT
5	vielleicht	_	ADV	ADV	_	4	MO	4	MO
6	ein	_	ART	ART	_	8	NK	8	NK
7	prächtiger	_	ADJA	ADJA	_	8	NK	8	NK
8	Diktator	_	NN	NN	_	4	PD	4	PD
9	''	_	$(	$(	_	4	PUNC	4	PUNC

The first sentence of the CoNLL 2006 test data:

1	Zwei	_	CARD	CARD	_	2	NK	2	NK
2	Themen	_	NN	NN	_	14	SB	14	SB
3	,	_	$,	$,	_	2	PUNC	2	PUNC
4	die	_	PRELS	PRELS	_	8	OA	8	OA
5	Perot	_	NE	NE	_	8	SB	8	SB
6	immer	_	ADV	ADV	_	7	MO	7	MO
7	wieder	_	ADV	ADV	_	8	MO	8	MO
8	anspricht	_	VVFIN	VVFIN	_	2	RC	2	RC
9	,	_	$,	$,	_	2	PUNC	2	PUNC
10	Rezession	_	NN	NN	_	2	APP	2	APP
11	und	_	KON	KON	_	10	CD	10	CD
12	Bürokratie	_	NN	NN	_	10	CJ	10	CJ
13	,	_	$,	$,	_	14	PUNC	14	PUNC
14	machen	_	VVFIN	VVFIN	_	0	ROOT	0	ROOT
15	ihnen	_	PPER	PPER	_	18	DA	18	DA
16	besonders	_	ADV	ADV	_	18	MO	18	MO
17	zu	_	PTKZU	PTKZU	_	18	PM	18	PM
18	schaffen	_	VVINF	VVINF	_	14	OC	14	OC
19	.	_	$.	$.	_	14	PUNC	14	PUNC

The first sentence of the CoNLL 2009 training data:

1

``

_

``

$(

_

4

PUNC

_

2

Ross

Roß

NE

NN

Nom|Sg|Masc

_

3

PNC

_

3

Perot

NE

Nom|Sg|Masc

_

4

SB

_

4

wäre

sein

VAFIN

3|Sg|Past|Subj

*|Sg|Past|Subj

0

ROOT

_

5

vielleicht

ADV

_

4

MO

_

6

ein

ART

Nom|Sg|Masc

*|Sg|*

8

NK

_

7

prächtiger

prächtig

ADJA

Pos|Nom|Sg|Masc

*|*|*|*

8

NK

_

8

Diktator

NN

Nom|Sg|Masc

*|Sg|Masc

4

PD

_

9

''

_

''

$(

_

4

PUNC

_

The first sentence of the CoNLL 2009 development data:

1

Maschinenbau

NN

Nom|Sg|Masc

*|Sg|Masc

0

4

ROOT

NK

_

2

/

_

/

$(

_

0

1

PUNC

_

3

(

_

(

$(

_

0

4

PUNC

_

4

Zusammenfassung

NN

Nom|Sg|Fem

*|Sg|Fem

0

ROOT

_

5

)

_

)

$(

_

0

1

PUNC

_

The first sentence of the CoNLL 2009 test data:

1

Gegen

gegen

APPR

_

2

eine

ein

ART

Acc|Sg|Fem

*|Sg|Fem

_

3

Erweiterung

NN

Acc|Sg|Fem

*|Sg|Fem

_

4

ihrer

ihr

PPOSAT

Gen|Sg|Fem

*|*|*

_

5

Organisation

NN

Gen|Sg|Fem

*|Sg|Fem

_

6

zu

APPR

_

7

einem

ein

ART

Dat|Sg|Neut

Dat|Sg|*

_

8

sicherheitspolitischen

sicherheitspolitisch

ADJA

Pos|Dat|Sg|Neut

Pos|*|*|*

_

9

Forum

NN

Dat|Sg|Neut

*|Sg|Neut

_

10

sprachen

sprechen

VVFIN

3|Pl|Past|Ind

*|Pl|Past|Ind

_

Y

11

sich

er|es|sie|Sie

PRF

3|Acc|Pl

*|*|*

_

12

die

der

d

ART

Nom|Pl|Masc

*|*|*

_

13

meisten

meister

meist

PIAT

Nom|Pl|Masc

*|*|*

_

14

Staaten

Staat

NN

Nom|Pl|Masc

*|Pl|Masc

_

15

beim

bei

beim

APPRART

Dat|Sg|Neut

Dat|Sg|*

_

16

Gipfeltreffen

NN

Dat|Sg|Neut

*|*|Neut

_

17

für

APPR

_

18

Asiatisch-Pazifische

asiatisch-pazifisch

Asiatisch-Pazifische

ADJA

NN

Pos|Acc|Sg|Fem

*|*|*

_

19

Wirtschaftskooperation

NN

Acc|Sg|Fem

*|Sg|Fem

_

20

(

_

(

$(

_

21

Apec

_

NE

Nom|Sg|Fem

_

22

)

_

)

$(

_

23

in

APPR

_

24

Osaka

NE

Dat|Sg|Neut

*|Sg|Neut

_

25

aus

PTKVZ

_

26

.

_

.

$.

