My task is to create a simple statistical machine translation engine, which, after appropriately processing the provided bilingual training set, will be able to translate whole sentences from one language to another, and to examine the quality of the translation generated by this program.
In my thesis, after the introduction, I briefly review the history of machine translation, then describe the general characteristics of the known translation methods, the criteria of their application, and how I started constructing my own translation program considering these issues. As we will see, according to its general model, statistical translation can be divided into three distinct problems. In the next chapters we will discuss these problems. I start these chapters with the short, but sufficiently thorough review of the literature, where I show the widespread solutions, then - if it is necessary or helps the understanding - present the carried out measurements and the details of implementation.