| Abstrakt: |
In recent years, the research of Chinese natural language processing on syntactic analysis has done a lot of work, and also has many achievements. The study of Chinese language processing has turned from the analysis on sentence level to analysis above the sentence level, such as the semantic role, function block, sentence chunks, etc. In this paper, the study of Chinese sentence constituent labeling is also an analysis above the sentence level, and this research has not been found yet before. In this paper, we apply the Hidden Markov model (HMM) for tagging constituents in each layer of the syntax structure tree, and also use other useful information, such as context information, auxiliary constituents, and heuristic language rules. Our method of sentence constituent labeling has reached 90.3% and 89.9% of the precision and recall rate respectively, and the results of constituent labeling have a promising research and application value in natural language applications. [ABSTRACT FROM PUBLISHER] |