_

TIGER is a mildly nonprojective treebank. 15875 of the 680,710 tokens in the CoNLL 2009 training+development datasets are attached nonprojectively (2.33%).

The results of the CoNLL 2006 shared task are available online. They have been published in (Buchholz and Marsi, 2006). The evaluation procedure was non-standard because it excluded punctuation tokens. These are the best results for German:

Parser (Authors)	LAS	UAS
MST (McDonald et al.)	87.34	90.38
Riedel et al.	86.24	89.76
Basis (O'Neil)	85.36	89.16
Malt (Nivre et al.)	85.82	88.76

The results of the CoNLL 2009 shared task are available online. They have been published in (Hajič et al., 2009). Unlabeled attachment score was not published. These are the best results for German:

Parser (Authors)	LAS
Bohnet	87.48
Merlo	87.29
Chen	86.24
Che	86.19

CoNLL 2007

There does not seem to be any regular distribution channel for the Greek Dependency Treebank. The CoNLL 2007 version had a restricted license for the duration of the shared task only. Republication of the CoNLL version in LDC is planned but it has not happenned yet. In the meantime, one can ask Prokopis Prokopidis (prokopis (at) ilsp (dot) gr) about availability of the corpus.

GDT was created by members of the Institute for Language and Speech Processing (Ινστιτούτο Επεξεργασίας του Λόγου, ILSP/ΙΕΛ), Επιδαύρου & Αρτέμιδος 6, Παράδεισος Αμαρουσίου, GR-15125 Αθήνα, Greece.

Website
- no website dedicated to the treebank
Data
- no separate citation
Principal publications
- Prokopis Prokopidis, Elina Desipri, Maria Koutsombogera, Harris Papageorgiou, Stelios Piperidis: Theoretical and Practical Issues in the Construction of a Greek Dependency Corpus In: Montserrat Civit, Sandra Kübler, Ma. Antònia Martí (eds.), Proceedings of The Fourth Workshop on Treebanks and Linguistic Theories (TLT 2005), pp. 149-160, Barcelona, Spain, 2005.
Documentation
- Description of tags and feature values is provided in the doc/README file in the CoNLL 2007 data distribution.

Mixed (“GDT consists of randomly selected textual fragments and texts in three domains: politics (current affairs, manual transcripts and minutes of European parliamentary sessions), health, and travel.”)

The CoNLL 2007 version contains 70223 tokens in 2902 sentences, yielding 24.20 tokens per sentence on average (CoNLL 2007 data split: 65419 tokens / 2705 sentences training, 4804 tokens / 197 sentences test).

The syntactic annotation style and the tagset for dependency relations (analytical functions) in GDT has been modeled after the Prague Dependency Treebank.

The first sentence of the CoNLL 2007 training data:

1	“	”	PUNCT	PUNCT	_	10	AuxG	_	_
2	Τα	ο	At	AtDf	Ne\|Pl\|Nm	3	Atr	_	_
3	αντισώματα	αντίσωμα	No	NoCm	Ne\|Pl\|Nm	5	Sb	_	_
4	IgG	IgG	Rg	RgFwOr	_	3	Atr	_	_
5	είναι	είμαι	Vb	VbMn	Id\|Pr\|03\|Pl\|Xx\|Ip\|Pv\|Xx	10	Obj_Co	_	_
6	σαν	σαν	Ad	Ad	Ba	5	Adv	_	_
7	μακροπρόθεσμη	μακροπρόθεσμος	Aj	Aj	Ba\|Fe\|Sg\|Nm	8	Atr	_	_
8	μνήμη	μνήμη	No	NoCm	Fe\|Sg\|Nm	6	Adv	_	_
9	,	,	PUNCT	PUNCT	_	10	AuxX	_	_
10	ενώ	ενώ	Cj	CjCo	_	26	Coord	_	_
11	το	ο	At	AtDf	Ne\|Sg\|Nm	12	Atr	_	_
12	IgA	IgA	Rg	RgFwOr	_	15	Sb	_	_
13	πιστεύεται	πιστεύεται	Vb	VbMn	Id\|Pr\|03\|Sg\|Xx\|Ip\|Pv\|Xx	10	Obj_Co	_	_
14	ότι	ότι	Cj	CjSb	_	13	AuxC	_	_
15	είναι	είμαι	Vb	VbMn	Id\|Pr\|03\|Sg\|Xx\|Ip\|Pv\|Xx	14	Sb	_	_
16	ένας	ένας	At	AtId	Ma\|Sg\|Nm	18	Atr	_	_
17	συγκεκριμένος	συγκεκριμένος	Aj	Aj	Ba\|Ma\|Sg\|Nm	18	Atr	_	_
18	δείκτης	δείκτης	No	NoCm	Ma\|Sg\|Nm	15	Pnom	_	_
19	για	για	AsPp	AsPpSp	_	18	AuxP	_	_
20	πρόσφατες	πρόσφατος	Aj	Aj	Ba\|Fe\|Pl\|Ac	21	Atr_Co	_	_
21	ή	ή	Cj	CjCo	_	23	Coord	_	_
22	χρόνιες	χρόνιος	Aj	Aj	Ba\|Fe\|Pl\|Ac	21	Atr_Co	_	_
23	λοιμώξεις	λοίμωξη	No	NoCm	Fe\|Pl\|Ac	19	Atr	_	_
24	“	”	PUNCT	PUNCT	_	10	AuxG	_	_
25	,	,	PUNCT	PUNCT	_	10	AuxX	_	_
26	εξηγεί	εξηγώ	Vb	VbMn	Id\|Pr\|03\|Sg\|Xx\|Ip\|Av\|Xx	0	Pred	_	_
27	η	ο	At	AtDf	Fe\|Sg\|Nm	28	Atr	_	_
28	Δρ	Δρ	Rg	RgFwTr	_	26	Sb	_	_
29	Αρκάρι	Αρκάρι	No	NoCm	Ne\|Sg\|Nm	28	Atr	_	_
30	.	.	PUNCT	PUNCT	_	0	AuxK	_	_

