Distributional models in the task of hypernym discovery

An approach to the solution of the first task of automatically taxonomy construction for the Russian language is described. This task consists in matching unknown input-words with hypernyms from the existing taxonomy. We show that useful results can be attained using pre-trained distribution models without additional training. © Springer Nature Switzerland AG 2020.

Авторы
Yadrintsev V. 1, 2 , Ryzhova A.3 , Sochenkov I. 2
Язык
Английский
Страницы
338-350
Статус
Опубликовано
Том
12412 LNAI
Год
2020
Организации
  • 1 Peoples Friendship University of Russia (RUDN University), Moscow, Russian Federation
  • 2 Federal Research Center “Computer Science and Control” of the Russian Academy of Sciences, Moscow, Russian Federation
  • 3 Skolkovo Institute of Science and Technology, Moscow, Russian Federation
Ключевые слова
BERT; Distributional models; ELMO; fastText; Hypernym discovery; Rusvectores; RuWordNet
Дата создания
02.11.2020
Дата изменения
02.11.2020
Постоянная ссылка
https://repository.rudn.ru/ru/records/article/record/65462/
Поделиться

Другие записи

Grebennikov Valery V., Dubrovin Vladimir V., Zhuravlev Valerii A., Kalinovskaya Victoria S., Shiryov Denis A.
Вопросы истории. Rossiiskaya Akademiya Nauk, Institut Istorii (Russian Academy of Sciences, Institute of General Hist. Том 2020. 2020. С. 255-260