Title: Study of semantic automatic error-detecting for Chinese text based on sememe matching of Hownet
Abstract: The study of automatic proofreading of Chinese text is always very difficult at the semantic level.The semantic matching error of Chinese text is discussed and an effective automatic error-detecting method is proposed based on the sememes matching of Hownet.This method includes two parts: the constructing of the semantic repository and the algorithm of automatic error-detecting.The procedure of constructing the repository is: ① Statistics huge amounts of the binary-combination of verbs and nouns from training corpus by a group of rules,than filter the records through the factors: mutual-information etc.② Use the sememe descriptions of words’ meanings defined in Hownet(for verbs choose the main sememe and for nouns choose the main sememe and the other four common sememes) to convert the binary-combination of verbs and nouns into the restricted multi-combination of sememes.The automatic errordetecting algorithm could be devised appropriately on the basis of the semantic repository.The experimental result shows that the recall rate is 35% and accuracy rate is 82.3%,which reflects the good application prospect.
Publication Year: 2010
Publication Date: 2010-01-01
Language: en
Type: article
Access and Citation
Cited By Count: 3
AI Researcher Chatbot
Get quick answers to your questions about the article from our AI researcher chatbot