The Research Of The Marked-based Chinese To English

The Example-Based Machine Translation (EBMT) system can efficiently make better translation result for a given domain, otherwise the system can learn translation knowledge by itself and it can be developed in a relatively short period. The research on the Example-Based Machine Translation model is of great theoretical and practical significance to the Corpus-Based Machine Translation and, further, to the Natural Language Processing.A basic problem for EBMT is how to construct the translation unit base. Although translation unit base is important to EBMT system, the research on it is not adequately study. In this dissertation, translation unit is extracted with"Markers", which is proved to be effective.In this paper, the thesis focuses on the following issues in adopting Marker in Chinese-English EBMT: 1. Defined the Chinese Marker. After introducing the English marker recently used, the Chinese marker is identified and the definition is validated via a survey on a large corpus;2. Introduced the method of extracting Marker-based Chinese-English translation unit. After identifying the Chinese marker, the thesis proposes three algorithms to extract Chinese-English translation unit using Marker hypothesis and word alignment, namely, Marker Improved Word Alignment Based method, Marker Based Segmentation Alignment method and the Merge algorithm of them. These method adopts the syntactic infection without introducing the errors in deep sentence analysis;3. Introduced the evaluation method for Chinese-English translation unit acquired by Markers, and enhance the performance of EBMT system. In this paper the evaluation method for Chinese-English translation unit merge with the system integrated evaluation is introduced. First, the principles for evaluation of translation unit and the quality classified are introduced, based on this the performance of translation unit which extracted by the method introduced in this...