本論文提出一個演算法去改善原模型(馬可夫模型標記器、TnT標記器)的詞性標記的準確率，並以七個錯誤率較高的特徵字作為研究對象。所設計的演算法是透過原標記器去標記句子裡每一個單字的詞性，再利用字彙資訊與相對機率比值，給予特徵字有第二次標記的機會。 數據探討分成兩部分，分別為(一)七個特徵字在馬可夫模型標記器伴隨字彙資訊與馬可夫標記器的整體錯誤率比較；(二) 七個特徵字在TnT標記器伴隨字彙資訊與TnT標記器的整體錯誤率比較。經數據的分析顯示，我們的演算法確實可以提升標記器的準確率。 This paper presents an algorithm to improve the original model (Markov model tagger, TnT tagger) of accuracy of speech tags and take the higher error rate feature word as the object of study. The algorithm we designed is through the original tagger to tag the part of speech of each word in the sentence and then use lexical information and relative probability ratio to give the feature word a second tagged chance. The probing of data is divided in two parts, respectively ( a ) Comparison of the overall error rate of seven feature words in Markov model tagger with lexical information and Markov model tagger, ( b ) Comparison of the overall error rate of seven feature words in TnT tagger with lexical information and TnT tagger. The data analysis shows that our algorithm can improve the accuracy of tagger exactly.