The first sentence of the CoNLL 2007 test data:

1	Η	ο	At	AtDf	Fe\|Sg\|Nm	2	Atr	_	_
2	Σίφνος	Σίφνος	No	NoPr	Fe\|Sg\|Nm	3	Sb	_	_
3	φημίζεται	φημίζομαι	Vb	VbMn	Id\|Pr\|03\|Sg\|Xx\|Ip\|Pv\|Xx	0	Pred	_	_
4	και	και	Cj	CjCo	_	5	AuxY	_	_
5	για	για	AsPp	AsPpSp	_	3	AuxP	_	_
6	τα	ο	At	AtDf	Ne\|Pl\|Ac	8	Atr	_	_
7	καταγάλανα	καταγάλανος	Aj	Aj	Ba\|Ne\|Pl\|Ac	8	Atr	_	_
8	νερά	νερό	No	NoCm	Ne\|Pl\|Ac	5	Obj	_	_
9	των	ο	At	AtDf	Fe\|Pl\|Ge	11	Atr	_	_
10	πανέμορφων	πανέμορφος	Aj	Aj	Ba\|Fe\|Pl\|Ge	11	Atr	_	_
11	ακτών	ακτή	No	NoCm	Fe\|Pl\|Ge	8	Atr	_	_
12	της	μου	Pn	PnPo	Fe\|03\|Sg\|Ge\|Xx	11	Atr	_	_
13	.	.	PUNCT	PUNCT	_	0	AuxK	_	_

Nonprojectivities in GDT are not frequent. Only 823 of the 70223 tokens in the CoNLL 2007 version are attached nonprojectively (1.17%).

The results of the CoNLL 2007 shared task are available online. They have been published in (Nivre et al., 2007). The evaluation procedure was changed to include punctuation tokens. These are the best results for Greek:

Parser (Authors)	LAS	UAS
Nakagawa	76.31	84.08
Keith Hall et al.	74.21	82.04
Carreras	73.56	81.37
Malt (Nilsson et al.)	74.65	81.22
Titov et al.	73.52	81.20
Chen	74.42	81.16
Duan	74.29	80.77
Attardi et al.	73.92	80.75
Malt (J. Hall et al.)	74.21	80.66

The two Malt parser results of 2007 (single malt and blended) are described in (Hall et al., 2007) and the details about the parser configuration are described here.

Penn Treebank 2 (1995)
Penn Treebank 3 (1999)
CoNLL 2007
CoNLL 2008
CoNLL 2009

The original Penn Treebank is distributed by the LDC under the catalogue number LDC99T42. It is free for LDC members 1999, price for non-members is unknown (contact LDC). The license in short:

non-commercial education and research usage
no redistribution
citation in publications not explicitly required but it is common decency

The CoNLL 2007, 2008 and 2009 versions are also licensed by the LDC and LDC members can keep them after the shared task. Those who have not participated in the shared task may inquire at the LDC about the availability of the datasets. Their republication in LDC is planned but it has not happenned yet.

The Penn Treebank was created by members of the Department of Computer and Information Science (CIS), School of Engineering, University of Pennsylvania, Levine Hall, 3330 Walnut Street, Philadelphia, PA 19104-6309, USA. The constituents-to-dependencies CoNLL 2007 conversion of the treebank was prepared by Ryan McDonald.

Website
- http://www.cis.upenn.edu/~treebank/
Data
- Mitchell P. Marcus, Beatrice Santorini, Mary Ann Marcinkiewicz, Ann Taylor: Treebank-3 (LDC99T42). Linguistic Data Consortium, Philadelphia, USA, 2001. ISBN 1-58563-163-9.
Principal publications
- Mitchell P. Marcus, Beatrice Santorini, Mary Ann Marcinkiewicz: Building a large annotated corpus of English: the Penn Treebank. Computational Linguistics, 19(2):313-330. 1993.
Documentation
- Tokenization
- Beatrice Santorini: Part-of-Speech Tagging Guidelines for the Penn Treebank Project, 3rd Revision, Philadelphia, USA, 1990.
- Ann Bies, Mark Ferguson, Karen Katz, Robert MacIntyre: Bracketing Guidelines for Treebank II Style, Penn Treebank Project, Philadelphia, USA, 1995.
- Robert MacIntyre: NP Heads and Base NPs (Treebank FAQ)
- Richard Johansson, Pierre Nugues: Extended constituent-to-dependency conversion for English. In: Proceedings of the 16th Nordic Conference on Computational Linguistics (NODALIDA), pp. 105-112, Tartu, Estonia, 2007.

