Quantifying Customer Review by Integrating Multiple Source of Knowledge

Abstract

The recent emergence of a large volume of customer reviews on e--commerce web sites has raised concerns on the provision of intuitive and comprehensive reputation comparisons of feature dimensions. In this paper, we propose and implement a product reputation mining prototype system. A multiple-knowledge based F-O pair extraction model, which is the center piece of our work, is presented for conducting analyses toward deeper sentence-level comprehension of sentiments in customer reviews. Experimental results demonstrate the effectiveness of the proposed method.