Predicting part-of-speech tags and morpho-syntactic relations using similarity-based technique

This paper describes a similarity-based technique which produces a good estimate of part-of-speech tags and their morpho-syntactic relations of Chinese compound words before they are fed into a tagger. The technique relies on a set of features from Chinese morphemes as well as a set of collocation markers which provide hints on the syntactic categories of the compound words. The technique is trained with a compound words database with more than 53,500 disyllabic words. Experimental results show the tagger with the technique outperforms its counterpart.