Financial news from the Wall Street Journal (1989). The constituent-based Treebank-3 also contains parsed versions of ATIS-3 and of the Brown Corpus. Only WSJ texts have been converted to dependencies for the CoNLL shared tasks.

CoNLL 2007: Wall Street Journal part of the Penn Treebank, sections 2-11 used for training, a subset of section 23 for testing.

All distributions of PDT are officially split to training, development (d-test) and test (e-test) data sets. PDT 2.0 contains data that are annotated only morphologically (M-layer), those that are annotated both morphologically and analytically (surface syntax; M+A layers), and the smallest subset is also annotated tectogrammatically (M+A+T layers). The statistics in this section cover the M+A subset, which is relevant for surface dependency parsing.

Size of CoNLL 2007 data was limited because some teams of CoNLL 2006 complained that they did not have enough time and resources to train the larger models. For CoNLL 2009, only that part of PDT was selected that contained also tectogrammatical annotation, because the 2009 task included semantic learning.

Parts of the following table have been taken from (Zeman 2004, page 21). Only non-empty sentences counted (e.g. PDT 1.0 had 81614 sentence tags but only 73088 non-empty ones).

Version	Train Sentences	Train Tokens	D-test Sentences	D-test Tokens	E-test Sentences	E-test Tokens	Total Sentences	Total Tokens	Sentence Length
PDT 0.5	19126	327,597	3697	63718	3787	65390	26610	456,705	17.16
PDT 1.0	73088	1,255,590	7319	126,030	7507	125,713	87914	1,489,748	16.95
PDT 2.0	68562	1,172,299	9270	158,962	10148	173,586	87980	1,504,847	17.10
CoNLL 2006	72703	1,249,408	365	5853			73068	1,255,261	17.18
CoNLL 2007	25364	432,296	286	4724			25650	437,020	17.04
CoNLL 2009	38727	652,544	5228	87988	4213	70348	48168	810,880	16.83

CoNLL 2007: Many function tags were removed from the non-terminals in the phrase-structure representation. The phrase structures were converted to dependency structures using the procedure described in Richard Johansson, Pierre Nugues: Extended constituent-to-dependency conversion for English. In: Proceedings of the 16th Nordic Conference on Computational Linguistics (NODALIDA), pp. 105-112, Tartu, Estonia, 2007.

PDT 1.0 is distributed in the CSTS format. PDT 2.0 uses the PML format. CoNLL 2006 and 2007 uses the CoNLL-X format; CoNLL 2009 format is slightly different (number and meaning of columns). Unlike the other formats, the CSTS format used the ISO-8859-2 character encoding.

The CSTS format (PDT 0.5 and 1.0) contains morphological annotation (lemmas and tags) both manual and by two taggers. The CoNLL 2009 version contains manual and one automatic disambiguation. The official distribution of PDT 2.0 and the CoNLL 2006 and 2007 versions contain only manual morphology.

The original PDT uses 15-character positional morphological tags. The CoNLL versions convert the tags to the two/three CoNLL columns, CPOS, POS and FEAT. In addition, the CoNLL versions contain the Sem feature, which is derived from the tags attached to lemma in PDT (see Hana and Zeman, 2005).

See above for documentation of the morphological tags. All CoNLL distributions contain a README file with a brief description of the parts of speech and features. Use DZ Interset to inspect the PDT and the CoNLL tagsets.

The guidelines for syntactic annotation are documented in the PDT annotation manual.

The first sentence of the PDT 1.0 training data:

<csts lang=cs>
<h>
<source>Českomoravský profit</source>
<markup>
<mauth>js
<mdate>1996-2000
<mdesc>Manual analytical annotation
</markup>
<markup>
<mauth>kk,lk
<mdate>1996-2000
<mdesc>Manual morphological annotation
</markup>
</h>
<doc file="s/inf/j/1994/cmpr9406" id="001">
<a>
<mod>s
<txtype>inf
<genre>mix
<med>j
<temp>1994
<authname>y
<opus>cmpr9406
<id>001
</a>
<c>
<p n=1>
<s id="cmpr9406:001-p1s1">
<p n=2>
<s id="cmpr9406:001-p2s1">
<f cap>Třikrát<l>třikrát`3<t>Cv-------------<MDl src="a">třikrát`3<MDt src="a">Cv-------------<MDl src="b">třikrát`3<MDt src="b">Cv-------------<A>Adv<r>1<g>2
<f>rychlejší<l>rychlý<t>AAFS1----2A----<MDl src="a">rychlý<MDt src="a">AANS1----2A----<MDl src="b">rychlý<MDt src="b">AAFS1----2A----<A>ExD<r>2<g>0
<f>než<l>než-2<t>J,-------------<MDl src="a">než-2<MDt src="a">J,-------------<MDl src="b">než-2<MDt src="b">J,-------------<A>AuxC<r>3<g>2
<f>slovo<l>slovo<t>NNNS1-----A----<MDl src="a">slovo<MDt src="a">NNNS4-----A----<MDl src="b">slovo<MDt src="b">NNNS1-----A----<A>ExD<r>4<g>3

