自动词性标注方法的比较  

Analysis and Comparison of the Part-of-Speech Tagging Techniques

作　　者：陈晓文[1] 

机构地区：[1]温州大学外国语学院

出　　处：《温州大学学报》2006年第1期53-57,共5页

摘　　要：对机器自动词性标注技术领域的三类主要理论方法(基于规则的方法、基于统计的方法和规则与统计相结合的方法)进行了研究分析和优缺点的对比,并在描述方式、标注依据、机器效率、鲁棒性、标注正确率和实用性等方面,对这三类方法进行认真的比较。比较结果显示规则与统计相结合的方法在各方面都占有较明显的优势,是目前最理想的标注方法。基于此类方法的自动词性标注技术可以较好地满足实际应用的要求。此外,本文还指出这类方法有待解决的三大难题。With the development of the natural language processing technology, diverse techniques of part-of-speech tagging have got boost in recent years. After the elaborate study of those techniques, we find that the core methodology of them falls into three groups： rule-based, statistics-based and the combination of rule and statistics. In this paper, we put the main effort on the comparison of the three types of methods and point out the advantage, the disadvantage and some serious problems. Furthermore, the article concludes that the combinatory method achieves the best results and possesses the applicable value. However, the combinatory method also leaves some haunting problems as well.

关 键 词：词性标注 规则 统计 概率 兼类词 

分 类 号：H08[语言文字—语言学]