基于清华汉语树库的时间表达式抽取模型构建研究

Construction of Temporal Expression Extraction Model Based on Tsinghua Chinese Treebank

基于清华汉语树库,利用条件随机场,构建时间表达式知识抽取模型。为了确保模型的性能,统计分析时间表达式的内部和外部特征,并把确定的特征应用到模型的构建中。在语言特征模板基础上,在摘要语料上抽取验证表明,时间表达式抽取模型的调和平均值最高达到80.83%。

Using the Conditional Random Fields （CRF）, the paper constructs the temporal expression extraction model based on Tsinghua Chinese Treebank. In order to ensure the perfornlance of the model, the internal and external syntactic features are obtained and arialyzed, and the definite features are applied to constructing the model. And this model is verified in the abstract corpus, the result shows that the best F-score of the temporal expression extraction model, which is obtained out of the template of the linguistic features, reaches up to 80.83%.