Paper: Applying a Mix Word-Pair Identifier to the Chinese Syllable-to-Word Conversion Problem

ACL ID I05-2010
Title Applying a Mix Word-Pair Identifier to the Chinese Syllable-to-Word Conversion Problem
Venue International Joint Conference on Natural Language Processing
Session poster-demo-tutorial
Year 2005
Authors
  • Jia-Lin Tsai (Tung Nan Institute of Technology, Taipei Taiwan)

This paper describes a mix word-pair mix-WP) identifier to resolve homo- nym/segmentation ambiguities as well as perform STW conversion effec- tively for Chinese input. The mix-WP identifier includes a specific word-pair (SWP) identifier and a common word- pair (CWP) identifier. It is designed as a supporting processing with Chinese input systems. Our experiments show that by applying the mix-WP identifier, together with the Microsoft input method editor 2003 (MSIME) and an optimized bigram model (BiGram), the tonal and toneless STW perform- ance of the two input systems can be improved.