Prague Dependency Treebank

Abstract

We present a richly annotated and genre-diversified language resource, the Prague Dependency Treebank-Consolidated 1.0 (PDT-C 1.0), the purpose of which is -as it always been the case for the family of the Prague Dependency Treebanks -to serve both as a training data for various types of NLP tasks as well as for linguistically-oriented research. PDT-C 1.0 contains four different datasets of Czech, uniformly annotated using the standard PDT scheme (albeit not everything is annotated manually, as we describe in detail here). The texts come from different sources: daily newspaper articles, Czech translation of the Wall Street Journal, transcribed dialogs and a small amount of user-generated, short, often non-standard language segments typed into a web translator. Altogether, the treebank contains around 180,000 sentences with their morphological, surface and deep syntactic annotation. The diversity of the texts and annotations should serve well the NLP applications as well as it is an invaluable resource for linguistic research, including comparative studies regarding texts of different genres. The corpus is publicly and freely available.

Are crossing dependencies really scarce?

“…Our treebanks come from version 2.0 of the HamleDT collection of treebanks [32,33]. This collection harmonizes previously existing treebanks for 30 different languages into two widely-used annotation guidelines: Universal Stanford dependencies [34] and Prague dependencies [35]. Therefore, this resource allows us to evaluate the baselines not only across a wide range of languages of different families, but also across two well-known annotation schemes.…”

Subcategorization of Adverbial Meanings Based on Corpus Data

“…While many Czech corpora has morphological annotation (done automatically), we have to take into account the syntax. Nowadays, several richly syntactically annotated corpora, collectively called Prague Dependency Treebanks (PDTs in the sequel; [9]), have been already developed. These corpora provide a large amount of valuable examples that are used as a basis for the determination of subcategorized meanings of adverbials.…”

“…The main features of the annotation style are: -well-developed dependency syntax theory which is known as the functional Generative Description (fGD in the sequel; see [22], [24], [25]), -interlinked hierarchical layers of standoff annotation, -deep syntactic layer. In the years 1996 through 2005, the first Prague Dependency treebank 1 (PDT in the sequel; [11], for the latest version 3.0 see [2]) was designed and built. The data in PDT are composed by articles from the Czech daily newspapers.…”

PAWS: A Multi-lingual Parallel Treebank with Anaphoric Relations

“…As two of the languages extensively use zero subjects, we could miss a lot of valuable information if we annotated coreference only on sur-face. Therefore, we adopted the style based on the theory of Functional Generative Description (Sgall et al, 1986), first used for Czech in Prague Dependency Treebank 2.0 (Hajič et al, 2006) and for Czech and English in Prague Czech-English Dependency Treebank 2.0 (Hajič et al, 2012). In this style, coreference and other anaphoric relations are annotated on the layer of deep syntax called tectogrammatical layer.…”