Eski Türkçe Metinlerin Günümüz Türkçesine Sadeleştirilmesi

Abstract

Changes in a language over time causes the text written in old times contain a lot of words that are not used at the present time. This makes it difficult for readers to understand old texts. The goal of text simplification task is to increase the readability and understandability of the text while preserving its content and meaning. In this thesis, it is aimed to reduce the complexity of the texts written in republican period Turkish with text simplification methods. Text simplification task is considered as an Intralingual translation problem. For simplification process; rule-based and statistical simplification models are developed. For rule-based model, an automatically built bilingual dictionary is used. For statistical model, a parallel corpus that is built from Nutuk is used. These systems are compared, as well as they are combined in a single hybrid system. The results are measured using BLEU metric that is used in the evaluation of machine translation systems. With this work, the complexity of old texts is reduced and the target audience of these texts is increased.