In Modern Chinese Language Structure Of Automatic Identification

Subjective-Object Structure is one of the most common phrases in Chinese used in a high frequency both in colloquial and written language. There are so many clear studies of Subjective-Object Phrases in the traditional linguistics, undergoing a procedure of raising the problem to arguing it and to the agreement at last from the middle of the last century. Most of them were limited in the Three Plane Language Theory without quota analysis. The study on the Chinese verb minor is very important, someone put forward the theory of (?)he Special Chinese Minor(?) hich consists of the directional verb and the willing verb, it is seldom to see the Subjective-Object Verb in the grammatical circles. Chinese teaching, Chinese studying and Chinese information processing all make the needs of investigating the characteristic of subjective-object structure in a large scale. There are no publishing results about the computer automatic identification at present; however this is a precondition of the investigation on a large scale. The automatic identification of the subjective object structure is also important to dealing Chinese with statistics.This paper is about the Subjective-Object Structure of Automatic Identification adopting the method of combining statistics and the rules. In this paper, the study found that V1 can be recognized by the subjective object structure automatically therefore the key point is to establish a detailed vocabulary of the automatic identification of it listed with the possibility of subjective object in order to make a effort to improve the recall rate. A comprehensive set of rules, that is, the building of the subjective object candidate, takes an important effort to identify the subjective object. Because there are several characteristics for one subjective object, and a appropriate way to organize them, so we design a calculating method to measure the support degree of each subjective object, calculating each feature of subjective object candidate quantitatively and effectively in order to organize and identify the subjective object. The automatic identification of V2 in subjective object is not a neglected step in the subjective object identification, but it involves the automatic identification of verb phrase, especially the right border delimitation which is very complicated and is beyond our study scope, so at the end of this paper V2 is just marked automatically.We tested our study of automatic identification on the Daily of 2 million words corpus of the January 1998, although we have established a more detailed V1 vocabulary during the process of our study, it can not reach 100 percent recall rate (it can theoretically), this is caused by quality of corpus itself for one hand, the rules set by this paper for the other. At the recognition of the subjective object, its candidate feature and its support degree take a supplementary role; some candidate feature makes support degree calculate simpler, which solves insuperable problems with candidate feature alone. The result shows that our identification can reach 80 percent of F-core, closer to the practical level.