The aim of this thesis is to propose the model of the Korean language processing for Korean-English machine translation. Based on Feature Computational Grammar(FCG), Korean language is analyzed and case alternation is adopted as the linguistic informa...
The aim of this thesis is to propose the model of the Korean language processing for Korean-English machine translation. Based on Feature Computational Grammar(FCG), Korean language is analyzed and case alternation is adopted as the linguistic information.
In chapter 2, I suggested the model of FCG and machine translation. First of all, I explained the position of FCG in the history of the linguistics and the system of FCG, especially the structure of lexicon and the procedure of parsing. Second, I proposed the model of Korean-English machine translation system. The system is presented with as follows.
(1) The model of Korean-English machine translation system.
◁표 삽입▷(원문을 참조로하세요)
In chapter 3, I argued for Korean case system and case alternation. I suggested the principle of case unification for the efficiency of the translation to parse and to generate the sentences. The principle of case unification is to unify high level case to lower level case. This means that high level case presents the general information and the lower case presents the specific information. Through the system I classified the patterns of case alternation and assigned the feature to verbs that are concerned in case alternation.
In chapter 4, I presented the parsing rules based on the information of case alternation. When one sentence is analysed, general parsing rules, i.e., subject assign rule, object assign rule and genitive assign rule, are applied. And then case alternation feature(caf) is searched. If caf is discovered, the case alternation parsing rule is applied to the sentence.
Case alternation parsing rule is followed the principle of case unification. The direction of case unification is presented with as follows.
(2) the direction of case unification
① structural case - structural case
i → lul, lul → ui, i → ui
② structural case - inherent case
lul → eise / lo / ei / eikei / wa / ei taihai,
i → ei(kei) / lo
③ inherent case - inherent case
ei → lo / wa
Using the information of the case alternation, I proposed the case alternation parsing rule and the preposition generating rule. Finally, I exemplified the process of the translation with sentences.
This thesis suggested the methodology of linguistic study for language information processing, especially Korean-English machine translation. Korean case alternation information is useful to parse the Korean sentences and to generate English sentences. If the further studies, i.e., ellipsis and the system of meaning, is achieved, the NLP(Natural Language Process) of Korean will be much more developed.