In recent years, cross-lingual sentiment classification has drawn much research attention. Many research studies have
been conducted in this area. These research studies are based on the use of annotated data in the source language (always
English) to compensate for the lack of labelled data in the various target languages. Most approaches have focused on
resource projection from one language to another with few sentiment resources. For example, Mihalcea et al. [21] generated
subjectivity analysis resources into a new language from English sentiment resources by using a bilingual dictionary. In other
works [2,3], automatic machine translation engines were used to translate the English resources for subjectivity analysis. In
[2], the authors showed that automatic machine translation was a viable alternative for the construction of resources for
subjectivity analysis in a new language. In two different experiments, they first translated the training data of subjectivity
classification from the source language into the target language. They then utilised this translated data to train a classifier in