APPARATUS AND METHOD FOR RECOGNIZING AND CLASSIFYING NAMED ENTITIES FROM TEXT DOCUMENT USING REPEATED LEARNING

摘要

PURPOSE: An apparatus and a method for recognizing and classifying named entities from text document using repeated learning are provided to gradually extend a range for recognizing the object name by automatically extending a pattern rule and a vocabulary dictionary through the repeated learning. CONSTITUTION: A language quality extractor(10) extracts a language quality needed for recognizing the object name from a general text document set(11) inputted from the outside. A vocabulary dictionary extender(20) decides an object name list(22) and a vocabulary candidate to be added by applying the current pattern rule to the document set, and extends the vocabulary dictionary(12). A pattern rule extender(30) extends the pattern rule list(21) by generating/certifying a new pattern rule candidate as applying the vocabulary dictionary to the document set.