NEREL: A Russian Dataset with Nested Named Entities, Relations and Events

Abstract

In this paper, we present NEREL, a Russian dataset for named entity recognition and relation extraction. NEREL is significantly larger than existing Russian datasets: to date it contains 56K annotated named entities and 39K annotated relations. Its important difference from previous datasets is annotation of nested named entities, as well as relations within nested entities and at the discourse level. NEREL can facilitate development of novel models that can extract relations between nested named entities, as well as relations on both sentence and document levels. NEREL also contains the annotation of events involving named entities and their roles in the events. The NEREL collection is available via https://github.com/nerel-ds/NEREL.

RuNNE-2022 Shared Task: Recognizing Nested Named Entities

“…Так, например в сущность класса Organization "Московский драматический театр имени М. Н. Ермоловой" вложена сущность типа Person -"М. Н. Ермоловой". Соревнование проводится на материале корпуса NEREL (Loukachevitch et al, 2021), собранного из новостных текстов WikiNews на русском языке. В корпусе NEREL представлено 29 классов различных сущностей, а глубина вложенности сущностей достигает 6 уровней разметки.…”

“…This limits its effectiveness in building large neural network models. Additional descriptions and statistics of Russian NER datasets can be found in (Loukachevitch et al, 2021).…”

“…The NEREL collection is developed for studying three levels of information extraction methods including named entity recognition, relation extraction and entity linking (Loukachevitch et al, 2021). Currently, NEREL is the largest Russian dataset annotated with entities and relations compared to the existing Russian datasets.…”

“…The best performing methods for nested NER utilize specialized neural network architectures (Straková et al, 2019;Yu et al, 2020;Jue et al, 2020). A recent dataset for nested NER in Russian, NEREL, is annotated with 29 entity types (Loukachevitch et al, 2021).…”

“…This way, the named entity "The Yermolova Theatre" of type ORGANIZATION houses another entity "Yermolova" of type PERSON. We adopt the Russian NEREL dataset (Loukachevitch et al, 2021) for the RuNNE Shared Task. NEREL comprises news texts written in the Russian language and collected from the Wikinews portal.…”