|
ИСТИНА |
Войти в систему Регистрация |
ИПМех РАН |
||
Machine Reading Comprehension Model in Domain-Transfer Taskстатья
Статья опубликована в журнале из списка RSCI Web of Science
Статья опубликована в журнале из перечня ВАКСтатья опубликована в журнале из списка Web of Science и/или Scopus
- Авторы: Rozhkov I.S., Loukachevitch N.V.
- Журнал: Lobachevskii Journal of Mathematics
- Том: 44
- Номер: 8
- Год издания: 2023
- Издательство: Kazanskii Gosudarstvennyi Universitet/Kazan State University
- Местоположение издательства: Russian Federation
- Первая страница: 3159
- Последняя страница: 3167
- Аннотация: Abstract—The paper studies domain-transfer capabilities of the machine reading comprehension model (MRC) in named entity recognition task. In such a task, the model is required to retrieve some knowledge from texts with prompts in order to extract named entities. Usually, this requires training on a large dataset. However in low-resource or domain-specific scenarios creating training dataset could be hard and non-effective. This is where domain transfer from available datasets can be useful. We check prompt design for the machine reading comprehension model to transfer knowledge between domains. We study the approach in transfer from general dataset NEREL to biomedical NEREL-BIO, which share some set of common name entities.
- Добавил в систему: Рожков Игорь Сергеевич