TREEBANK DESCRIPTION
The original treebank is HPSG-based. It consists of 215 000 tokens (over 15 000 sentences). It consists of two parts: core set, which is a test suite for Bulgarian, consisting of 1500 sentences, extracted from Bulgarian grammars; remaining set, which comprizes isolated random sentences and running texts from Bulgarian newspapers, literature and administrative documents. All the texts were processed automatically at tokenization, morphological and chunk level. Then, the full syntactic analysis were perfomed manually by trained annotators.
The BTB-UD contains in its third release 11 138 sentences which exclude ellipses and some rare phenomena. The conversion was done semi-automatically, with the application of set of rules and constraints for result consistency.
In the third release we did not add new sentences. We have improved the annotation through checking and correcting the following relations:
- appositions (the wrongly annotated vocatives and appositions were corrected as nmod)
- xcomp (the ccomp relations were checked for control structures and re-annotated as xcomp)
- nsubjpass and csubjpass in se-passives were annotated
- cop relation direction was corrected
- the relation between nouns and short possessive pronouns was re-annotated from case to det
We also have changed the feature VerbForm=Fin to VerbForm=Part for participles.
The rest of the sentences will be converted for the next release.
SPLIT DATA DESCRIPTION
The files of the original treebank were concatenated in one file. Then:
- each tenth sentence was selected for the development set
- each first sentence was selected for the test part
- the rest sentences are in the training set
Remark: Each tenth sentence means the following: when the ID of the sentence is divided by 10, the reminder is 0. Each first sentence means the following: when the ID of the sentence is divided by 10, the reminder is 1.
STATISTICS
Number of Sentences: 11 138 Number of Tokens: 156 323
ACKNOWLEDGEMENTS AND REFERENCES
Acknowledgements
The original treebank was developed in a project (2001-2004), funded by the Volkswagen Stiftung, Federal Republic of Germany under the Programme "Cooperation with Natural and Engineering Scientists in Central and Eastern Europe". The project was carried out mainly at IICT-BAS in tight cooperation with researchers at the Seminar für Sprachwissenschaft (SfS), Eberhard-Karls-Universität, Tübingen, Germany. Link: http://bultreebank.org/ The conversion of BulTreeBank into Universla Dependency format was supported by the EU Project QTLeap. Link: http://qtleap.eu/
References
Petya Osenova and Kiril Simov. BTB-TR05: BulTreeBank Stylebook. BulTreeBank Project Technical Report № 05. 2004. Link: http://www.bultreebank.org/TechRep/BTB-TR05.pdf
Kiril Simov and Petya Osenova. Practical Annotation Scheme for an HPSG Treebank of Bulgarian. In: Proc. of the 4th International Workshop on Linguistically Interpreteted Corpora (LINC-2003), Budapest, Hungary. 2003.
Kiril Simov, Gergana Popova, Petya Osenova. HPSG-based syntactic treebank of Bulgarian (BulTreeBank). In: "A Rainbow of Corpora: Corpus Linguistics and the Languages of the World", edited by Andrew Wilson, Paul Rayson, and Tony McEnery; Lincom-Europa, Munich 2002, pp. 135-142.
Kiril Simov, Petya Osenova and Milena Slavcheva. BTB-TR03: BulTreeBank Morphosyntactic Tagset. BulTreeBank Project Technical Report № 03. 2004
Kiril Simov, Petya Osenova, Alexander Simov, Milen Kouylekov. Design and Implementation of the Bulgarian HPSG-based Treebank. In Erhard Hinrichs and Kiril Simov, editors, Journal of Research on Language and Computation, Special Issue, Kluwer Academic Publishers, pp. 495-522.
Documentation status: partial Data source: semi-automatic Data available since: UD v1.1 License: CC BY-NC-SA 3.0 Genre: news legal fiction misc Contributors: Simov, Kiril; Osenova, Petya