The Research and Implementation for Dealing with the Difficulties Arose by More Verbs in Chinese Understanding
by JIN Yao-Hong
Under the framework of the theory of Hierarchical Network of Concepts (HNC), the research to deal with the difficulties arose by more verbs when the computer analyzes the semantic structure of the sentence, is presented here. The goal of this research is to make sure which verb is the center of the sentence.
Also the strategy of analyzing these difficulties (HNC-MV) is designed in the Sentence Category Analysis system (SCA) by means of the hypothesis and test. The main work is:
At first the verbs are classified to three classes, which are normal verb, the verb need attach other verb, and the verb need attach other noun. And the eigen chunk (we call it as E) is classified to three primitive partner which is Ep-Er, Eg-El and E1-E2 by means of the class of semantic attribute to verb. Here Eg is global E, and Ep and E1 are the other name of Eg. Er is the E in one sentence which is chunk extended into. El is local E in one chunk which is one sentence degenerated into.
And then implemented these aspects:
1. propose the class of semantic attribute for verb. These attributes are abstract of knowledge.
2. propose the range controlled by E, and define the range of Eg, El, Er, E2.
3. The E perceiving rule is done for Eg hypothesis and test;
4. The Sentence Category Deduction Rule (SCDR) is implemented for the E hypothesis which is Er, E2, or El.
HNC-MV is especially important for the Chinese NLP. HNC-MV have gotten some great improvement in the selection for the center of sentence, which is a most puzzle in Chinese NLP. This paper is also given a brief introduction of the Chinese Stand Determining System which is designed and developed based on HNC-MV.