Towards a Universal Semantic Dictionary
A novel method for finding linear mappings among word embeddings for several languages, taking as pivot a shared, multilingual embedding space, is proposed in this paper. Previous approaches learned translation matrices between two specific languages, while this method learns translation matrices be...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2019-09-01
|
Series: | Applied Sciences |
Subjects: | |
Online Access: | https://www.mdpi.com/2076-3417/9/19/4060 |
id |
doaj-b916c25ae44546058b23d7d406b141dc |
---|---|
record_format |
Article |
spelling |
doaj-b916c25ae44546058b23d7d406b141dc2020-11-25T02:03:11ZengMDPI AGApplied Sciences2076-34172019-09-01919406010.3390/app9194060app9194060Towards a Universal Semantic DictionaryMaria Jose Castro-Bleda0Eszter Iklódi1Gábor Recski2Gábor Borbély3VRAIN Valencian Research Institute for Artificial Intelligence, Universitat Politècnica de València, 46022 Valencia, SpainBudapest University of Technology and Economics, 1111 Budapest, HungaryBudapest University of Technology and Economics, 1111 Budapest, HungaryBudapest University of Technology and Economics, 1111 Budapest, HungaryA novel method for finding linear mappings among word embeddings for several languages, taking as pivot a shared, multilingual embedding space, is proposed in this paper. Previous approaches learned translation matrices between two specific languages, while this method learns translation matrices between a given language and a shared, multilingual space. The system was first trained on bilingual, and later on multilingual corpora as well. In the first case, two different training data were applied: Dinu’s English−Italian benchmark data, and English−Italian translation pairs extracted from the PanLex database. In the second case, only the PanLex database was used. The system performs on English−Italian languages with the best setting significantly better than the baseline system given by Mikolov, and it provides a comparable performance with more sophisticated systems. Exploiting the richness of the PanLex database, the proposed method makes it possible to learn linear mappings among an arbitrary number of languages.https://www.mdpi.com/2076-3417/9/19/4060natural language processingsemanticsword embeddingsmultilingual embeddingstranslationartificial neural networks |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Maria Jose Castro-Bleda Eszter Iklódi Gábor Recski Gábor Borbély |
spellingShingle |
Maria Jose Castro-Bleda Eszter Iklódi Gábor Recski Gábor Borbély Towards a Universal Semantic Dictionary Applied Sciences natural language processing semantics word embeddings multilingual embeddings translation artificial neural networks |
author_facet |
Maria Jose Castro-Bleda Eszter Iklódi Gábor Recski Gábor Borbély |
author_sort |
Maria Jose Castro-Bleda |
title |
Towards a Universal Semantic Dictionary |
title_short |
Towards a Universal Semantic Dictionary |
title_full |
Towards a Universal Semantic Dictionary |
title_fullStr |
Towards a Universal Semantic Dictionary |
title_full_unstemmed |
Towards a Universal Semantic Dictionary |
title_sort |
towards a universal semantic dictionary |
publisher |
MDPI AG |
series |
Applied Sciences |
issn |
2076-3417 |
publishDate |
2019-09-01 |
description |
A novel method for finding linear mappings among word embeddings for several languages, taking as pivot a shared, multilingual embedding space, is proposed in this paper. Previous approaches learned translation matrices between two specific languages, while this method learns translation matrices between a given language and a shared, multilingual space. The system was first trained on bilingual, and later on multilingual corpora as well. In the first case, two different training data were applied: Dinu’s English−Italian benchmark data, and English−Italian translation pairs extracted from the PanLex database. In the second case, only the PanLex database was used. The system performs on English−Italian languages with the best setting significantly better than the baseline system given by Mikolov, and it provides a comparable performance with more sophisticated systems. Exploiting the richness of the PanLex database, the proposed method makes it possible to learn linear mappings among an arbitrary number of languages. |
topic |
natural language processing semantics word embeddings multilingual embeddings translation artificial neural networks |
url |
https://www.mdpi.com/2076-3417/9/19/4060 |
work_keys_str_mv |
AT mariajosecastrobleda towardsauniversalsemanticdictionary AT eszteriklodi towardsauniversalsemanticdictionary AT gaborrecski towardsauniversalsemanticdictionary AT gaborborbely towardsauniversalsemanticdictionary |
_version_ |
1724948814612135936 |