Dependency Parsing for Long Chinese Sentence

This book explores the problem of parsing long sentences, and Chinese is the target language. The result shows that the best performance is achieved by the algorithm using Chinese comma (algorithm 1) as the boundary of sub-sentence. It improves the overall accuracy and the accuracy on long distance dependency of the baseline parser by around 1.5% and 6.0% separately. According to the design of the parsing method, if a better baseline parser is adopted, the accuracy can be further improved according to the design of the parsing method.

Autorentext

Bo Li, Mestre em Informática na Universidade de Edimburgo.