Uyghur Word Stemming Based on Stem and Affix Features

Uyghur is an agglutinative language with complex morphology, and word stemming is one of the essentials in Uyghur information processing. However, the performance of Uyghur word-stem segmentation still leaves much room for improvement. In this study, stemming was performed on Uyghur words using an affix-occurred probability feature, which provided the stemming accuracy of 88.59% for a baseline system. The performance of this stemmer was further improved by using parameter ‘α’ in combination with the proposed method.