Don't Just Scratch the Surface: Enhancing Word Representations for Korean with Hanja

TL;DR

This paper introduces a method to improve Korean word representations by incorporating Hanja through cross-lingual transfer learning, enhancing their quality for various NLP tasks.

Contribution

It presents a novel approach leveraging Hanja and cross-lingual transfer learning to enhance Korean word embeddings, validated through intrinsic and downstream evaluations.

Findings

Improved performance on word analogy and similarity tests.

Enhanced results on Korean news headline generation.

Effective use of Hanja for cross-lingual transfer in Korean NLP.

Abstract

We propose a simple yet effective approach for improving Korean word representations using additional linguistic annotation (i.e. Hanja). We employ cross-lingual transfer learning in training word representations by leveraging the fact that Hanja is closely related to Chinese. We evaluate the intrinsic quality of representations learned through our approach using the word analogy and similarity tests. In addition, we demonstrate their effectiveness on several downstream tasks, including a novel Korean news headline generation task.