TY - JOUR
AU - 
AB - 1 2 2 1 Shyam Upadhyay Manaal Faruqui Chris Dyer Dan Roth Department of Computer Science, University of Illinois, Urbana-Champaign, IL, USA School of Computer Science, Carnegie Mellon University, Pittsburgh, PA, USA upadhya3@illinois.edu, mfaruqui@cs.cmu.edu cdyer@cs.cmu.edu, danr@illinois.edu Abstract al., 2015) or word level (Faruqui and Dyer, 2014; Gouws and Søgaard, 2015), while some require Despite interest in using cross-lingual both sentence and word alignments (Luong et al., knowledge to learn word embeddings for 2015). However, a systematic comparison of these various tasks, a systematic comparison of models is missing from the literature, making it the possible approaches is lacking in the difficult to analyze which approach is suitable for a literature. We perform an extensive eval- particular NLP task. In this paper, we fill this void uation of four popular approaches of in- by empirically comparing four cross-lingual word ducing cross-lingual embeddings, each re- embedding models each of which require different quiring a different form of supervision, form of alignment(s) as supervision, across several on four typologically different language dimensions. To this end, we train these models on pairs. Our evaluation setup spans four dif- four different language pairs, and evaluate them on ferent tasks, including intrinsic evaluation both monolingual 
TI - Cross-lingual Models of Word Embeddings: An Empirical Comparison
JF - Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
DO - 10.18653/v1/p16-1157
DA - 2016-01-01
UR - https://www.deepdyve.com/lp/unpaywall/cross-lingual-models-of-word-embeddings-an-empirical-comparison-KCKG2UWU1O
DP - DeepDyve
ER -