Iterative Feature Selection of Translation Texts for Translator Identification

Abstract

AbstractTranslation is an activity to transform linguistic information to another language. Translation product is a written text in a target-language (TL), which represents the result of a translation process, has been described by a comparison with the respective source-language (SL) text. The relation between the SL text and the TL text is a kind of the numerous and highly abstract models of equivalence (Koller 1978; 21983: 95; Ladmiral 1981: 393). In most cases, these models are very limited use for the practical translator. Problems in translating are caused at least as much by discrepancies in conceptual and texts use as by discrepancies in languages. Mankind has been engaged in transform for thousands of years, it affects the development of culture and language. “No language can exist unless it is steeped in the context of culture; and no culture can exist which does not have at its center, the structure of natural language.” Translation activities can promote exchanges between different cultures and languages, is also very important in the spread and development of religion. This study works for finding the discriminative features in Buddhist translation texts and other translation texts. Using these features can set up a training model for the identification of translator. Translator Identification is a process of examining the characteristics of translation texts to distinguish who is the translator. The similar processes had been used in authorship identification, writing forensics and similarity detection, which refer to statistical analysis of literary style.KeywordsDiscriminative FeatureIteration FilterLexical FeatureTranslation FeatureLiterary StyleThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.