Conference paper Open Access

A Comparison of Unsupervised Methods for Ad hoc Cross-Lingual Document Retrieval

Zosa, Elaine; Granroth- Wilding, Mark; Pivovarova, Lidia

We address the problem of linking related documents across languages in a multilingual collection. We evaluate three diverse unsupervised methods to represent and compare documents: (1) multilingual topic model; (2) cross-lingual document embeddings; and (3) Wasserstein distance. We test the performance of these methods in retrieving news articles in Swedish that are known to be related to a given Finnish article. The results show that ensembles of the methods outperform the stand-alone methods, suggesting that they capture complementary characteristics of the documents.

Files (131.6 kB)
Name Size
LREC_cross_lingual_news_linking.pdf
md5:f419cfd229d47c8a86a09e76aac3cdc9
131.6 kB Download
25
24
views
downloads
All versions This version
Views 2525
Downloads 2424
Data volume 3.2 MB3.2 MB
Unique views 2121
Unique downloads 2222

Share

Cite as