Cross-lingual Word Analogies using Linear Transformations between Semantic Spaces

TL;DR

This paper introduces a new intrinsic evaluation method for cross-lingual semantic spaces using linear transformations to perform word analogies across six languages, demonstrating effective preservation of semantic relationships.

Contribution

It generalizes the word analogy task to multiple languages and compares various linear transformations for aligning semantic spaces across languages.

Findings

Linear transformations effectively preserve word relationships.

Achieved average accuracy of 51.1% monolingual, 43.1% bilingual, 38.2% multilingual.

Transformations enable cross-lingual semantic space alignment.

Abstract

We generalize the word analogy task across languages, to provide a new intrinsic evaluation method for cross-lingual semantic spaces. We experiment with six languages within different language families, including English, German, Spanish, Italian, Czech, and Croatian. State-of-the-art monolingual semantic spaces are transformed into a shared space using dictionaries of word translations. We compare several linear transformations and rank them for experiments with monolingual (no transformation), bilingual (one semantic space is transformed to another), and multilingual (all semantic spaces are transformed onto English space) versions of semantic spaces. We show that tested linear transformations preserve relationships between words (word analogies) and lead to impressive results. We achieve average accuracy of 51.1%, 43.1%, and 38.2% for monolingual, bilingual, and multilingual semantic spaces, respectively.