Research the Method of Joint Segmentation and POS Tagging for Tibetan using BiGRU-CRF

Tibetan word segmentation and part-of-speech tagging are the most basic parts of Tibetan natural language processing, and its accuracy and performance have a crucial impact on many subsequent tasks. Considering the insufficiency of the pipeline model of word segmentation and part-of-speech tagging, this paper uses an integrated model of BiGRU-CRF word segmentation and part-of-speech tagging based on deep learning to simultaneously process two tasks of Tibetan word segmentation and part-of-speech tagging in one step. After conducting experiments on the Tibetan corpus collected in "Humanistic Tibet", the joint F1 value of Tibetan word segmentation and part-of-speech tagging was 92.48%.