CoBaLD Annotation Project
CoBaLD Annotation Project
CoBaLD Annotation Project is aimed at elaborating full three-level linguistic annotation standard, which includes morphological, syntactic and semantic markup, the creation of text corpora annotated according to the standard, and the development of neural network-based parsers.
The morphosyntax level relies on the Universal Dependencies (UD) markup – currently, the most popular scheme of morpho-syntactic annotation. UD uses the CONLL (or CONLL Plus) format, which makes it convenient for scripting and automatic parsing purposes. The semantic pattern is based on the Compreno model, which provides the annotation of word meanings and the relations between words. Namely, CoBaLD (Compreno-Based Linguistic Data) Annotation is a kind of the UD markup enriched with the Compreno semantic pattern.
CoBaLD Annotation offers comprehensive linguistic markup, combining UD-based morphosyntax with enriched semantic patterns from Compreno. It annotates word meanings and their interrelations, providing a robust tool for detailed language analysis.
As in UD, Cobald annotation is presented in the CONLL-U Plus file standard, which looks like a table, where each column corresponds to some annotated meaning and each row – to the annotated token. As the key purpose of the CoBaLD standard is to add the semantic domain to the UD/E-UD morphosyntax, it just adds two new columns with the semantic information in the CONLL-U Plus, that is, a column for the word meanings and a column for the semantic relations of the words:
The project includes:
The standard is ready to use, nevertheless, we still work on its improving – restore the elided nodes, discuss copula constructions, experiment with using other ontologies aimed at different tasks, and so on.
There are two parsers for CoBaLD annotation: one for the annotation according to the Base UD rules, and the other – for the annotation according to the Enhanced UD rules.
These are the shared tasks aimed at the creation of the automatic semantic markup in order to investigate the capabilities of the format and to elaborate better tools for annotating in CoBaLD.
M Petrova, A Ivoylova, I Bayuk, D Dyachkova, and M Michurina. 2023. The CoBaLD Annotation Project: the Creation and Application of the full Morpho-Syntactic and Semantic Markup Standard. // International Conference on Computational Linguistics and Intellectual Technologies «Dialog»
A Ivoylova, D Dyachkova, M Petrova, and M Michurina. 2023. The problem of linguistic markup conversion: the transformation of the compreno markup into the ud format. // International Conference on Computational Linguistics and Intellectual Technologies «Dialog»
Navigation
All-in-one platform
Datasets, parsers and competitions for CoBaLD.