The first two sentences of the PDT 1.0 d-test data:

<csts lang=cs>
<h>
<source>Lidové noviny</source>
<markup>
<mauth>zu
<mdate>1996-2000
<mdesc>Manual analytical annotation
</markup>
</h>
<doc file="s/pub/nws/1994/ln94206" id="1">
<a>
<mod>s
<txtype>pub
<genre>mix
<med>nws
<temp>1994
<authname>y
<opus>ln94206
<id>1
</a>
<c>
<p n=1>
<s id="ln94206:1-p1s1">
<i>ti
<f cap>Lidé<MDl src="a">člověk<MDt src="a">NNMP1-----A---1<MDl src="b">člověk<MDt src="b">NNMP1-----A---1<A>ExD<r>1<g>0
<p n=2>
<s id="ln94206:1-p2s1">
<f upper.abbr>ING<MDl src="a">Ing-1_:B_^(inženýr)<MDt src="a">NNMXX-----A---8<MDl src="b">Ing-1_:B_^(inženýr)<MDt src="b">NNMXX-----A---8<A>Atr<r>1<g>4
<D>
<d>.<MDl src="a">.<MDt src="a">Z:-------------<MDl src="b">.<MDt src="b">Z:-------------<A>AuxG<r>2<g>1
<f upper>PETR<MDl src="a">Petr_;Y<MDt src="a">NNMS1-----A----<MDl src="b">Petr_;Y<MDt src="b">NNMS1-----A----<A>Atr<r>3<g>4
<f upper>KARAS<MDl src="a">karas<MDt src="a">NNMS1-----A----<MDl src="b">karas<MDt src="b">NNMS1-----A----<A>Sb_Ap<r>4<g>11
<D>
<d>,<MDl src="a">,<MDt src="a">Z:-------------<MDl src="b">,<MDt src="b">Z:-------------<A>AuxX<r>5<g>6
<f mixed>CSc<MDl src="a">CSc-1_:B_^(kandidát_věd)<MDt src="a">NNMXX-----A---8<MDl src="b">CSc-1_:B_^(kandidát_věd)<MDt src="b">NNMXX-----A---8<A>Atr<r>6<g>4
<D>
<d>.<MDl src="a">.<MDt src="a">Z:-------------<MDl src="b">.<MDt src="b">Z:-------------<A>AuxG<r>7<g>6
<d>(<MDl src="a">(<MDt src="a">Z:-------------<MDl src="b">(<MDt src="b">Z:-------------<A>ExD<r>8<g>4
<D>
<f num>53<MDl src="a">53<MDt src="a">C=-------------<MDl src="b">53<MDt src="b">C=-------------<A>ExD_Pa<r>9<g>4
<D>
<d>)<MDl src="a">)<MDt src="a">Z:-------------<MDl src="b">)<MDt src="b">Z:-------------<A>ExD<r>10<g>4
<D>
<d>,<MDl src="a">,<MDt src="a">Z:-------------<MDl src="b">,<MDt src="b">Z:-------------<A>Apos<r>11<g>20
<f>generální<MDl src="a">generální<MDt src="a">AAMS1----1A----<MDl src="b">generální<MDt src="b">AAMS1----1A----<A>Atr<r>12<g>13
<f>ředitel<MDl src="a">ředitel<MDt src="a">NNMS1-----A----<MDl src="b">ředitel<MDt src="b">NNMS1-----A----<A>Sb_Co<r>13<g>15
<f upper>ČEZ<MDl src="a">ČEZ-1_:B_;K_^(České_energetické_závody)<MDt src="a">NNIPX-----A---8<MDl src="b">ČEZ-1_:B_;K_^(České_energetické_závody)<MDt src="b">NNIPX-----A---8<A>Atr<r>14<g>13
<f>a<MDl src="a">a-1<MDt src="a">J^-------------<MDl src="b">a-1<MDt src="b">J^-------------<A>Coord_Ap<r>15<g>11
<f>předseda<MDl src="a">předseda<MDt src="a">NNMS1-----A----<MDl src="b">předseda<MDt src="b">NNMS1-----A----<A>Sb_Co<r>16<g>15
<f>jeho<MDl src="a">jeho_^(přivlast.)<MDt src="a">PSXXXZS3-------<MDl src="b">jeho_^(přivlast.)<MDt src="b">PSXXXZS3-------<A>Atr<r>17<g>18
<f>představenstva<MDl src="a">představenstvo<MDt src="a">NNNS2-----A----<MDl src="b">představenstvo<MDt src="b">NNNS2-----A----<A>Atr<r>18<g>16
<D>
<d>,<MDl src="a">,<MDt src="a">Z:-------------<MDl src="b">,<MDt src="b">Z:-------------<A>AuxX<r>19<g>11
<f>je<MDl src="a">být<MDt src="a">VB-S---3P-AA---<MDl src="b">být<MDt src="b">VB-S---3P-AA---<A>Pred<r>20<g>0
<f>absolventem<MDl src="a">absolvent<MDt src="a">NNMS7-----A----<MDl src="b">absolvent<MDt src="b">NNMS7-----A----<A>Pnom<r>21<g>20
<f>elektrotechnické<MDl src="a">elektrotechnický<MDt src="a">AAFS2----1A----<MDl src="b">elektrotechnický<MDt src="b">AAFS2----1A----<A>Atr<r>22<g>23
<f>fakulty<MDl src="a">fakulta<MDt src="a">NNFS2-----A----<MDl src="b">fakulta<MDt src="b">NNFS2-----A----<A>Atr_Co<r>23<g>25
<f upper>ČVUT<MDl src="a">ČVUT-1_:B_;K_^(České_vysoké_učení_technické)<MDt src="a">NNNXX-----A---8<MDl src="b">ČVUT-1_:B_;K_^(České_vysoké_učení_technické)<MDt src="b">NNNXX-----A---8<A>Atr<r>24<g>23
<f>a<MDl src="a">a-1<MDt src="a">J^-------------<MDl src="b">a-1<MDt src="b">J^-------------<A>Coord<r>25<g>21
<f>postgraduálního<MDl src="a">postgraduální<MDt src="a">AANS2----1A----<MDl src="b">postgraduální<MDt src="b">AANS2----1A----<A>Atr<r>26<g>27
<f>studia<MDl src="a">studium<MDt src="a">NNNS2-----A----<MDl src="b">studium<MDt src="b">NNNS2-----A----<A>Atr_Co<r>27<g>25
<f>v<MDl src="a">v-1<MDt src="a">RR--6----------<MDl src="b">v-1<MDt src="b">RR--6----------<A>AuxP<r>28<g>29
<f>oboru<MDl src="a">obor_^(lidské_činnosti)<MDt src="a">NNIS6-----A----<MDl src="b">obor_^(lidské_činnosti)<MDt src="b">NNIS6-----A----<A>AuxP<r>29<g>27
<f>metod<MDl src="a">metoda<MDt src="a">NNFP2-----A----<MDl src="b">metoda<MDt src="b">NNFP2-----A----<A>Atr<r>30<g>29
<f>operační<MDl src="a">operační<MDt src="a">AAFS2----1A----<MDl src="b">operační<MDt src="b">AAFS2----1A----<A>Atr<r>31<g>32
<f>analýzy<MDl src="a">analýza<MDt src="a">NNFS2-----A----<MDl src="b">analýza<MDt src="b">NNFS2-----A----<A>Atr<r>32<g>30
<D>
<d>.<MDl src="a">.<MDt src="a">Z:-------------<MDl src="b">.<MDt src="b">Z:-------------<A>AuxK<r>33<g>0

