Using Mutual Information Criterion to Design an Effective Lexicon for Chinese Pinyin-to-Character Conversion

Pinyin-to-character (P2C) conversion is mostly used to input Chinese characters into a computer. Its main problem is homophone words, which is solved through exploiting contextual information provided by lexicon and n-gram language model (LM). Our investigation about the state-of-the-art P2C technologies reveals that the methods of conventional optimization for them were almost based on minimizing text perplexity, however it is not directly related to the optimization of P2C performance. Therefore, we propose to use a new optimization criterion: mutual information (MI) between text corpus and its Pinyin script, to do self-supervised word segmentation, build a lexicon and estimate an n-gram LM, then use them to build P2C system. We realized the P2C system using newspaper corpus. Compared with the two baseline systems using handcrafted lexicon and perplexity based optimized lexicon, our system got relatively 19.7% and 10.3% error reductions on testing corpus respectively. The results show the efficiency of our proposal.