查看文章

One of the fundamental works in natural language processing is creating a feasible corpus for evaluating effectiveness of different algorithms. In this paper, the authors report creation of test corpus of automatic part of speech tagging purposes based on the Persian tagged corpus of Prof. Bijankhan. This study includes preprocessing, statistical analysis and experiments with simple statistical POS tagging methods done on this corpus. Part of speech tagging experimental results show that even with a simple POS method such as Maximum Likelihood Estimation (MLE), we could reach to an get acceptable accuracy of 93.16 percent.