Word Segmentation for Javanese Character Using Dictionary, SVM, and CRF
Dipta Tanaya, Mirna Adriani · 2018
Word segmentation method based on dictionary for Javanese character still cannot overcome the problem of ambiguity, derivational word segmentation, and unknown words identification. We propose a new approach for Javanese character segmentation, i.e. a machine learning based method using CRF and SVM with a set feature of Javanese character's categorization and its combination with dictionary-based method. Through several experiments, it is proven that combination of dictionary-based method and CRF performs best than word segmentation using CRF, SVM, and traditional dictionary-based word segmentation.