Ambiguïté des kanji et stratégie de désambiguïsation par le contexte

Résumé

In this article we will examine the ambiguity of kanji, characters of Chinese origin, and one of the character sets used to write the Japanese language. We will first describe the ambiguity phenomenon by presenting the origin of this problem and its multilevel effects on the Japanese texts decoding. Then, we will expose a context based method of disambiguating kanji which draws on Japanese graphotax specificities themselves. Lastly, we will show the effectiveness of this disambiguating method when applied to computational linguistics and particularly to an automatic morphological analysis and segmentation system for Japanese texts.