The first sentence of the PDT 1.0 e-test data:

<csts lang=cs>
<h>
<source>Lidové noviny</source>
<markup>
<mauth>zu
<mdate>1996-2000
<mdesc>Manual analytical annotation
</markup>
</h>
<doc file="s/pub/nws/1994/ln94209" id="1">
<a>
<mod>s
<txtype>pub
<genre>mix
<med>nws
<temp>1994
<authname>y
<opus>ln94209
<id>1
</a>
<c>
<p n=1>
<s id="ln94209:1-p1s1">
<f cap>Přádelny<MDl src="a">přádelna<MDt src="a">NNFP1-----A----<MDl src="b">přádelna<MDt src="b">NNFP1-----A----<A>Sb<r>1<g>2
<f>mají<MDl src="a">mít<MDt src="a">VB-P---3P-AA---<MDl src="b">mít<MDt src="b">VB-P---3P-AA---<A>Pred<r>2<g>0
<f>dvojnásob<MDl src="a">dvojnásob<MDt src="a">Db-------------<MDl src="b">dvojnásob<MDt src="b">Db-------------<A>Obj<r>3<g>2
<f>vad<MDl src="a">vada<MDt src="a">NNFP2-----A----<MDl src="b">vada<MDt src="b">NNFP2-----A----<A>Atr<r>4<g>3

Morphological annotation of the first amw training file of the PDT 2.0:

<mdata xmlns="http://ufal.mff.cuni.cz/pdt/pml/">
 <head>
  <schema href="mdata_schema.xml" />
  <references>
   <reffile id="w" name="wdata" href="cmpr9406_001.w.gz" />
  </references>
 </head>
 <meta>
  <lang>cs</lang>
  <annotation_info id="manual">
   <desc>Manual annotation</desc>
  </annotation_info>
 </meta>
 <s id="m-cmpr9406-001-p2s1">
  <m id="m-cmpr9406-001-p2s1w1">
   <src.rf>manual</src.rf>
   <w.rf>w#w-cmpr9406-001-p2s1w1</w.rf>
   <form>Třikrát</form>
   <lemma>třikrát`3</lemma>
   <tag>Cv-------------</tag>
  </m>
  <m id="m-cmpr9406-001-p2s1w2">
   <src.rf>manual</src.rf>
   <w.rf>w#w-cmpr9406-001-p2s1w2</w.rf>
   <form>rychlejší</form>
   <lemma>rychlý</lemma>
   <tag>AAFS1----2A----</tag>
  </m>
  <m id="m-cmpr9406-001-p2s1w3">
   <src.rf>manual</src.rf>
   <w.rf>w#w-cmpr9406-001-p2s1w3</w.rf>
   <form>než</form>
   <lemma>než-2</lemma>
   <tag>J,-------------</tag>
  </m>
  <m id="m-cmpr9406-001-p2s1w4">
   <src.rf>manual</src.rf>
   <w.rf>w#w-cmpr9406-001-p2s1w4</w.rf>
   <form>slovo</form>
   <lemma>slovo</lemma>
   <tag>NNNS1-----A----</tag>
  </m>
 </s>

