The Research On The Recognition Of Redundant Negation Formats

This paper brings forth new approaches to the recognition of various formats of redundant negation.The redundant negation phenomenon refers to the pragmatic fact that some components,while posses negative lexical meanings,e.g.“不、没、没有、别、非、未”,do not semantically exhibit negation in certain contexts.Former researches on redundant negation were primarily focused on linguistics aspect,yet no known research on the application of Natual Language Processing(NLP)in the recognition of redundant negation has been presented.This paper is devoted to the research on the recognition of redundant negation formats.The outcome will not only contribute to semantic recognition and computational interpretation of Chinese language but can also be introduced to redress incorrect word segmentation.This paper illustrates the recognition approach for each format by determining the syntactic features of several typical redundant negation formats,identifying the recognition strategy,working out the recognition programs using Python and analyzing the recognition result.The main research objects are the following formats:“好不XP”“难免不/没(有)XP”“没(有)XP之前”“差(一)点(儿)没(有)XP”.Other formats,such as“小心别XP”“险些没XP”“�'不XP”,are excluded due to inadequate corpus.The recognition strategy subtly varies for each format.In general,however,the approaches introduced in this paper follow rules instead of statistic methods as a result of the particularity of this phenomenon and rarity of corpus.Firstly I organized and analyzed the redundant negation formats in the training corpus and explored the regularity and formalized it into computationally processible algorithm.Then I ran the algorithm against new input corpus and decided whether the formats are redundant negation.We can infer from the experiments in this paper that the F-score in the test corpus exceeds 90%,and more excitingly,the F-score of the formats“没(有)XP之前”and“差(一)点(儿)没(有)XP”can reach 95%.We can thus bring forth the conclusion that recognition based on language knowledge rules is very outstanding.