基于Rough Sets和模糊神经网络的汉语兼类词词性标注规则的获取方法

文章提出了基于RoughSets的汉语兼类词初始标注规则的获取方法,并通过模糊神经网络(FNN)进行优化,最后再进行简化获取模糊规则;文章以人工标注过的句子作为训练集和测试集,得出了训练集左3、左4、右3、右4个兼类词标注规则库;对同样的训练集和测试集,采用统计二元模型进行标注后,再利用该方法(粗糙模糊神经网络方法,简称RSFNN)进行二次标注,结果表明RSFNN方法优于统计二元模型方法。最后实例说明汉语兼类词词性标注规则的获取方法。

The Acquiring Method of Chinese Ambiguity Word POS Tagging Rules Based on Rough Sets and Fuzzy Neural Network

This paper proposes a method of acquiring Chinese ambiguity word part of speech(POS)tagging fuzzy rules.Firstly,It acquires original fuzzy rules of ambiguity POS tagging based on Rough Sets,optiming the fuzzy rules based on a fuzzy neural network,lastly,reducing the fuzzy rules.Annotated sentences is used as training set and testing set,and acquiring ambiguity word tagging rules of left-3,left-4,right-3,right-4.To the same training set and testing set,using this method(the method of Rough Fuzzy Neural,etc RSFNN)to go on the second POS tagging after using statistical2-model carry out POS tagging,the result shows that the method of RSFNN is superior to the effect of POS tagging based on statistical model.Finally,a example is used to illustrate the acquiring method of ambiguity word POS tagging rules.