Analytical (surface-syntactic) annotation of the first amw training file of the PDT 2.0:

<adata xmlns="http://ufal.mff.cuni.cz/pdt/pml/">
 <head>
  <schema href="adata_schema.xml" />
  <references>
   <reffile id="m" name="mdata" href="cmpr9406_001.m.gz" />
   <reffile id="w" name="wdata" href="cmpr9406_001.w.gz" />
  </references>
 </head>
 <meta>
  <annotation_info>
   <desc>Manual annotation</desc>
  </annotation_info>
 </meta>
 <trees>
  <LM id="a-cmpr9406-001-p2s1">
   <s.rf>m#m-cmpr9406-001-p2s1</s.rf>
   <ord>0</ord>
   <children>
    <LM id="a-cmpr9406-001-p2s1w2">
     <m.rf>m#m-cmpr9406-001-p2s1w2</m.rf>
     <afun>ExD</afun>
     <ord>2</ord>
     <children>
      <LM id="a-cmpr9406-001-p2s1w1">
       <m.rf>m#m-cmpr9406-001-p2s1w1</m.rf>
       <afun>Adv</afun>
       <ord>1</ord>
      </LM>
      <LM id="a-cmpr9406-001-p2s1w3">
       <m.rf>m#m-cmpr9406-001-p2s1w3</m.rf>
       <afun>AuxC</afun>
       <ord>3</ord>
       <children>
        <LM id="a-cmpr9406-001-p2s1w4">
         <m.rf>m#m-cmpr9406-001-p2s1w4</m.rf>
         <afun>ExD</afun>
         <ord>4</ord>
        </LM>
       </children>
      </LM>
     </children>
    </LM>
   </children>
  </LM>

The first two sentences of the CoNLL 2006 and 2007 training data:

1	Třikrát	třikrát`3	C	v	_	2	Adv	_	_
2	rychlejší	rychlý	A	A	Gen=F\|Num=S\|Cas=1\|Gra=2\|Neg=A	0	ExD	_	_
3	než	než-2	J	,	_	2	AuxC	_	_
4	slovo	slovo	N	N	Gen=N\|Num=S\|Cas=1\|Neg=A	3	ExD	_	_

1	Faxu	fax	N	N	Gen=I\|Num=S\|Cas=3\|Neg=A	2	Obj	_	_
2	škodí	škodit	V	B	Num=P\|Per=3\|Ten=P\|Neg=A\|Voi=A	0	Pred	_	_
3	především	především	D	b	_	6	AuxZ	_	_
4	přetížené	přetížený	A	A	Gen=F\|Num=P\|Cas=1\|Gra=1\|Neg=A	6	Atr	_	_
5	telefonní	telefonní	A	A	Gen=F\|Num=P\|Cas=1\|Gra=1\|Neg=A	6	Atr	_	_
6	linky	linka	N	N	Gen=F\|Num=P\|Cas=1\|Neg=A	2	Sb	_	_
7	*	*	Z	:	_	2	AuxG	_	_

The first sentence of the CoNLL 2006 test data:

1	Podobně	podobně	D	g	Gra=1\|Neg=A	5	Adv	_	_
2	,	,	Z	:	_	3	AuxX	_	_
3	myslím	myslit	V	B	Num=S\|Per=1\|Ten=P\|Neg=A\|Voi=A	5	Pred_Pa	_	_
4	,	,	Z	:	_	3	AuxX	_	_
5	postupuje	postupovat	V	B	Num=S\|Per=3\|Ten=P\|Neg=A\|Voi=A	0	Pred	_	_
6	většina	většina	N	N	Gen=F\|Num=S\|Cas=1\|Neg=A	5	Sb	_	_
7	českých	český	A	A	Gen=F\|Num=P\|Cas=2\|Gra=1\|Neg=A	8	Atr	_	_
8	bank	banka	N	N	Gen=F\|Num=P\|Cas=2\|Neg=A	6	Atr	_	_
9	,	,	Z	:	_	11	AuxX	_	_
10	zejména	zejména	D	b	_	12	AuxZ	_	_
11	v	v-1	R	R	Cas=6	5	AuxP	_	_
12	případech	případ	N	N	Gen=I\|Num=P\|Cas=6\|Neg=A	11	Adv	_	_
13	,	,	Z	:	_	17	AuxX	_	_
14	kdy	kdy	D	b	_	17	Adv	_	_
15	by	být	V	c	Num=X\|Per=3	17	AuxV	_	_
16	se	se	P	7	Num=X\|Cas=4	18	AuxT	_	_
17	mělo	mít	V	p	Gen=N\|Num=S\|Per=X\|Ten=R\|Neg=A\|Voi=A	12	Atr	_	_
18	jednat	jednat	V	f	Neg=A	17	Obj	_	_
19	o	o-1	R	R	Cas=4	18	AuxP	_	_
20	větší	velký	A	A	Gen=F\|Num=P\|Cas=4\|Gra=2\|Neg=A	21	Atr	_	_
21	částky	částka	N	N	Gen=F\|Num=P\|Cas=4\|Neg=A	19	Obj	_	_
22	.	.	Z	:	_	0	AuxK	_	_

