APPARATUS AND METHOD FOR RECOGNIZING THE NAMED ENTITY USING BACKOFF N-GRAM FEATURES

摘要

A device and a method for recognizing an entity name by using a back-off n-gram quality are provided to solve a problem caused by scarcity of data and guarantee reliability for recognizing the entity name of an unregistered or low frequency word. A morphologically analyzed trained corpus(41) is made by passing the corpus through a lexical analyzer(20) and is inputted to a quality information extractor(33). The quality information extractor processes a morphologically analyzed input sentence(42) inputted from the outside through a sentence input unit and the lexical analyzer. The quality information extractor provides training quality(43) to be used in a model trainer(34) by extracting the quality for the morphologically analyzed training corpus based on the back-off n-gram quality. The quality information extractor provides test quality(44) to be used in a candidate entity name extractor by using the quality for the morphologically analyzed input sentence .