구글 번역기를 통해서 본 통계기반방식 번역기의 특징

A Study on Google Translation

This study aims to review key features found at Google French - Korean machine translation system (a translation system based on statistical methods). In order to analyze the Google machine translation process, I first reviewed the theoretical aspect of the machine translation system based on statistical methods. Then, I randomly selected some articles from daily newspapers or weekly magazines from Yahoo France, had them translated through Google machine translation system, and analyzed those results. I started with the analysis of the original language focusing on synthetic/functional/dialectic units. Then, I abstracted key words by syntax. As a result, I concluded that although Google machine translation system occasionally generates some natural expressions, in most of the cases, it has difficulties in accurately grasping the syntax structures of the original context and therefore has difficulty in understanding/considering full context and semantic field, which leads to translated materials with less appropriateness and less accuracy.