Machine Translation Based on HNC theory

by LI Yao-yong

Directed by HUANG Zeng-yang


Machine translation is very important but also very difficult. It has a lot of applications. What machine translation needs mostly is a new linguistic theory which can analyzes the natural language deeply and can be implemented in computer easily. The HNC theory, founded by Prof. Zengyang Huang, is such a linguistic theory. In this report we study the machine translation based on HNC theory. The main content of the report is as the following.

Give a complete description of language concept space, which includes the Hierachical Network of Concepts, the combination of concepts, sentence, discourse, and paragraph, etc..

Furthermore, we give a definition of the discourse in language concept space and in language space, respectively. Present the prominent characteristics of HNC theory by comparing the HNC theory with the syntactic and semantic theory.

Explain the strategies of the HNC based machine translation.

Propose an example-based machine translation scheme by using the analysis result of HNC theory.

Describe two programs. One is to align the sentences for English-Chinese corpus. Another is to transform the order of the main chunks of the Chinese basic action sentence into that of English.


Keywords: machine translation, natural language understanding, the HNC theory,
syntactic and semantic theory.