Hierarchical Mapping for Crosslingual Word Embedding Alignment

Ion Madrazo Azpiazu, Maria Soledad Pera


Abstract
The alignment of word embedding spaces in different languages into a common crosslingual space has recently been in vogue. Strategies that do so compute pairwise alignments and then map multiple languages to a single pivot language (most often English). These strategies, however, are biased towards the choice of the pivot language, given that language proximity and the linguistic characteristics of the target language can strongly impact the resultant crosslingual space in detriment of topologically distant languages. We present a strategy that eliminates the need for a pivot language by learning the mappings across languages in a hierarchical way. Experiments demonstrate that our strategy significantly improves vocabulary induction scores in all existing benchmarks, as well as in a new non-English–centered benchmark we built, which we make publicly available.
Anthology ID:
2020.tacl-1.24
Volume:
Transactions of the Association for Computational Linguistics, Volume 8
Month:
Year:
2020
Address:
Cambridge, MA
Editors:
Mark Johnson, Brian Roark, Ani Nenkova
Venue:
TACL
SIG:
Publisher:
MIT Press
Note:
Pages:
361–376
Language:
URL:
https://aclanthology.org/2020.tacl-1.24
DOI:
10.1162/tacl_a_00320
Bibkey:
Cite (ACL):
Ion Madrazo Azpiazu and Maria Soledad Pera. 2020. Hierarchical Mapping for Crosslingual Word Embedding Alignment. Transactions of the Association for Computational Linguistics, 8:361–376.
Cite (Informal):
Hierarchical Mapping for Crosslingual Word Embedding Alignment (Azpiazu & Pera, TACL 2020)
Copy Citation:
PDF:
https://aclanthology.org/2020.tacl-1.24.pdf