The first sentence of the CoNLL 2007 test data:

1	Proč	proč	D	b	_	2	Adv	_	_
2	mají	mít	V	B	Num=P\|Per=3\|Ten=P\|Neg=A\|Voi=A	0	Pred	_	_
3	každý	každý	A	A	Gen=I\|Num=S\|Cas=4\|Gra=1\|Neg=A	4	Atr	_	_
4	rok	rok	N	N	Gen=I\|Num=S\|Cas=4\|Neg=A	5	Adv	_	_
5	fasovat	fasovat	V	f	Neg=A	2	Obj	_	_
6	speciální	speciální	A	A	Gen=F\|Num=S\|Cas=4\|Gra=1\|Neg=A	7	Atr	_	_
7	taxu	taxa	N	N	Gen=F\|Num=S\|Cas=4\|Neg=A	5	Obj	_	_
8	na	na	R	R	Cas=4	7	AuxP	_	_
9	oblečení	oblečení	N	N	Gen=N\|Num=S\|Cas=4\|Neg=A	8	AtrAdv	_	_
10	?	?	Z	:	_	0	AuxK	_	_

The first sentence of the CoNLL 2009 training data:

1

Celní

celní

A

2

Atr

Y

celní

_

RSTR

_

2

unie

N

SubPOS=N|Gen=F|Num=S|Cas=1|Neg=A

0

ExD

Y

unie

_

3

v

R

SubPOS=R|Cas=6

2

AuxP

_

4

ohrožení

N

SubPOS=N|Gen=N|Num=S|Cas=6|Neg=A

3

Atr

Y

v-w3017f1

_

The first sentence of the CoNLL 2009 development data:

1

|

Z

SubPOS=:

0

3

ExD

AuxG

_

2

Daňový

daňový

A

3

Atr

Y

daňový

_

RSTR

3

poradce

N

SubPOS=N|Gen=M|Num=S|Cas=1|Neg=A

0

ExD

Y

poradce

_

4

|

Z

SubPOS=:

0

3

AuxK

AuxG

_

The first sentence of the CoNLL 2009 test data:

1

Názor

názor

N

SubPOS=N|Gen=I|Num=S|Cas=1|Neg=A

_

Y

2

experta

expert

N

SubPOS=N|Gen=M|Num=S|Cas=2|Neg=A

_

Y

PDT is a mildly nonprojective treebank. 8351 of the 437,020 tokens in the CoNLL 2007 version are attached nonprojectively (1.91%).

There is an online summary of known results in Czech parsing.

The results of the CoNLL 2006 shared task are available online. They have been published in (Buchholz and Marsi, 2006). The evaluation procedure was non-standard because it excluded punctuation tokens. These are the best results for Czech:

Parser (Authors)	LAS	UAS
MST (McDonald et al.)	80.18	87.30
Basis (O'Neil)	76.60	85.58
Malt (Nivre et al.)	78.42	84.80
Nara (Yuchang Cheng)	76.24	83.40

The results of the CoNLL 2007 shared task are available online. They have been published in (Nivre et al., 2007). The evaluation procedure was changed to include punctuation tokens. These are the best results for Czech:

Parser (Authors)	LAS	UAS
Nakagawa	80.19	86.28
Carreras	78.60	85.16
Titov et al.	77.94	84.19
Malt (Nilsson et al.)	77.98	83.59
Attardi et al.	77.37	83.40
Malt (Hall et al.)	77.22	82.35

The two Malt parser results of 2007 (single malt and blended) are described in (Hall et al., 2007) and the details about the parser configuration are described here.

The results of the CoNLL 2009 shared task are available online. They have been published in (Hajič et al., 2009). Unlabeled attachment score was not published. These are the best results for Czech:

Parser (Authors)	LAS
Merlo (Gesmundo et al.)	80.38
Bohnet	80.11
Che et al.	80.01

Institute of Formal and Applied Linguistics Wiki

Table of Contents

Treebanks for Various Languages

Arabic (ar)

Versions

Obtaining and License

References

Domain

Size

Inside

Sample

Parsing

Bulgarian (bg)

Versions

Obtaining and License

References

Domain

Size

Inside

Sample

Parsing

Bengali (bn)

Versions

Obtaining and License

References

Domain

Size

Inside

Sample

Parsing

Catalan (ca)

Versions

Obtaining and License

References

Domain

Size

Inside

Sample

Parsing

Czech (cs)

Versions

Obtaining and License

References

Domain

Size

Inside

Sample

Parsing

Danish (da)

Versions

Obtaining and License

References

Domain

Size

Inside

Sample

Parsing

German (de)

Versions

Obtaining and License

References

Domain

Size

Inside

Sample

Parsing

Greek (el)

Versions

Obtaining and License

References

Domain

Size

Inside

Sample

Parsing

English (en)

Versions

Obtaining and License

References

Domain