Quantification of emotional values for attributes from sentiment analysis based on mobile phone reviews

Abstract

E-commerce online reviews are commonly short in length, lack of contextual semantics, and are mixed with explicit emotions and implicit emotions. The subjects of implicit emotions are not easy to be extracted and be defined. Now product-oriented attributes are used to define implicit and fine-grained emotions with the help of the domain emotion dictionary that is constructed by extracted explicit attribute-emotion pairs through using a variety of part-of-speech rules. And implicit emotion and attribute pairs for product "mobile phone" are mined with the help of explicit attribute-emotion pairs and probability calculation method. At last, explicit and implicit emotion values are calculated by considering degree and negative modifiers. The experimental results show that proposed method is effective and has certain commercial value. And the process method is easy to extend to other product fields.