Analysis of Joint Multilingual Sentence Representations and Semantic K-Nearest Neighbor Graphs

Holger Schwenk, Douwe Kiela, Matthijs Douze

AAAI 2019 pp. 6982-6990

doi:10.1609/AAAI.V33I01.33016982 /aaai/2019/schwenk2019aaai-analysis/

Abstract

Multilingual sentence and document representations are becoming increasingly important. We build on recent advances in multilingual sentence encoders, with a focus on efficiency and large-scale applicability. Specifically, we construct and investigate the k-nn graph over the joint space of 566 million news sentences in seven different languages. We show excellent multilingual retrieval quality on the UN corpus of 11.3M sentences, which extends to the zero-shot case where we have never seen a language. We provide a detailed analysis of both the multilingual sentence encoder for twenty-one European languages and the learned graph. Our sentence encoder is language agnostic and supports code switching.

PDF AAAI Semantic Scholar

Cite

Text

Schwenk et al. "Analysis of Joint Multilingual Sentence Representations and Semantic K-Nearest Neighbor Graphs." AAAI Conference on Artificial Intelligence, 2019. doi:10.1609/AAAI.V33I01.33016982

Markdown

[Schwenk et al. "Analysis of Joint Multilingual Sentence Representations and Semantic K-Nearest Neighbor Graphs." AAAI Conference on Artificial Intelligence, 2019.](https://mlanthology.org/aaai/2019/schwenk2019aaai-analysis/) doi:10.1609/AAAI.V33I01.33016982

BibTeX

@inproceedings{schwenk2019aaai-analysis,
  title     = {{Analysis of Joint Multilingual Sentence Representations and Semantic K-Nearest Neighbor Graphs}},
  author    = {Schwenk, Holger and Kiela, Douwe and Douze, Matthijs},
  booktitle = {AAAI Conference on Artificial Intelligence},
  year      = {2019},
  pages     = {6982-6990},
  doi       = {10.1609/AAAI.V33I01.33016982},
  url       = {https://mlanthology.org/aaai/2019/schwenk2019aaai-analysis/}
}