Author(s): Anders Søgaard, Ivan Vulić, Sebastian Ruder, Manaal Faruqui, Graeme Hirst
Series: Synthesis Lectures on Human Language Technologies
Publisher: Morgan & Claypool Publishers
Year: 2019
Language: English
Pages: 134
Preface......Page 13
Introduction......Page 15
Monolingual Word Embedding Models......Page 23
Cross-Lingual Word Embedding Models: Typology......Page 27
A Brief History of Cross-Lingual Word Representations......Page 35
Cross-Lingual Word Representations using Bilingual Lexicons......Page 37
Cross-Lingual Word Embeddings and Word Alignments......Page 41
Representations Based on Latent and Explicit Cross-Lingual Concepts......Page 43
Summary......Page 46
Word-level Alignment Methods with Parallel Data......Page 47
Mapping-Based Approaches......Page 48
Pseudo-Mixing Approaches......Page 55
Joint Approaches......Page 56
Sometimes Mapping, Joint, and Pseudo-Bilingual Approaches are Equivalent......Page 58
Word-Level Alignment Methods with Comparable Data......Page 60
Sentence-Level Methods with Parallel Data......Page 63
Sentence Alignment with Comparable Data......Page 67
Document Alignment with Comparable Data......Page 69
Multilingual Word Embeddings from Word-Level Information......Page 73
Multilingual Word Embeddings from Sentence-Level and Document-Level Information......Page 78
Unsupervised Learning of Cross-Lingual Word Embeddings......Page 81
Seed Dictionary Induction......Page 82
Refinement and Heuristics......Page 86
Limitations of Unsupervised Approaches......Page 87
Intrinsic Evaluation......Page 89
Extrinsic Evaluation Through Cross-Lingual Transfer......Page 90
Multi-Modal and Cognitive Approaches to Evaluation......Page 94
Monolingual Resources......Page 97
Cross-Lingual Word Embedding Models......Page 98
Evaluation and Application......Page 100
General Challenges and Future Directions......Page 103
Bibliography......Page 107
Authors' Biographies......Page 133
Blank Page......Page 2
