Paper: A Chinese LPCFG Parser with Hybrid Character Information

ACL ID W10-4150
Title A Chinese LPCFG Parser with Hybrid Character Information
Venue Joint Conference on Chinese Language Processing
Session Main Conference
Year 2010
Authors

We present a new probabilistic model based on the lexical PCFG model, which can easily utilize the Chinese character in- formation to solve the lexical information sparseness in lexical PCFG model. We discuss in particular some important fea- tures that can improve the parsing perfor- mance, and describe the strategy of mod- ifying original label structure to reduce the label ambiguities. Final experiment demonstrates that the character informa- tion and label modification improve the parsing performance.