词性标注的方法研究——结合条件随机场和基于转换学习的方法进行词性标注

Research on the Part-of-Speech Tagging Method——POS Tagging using Conditional Random Fields and Transformation Based Learning

词性标注是语料库建设中的重要环节,同时也是自然语言处理领域的基础研究课题。针对词性标注任务中统计处理和规则处理两种方法各自的特点和局限,提出融合条件随机场模型和基于转换学习的方法来进行自动词性标注的方案。实验结果表明,该方案能够有效地提高词性标注的正确率。

POS tagging is an important part of corpora building and a basic research in the field of NLP. After comparing the advantage and weakness of the rule - based methods and the statistical methods, an automatic POS tagging method based on both CRF and TBL is presented. And the tests prove that the method can improve the accuracy of words tagging.