Cleaning the Textual Variables

• Deleting Descrition
• Replace common French HTML entities with their characters
• Language Recognition with langdetect
• sent_tokenize with french language (?!?)
• Removal of special characters and numbers except [^a-zA-ZéàèêëîïôùüçÀÉÈÊËÎÏÔÙÜÇ]
• word_tokenize --> dic
• Stemming with detected language and SnowballStemmer in en/de/fr
• join dic to string

Proceed to

• Simple_Prediction_Model