Part of Speech (POS) Tagger for Kokborok
Braja Gopal Patra, Khumbar Debbarma, Dipankar Das, Sivaji Bandyopadhyay · International Conference on Computational Linguistics · 2012
The Part of Speech (POS) tagging refers to the process of assignin g appropriate lexical category to individual word in a sentence of a natural language. This paper describes the d evelopment of a POS tagger using rule based and supervised methods in Kokborok, a reso urce constrained and less computerized Indian language. In case of rule based POS tagging, we took the help of a morphological analyzer while for supervised methods, we employed two machine learning classifiers, Conditional Random Field (CRF) and Support Vector Machines (SVM). A total of 42,537 words were POS tagged. Manual checking achieves the accuracies of 70% and 84% in case of rule based and supervised POS tagging, respectively.