Abstract:

A rulebased tagging method for unsupervised partofspeech was proposed. More than 200 grammar rules were used to create 26 kinds of rules functions. After it was preprocessed, the initial tags of words in the input sentence were obtained, the 26 kinds of rules functions were applied to each word to attain all the tags of the input sentence. Experimental results on Brown corpus show that the accuracy of our method is up to 93.95%, thus, our rulebased method is feasible and effective, and improves the accuracy and the simplicity of English partofspeech tagging.

Key words: partofspeech tagging, rulebased, unsupervised learning, rules function