Polyphone Disambiguation in Mandarin Chinese with Semi-Supervised Learning
This work addresses a domain-specific challenge in natural language processing for Mandarin Chinese, with incremental improvements in method and dataset availability.
The paper tackles the problem of polyphone disambiguation in Mandarin Chinese, which is crucial for speech-related tasks, by proposing a semi-supervised learning framework that leverages unlabeled text data and achieves state-of-the-art performance, while also releasing a new dataset to support further research.
The majority of Chinese characters are monophonic, while a special group of characters, called polyphonic characters, have multiple pronunciations. As a prerequisite of performing speech-related generative tasks, the correct pronunciation must be identified among several candidates. This process is called Polyphone Disambiguation. Although the problem has been well explored with both knowledge-based and learning-based approaches, it remains challenging due to the lack of publicly available labeled datasets and the irregular nature of polyphone in Mandarin Chinese. In this paper, we propose a novel semi-supervised learning (SSL) framework for Mandarin Chinese polyphone disambiguation that can potentially leverage unlimited unlabeled text data. We explore the effect of various proxy labeling strategies including entropy-thresholding and lexicon-based labeling. Qualitative and quantitative experiments demonstrate that our method achieves state-of-the-art performance. In addition, we publish a novel dataset specifically for the polyphone disambiguation task to promote further research.