Russian-oriented Information Processing, Machine Translation Experiment

In order to develop a practical Machine Translation (MT) system, we must have an integrated and logical translation theory in place that is adaptable to computer processing. However, after studying the history and present status of Russian MT, as well as the issues of linguistic and program assurance with regard to the design of MT systems, we found that there is not yet a single well-developed and effective theory for Russian-Chinese MT in China. In view of this fact, the thesis focuses on the ways to improve the accuracy of MT of Russian texts, especially military texts.On the basis of studying the application of foreign rule-based Russian-Chinese MT software, we discovered that, if we continue to use traditional grammatical rule-based automatic translation methods, it will be extremely difficult to significantly improve the accuracy of MT of Russian military texts. After comparing a number of approaches to MT, we proposed that CAT (Computer Aided Translation) systems which are based on Translation Memory technology would be the optimal choice, from the perspective of engineering. On the one hand, military texts feature a high similarity in terms of vocabulary, sentence patters, and structures, and a high repetition rate of terms and sentences. This is where TM works best. On the other hand, translation of military texts usually is required more timely and accurate. This is where other MT approaches fail.The theoretical significance of the present thesis lies in: a systematic summary of the history and state-of-the-art of Russian MT studies; an objective analysis of the future development of MT; and a proposal of the core technology and realizing methods for TM-based CAT systems. Our study will no doubt help to expand the scope of MT application studies, to improve the performance of MT systems, and to accumulate some experience in research and development of Russian-Chinese MT systems for military uses.The thesis consists of an Introduction, Chapters 1-4, a Conclusion and a Bibliography.The introductory part presents the motivation, objective, significance, and creativeness of this study as well as the data sources of the thesis.Chapter One, an overview of MT in the context of computerization in global science, first draws a distinction between translation and MT, then summarized the history and the present status of MT studies, and finally elaborates on a number of basic approaches to MT studies.In Chapter Two, which is a parsing of MT systems in Soviet-Russia, we begin with an introduction to the linguistic model theory of"Meaning ? Text". Then, we move on to analyze several typical MT platforms, including ETAP, RETRANS and LINGTON, and point out the experiences accumulated by and deficiencies in each of them regarding Russian text processing.Chapter Three is an assessment of MT systems. In this chapter, we first introduce the theory and methods for the assessment of MT. Second, we apply such MT assessment theories to conduct a comprehensive evaluation of three online Russian-Chinese MT engines (including the Russian-Chinese Translator Ehan Yitong) with the help of online MT resources.Chapter Four is the design and realization of a Russian-Chinese CAT system for military official documents. In accordance with the style characteristics of military texts and leveraging on the benefits of TM-based CAT systems, we put forward a discourse upon the linguistic and program assurance for the development of these MT systems.The concluding part summarizes the theory advanced in this thesis and its possible applications. In addition, a number of pending issues are pointed out for further studies.