3.3　Pretreatment of Data

In order to make sure the integrality，all the texts are com pletely embodied in the corpus although texts are notexactly of the same size as each other.Before annotation，each text is cleaned and saved as a separate plain text（.txt）encoding in UTF-8.The head which includes the basic information such as title（name of tourist attractions），source language，target language，country（location of tourist attractions），type of landscape（natural or cultural），size（how many words in the text）and so on is inserted at the very beginning of each text.(https://www.daowen.com)