Many difficulties remain in identifying the sentence skeleton of Chinese sentences. The purpose of this paper is to automatically identify the sentence skeleton of Chinese sentences on the event level. Based on the ev...Many difficulties remain in identifying the sentence skeleton of Chinese sentences. The purpose of this paper is to automatically identify the sentence skeleton of Chinese sentences on the event level. Based on the event model, this paper discusses the relationship between events and concepts, and presents an event conceptualization method. It analyzes the event composition of Chinese sentences, classifies the re- lationships among events, and proposes a recursive description of the event composition of Chinese sen- tences. Based on these analyses, an algorithm for automatically identifying the sentence skeleton of Chi- nese sentences based on the event model is presented. The results of the final experiment show an 89% success rate, which demonstrates that the proposed algorithm is feasible and effective.展开更多
基金financially supported by the National Natural Science Funds for Distinguished Young Scholars(21725401)the National Key R&D Program of China(2017YFA0207800)the National Natural Science Foundation of China(22161142021)。
基金Supported by the National Natural Science Foundation of China (No. 51105290)
文摘Many difficulties remain in identifying the sentence skeleton of Chinese sentences. The purpose of this paper is to automatically identify the sentence skeleton of Chinese sentences on the event level. Based on the event model, this paper discusses the relationship between events and concepts, and presents an event conceptualization method. It analyzes the event composition of Chinese sentences, classifies the re- lationships among events, and proposes a recursive description of the event composition of Chinese sen- tences. Based on these analyses, an algorithm for automatically identifying the sentence skeleton of Chi- nese sentences based on the event model is presented. The results of the final experiment show an 89% success rate, which demonstrates that the proposed algorithm is feasible and effective.
