|
ИСТИНА |
Войти в систему Регистрация |
ИПМех РАН |
||
Multiword Terms and Machine Translationстатья
- Автор: Потемкин С.Б.
- Сборник: Proceedings of the Third International Conference EUROPHRAS 2019
- Год издания: 2019
- Место издания: Malaga Spain
- Первая страница: 133
- Последняя страница: 139
- Аннотация: Abstract. In this article we discuss, using morphological analysis and foreign equivalent selection, the following issues: the definition of syntax and the logical-semantic structure of terms using morphological analysis and foreign equivalent selection. Texts of articles in the “Voprosy Psichologii” (Psychological Problems) journal were used as the source corpus. Statistics of multiword terms in this corpus were calculated as well as the pointwise mutual information (PMI) between terms. We defined the distance between multiword terms in a multidimensional space, and the measure of terms proximity, as the minimal semantic distance between them. Then, we performed multidimensional scaling for visualising the terms space. Machine translation was used as a means to finding equivalents of scientific and technical terms in two arbitrary languages. With the help of forward and reverse translation of terms, using online machine translation tools, the meaningful equivalents were evaluated. We measured the proximity between terms as the Levenshtein distance between the original term and its direct and reverse translation, and tried to minimise this measure.
- Добавил в систему: Потемкин Сергей Борисович