Natural Langauge Processing by Probabilistic/Stocastic Methos

Unknown words appearing in texts give us a serious problem when natural language processing. We have been developing Probabilistic/Stocastic Model for new word generation. By applying this model we can infer lexical, syntactic, and semantic attribute which enables us to continue the processing without hunging up it.

In order to save processing time we have been developing a lexicon for some of such unkonown words. Such unknown words are usually delivatives or compounds. Even if they are not included in general dictionaries, they might be very popular words in some specific area. We have already developed such a delivetive lexicon which contains more than eighty thousands delivetives.

To Linguistic Data　(Access Restricted)

TO OUR HOMEPAGE

Back to Home Page