ИСТИНА

Creating a corpus annotated for coreferential relations for the Russian language is required for both empirical studies of anaphora and coreference resolution and the needs of computational linguistics. Although large-scale annotated corpora and annotation schemes for coreference are available, these resources are still limited to the major European languages. In our study we present the first open source corpus for the Russian language manually annotated for coreference and introduce the annotation guidelines and the software to perform the annotations created at the Moscow State University in 2013/2014. This corpus was developed as a set for the coreference resolution evaluation campaign Ru-eval-2014 (Russian forum for evaluation of Russian NLP resources http://www.dialog-21.ru/en/parser/).