Research On Combination Class New Word Identification Technology Based On Rules And Statistics And Its Application In The Design Of Mechanical Products

This paper studies the Chinese new word identification in the field of NaturalLanguage Understanding, finally with the method of understand and analyze the userneeds which expressed in the form of natural language, provide support for thesubsequent design.This paper firstly analyzes several commonly knowledge representations in NaturalLanguage Understanding, elaborate the function of Conceptual Dependency Theory indescribing the connotation and extension of knowledge, as a theoretical basis foridentifying the combination class Chinese new word. Secondly, as the theoretical basisof linguistic knowledge, analyze the structure mode of the combination class Chinesenew word; based on word formation structure, determine the classification criteria,divide the combination class Chinese new word into three categories:“single charactermode”,“suffix character mode” and “other mode”; through word disable table, part ofspeech disable table and affix table filtering possible word string; by means ofConceptual Dependency Tree Model, analyze the relationship among the combinationelements. and then collect the information about the context of the possible word string,by means of Statistical language model, inspecting internal combination and externallinguistic environment of the possible word string, which are on the basis of the insideword probability, the mutual information and the left and right entropy. This paperadopts the method of combined rules and statistic to identify the combination classChinese new word. Finally design the process for the identification of the combinationclass Chinese new word, and through the instance of mechanical design, put the analysismodel of the identification of the combination class Chinese new word applied toproduce design in the field of Natural Language Understanding.