Joint Word Segmentation and Stemming for Myanmar Language Based on Conditional Random Fields

In this paper, we describe a joint work on wordsegmentation and stemming of Myanmar sentenceswith syllabled-based tagging under ConditionalRandom Fields(CRF) framework. A manuallysegmentedcorpus was developed to train thesegmenter, and we implement it as a 7-tag syllablebasedtagging and stemming with conditional randomfields(CRF). And then, the trained CRF segmenter wascompared to a baseline approached based on longestmatching that used a dictionary extracted frommanually segmented corpus. In our approach, we canachieve comparative performances compared to 4-tagsyllable tagging approach. The experimental resultsshow that the CRF with 7-tag set and word featureimprove the stemming performance