Feature Extraction and Selection for Detecting Manipulation Online Review

摘要

With the proliferation of e-commerce, internet has become an excellent platform for gathering and sharing consumers' personal views on, preferences for, and experiences with products. With the popularity of text based communication tools, customers can easily express their opinions about purchased products or services. Generally speaking, the on-line reviews should be unbiased reflections of the consumers experiences with the products or services. However, some comments are biased "manipulation", which might reduce consumers' purchase intentions and bring a great damage to enterprisers. Although the existence of manipulation has been assumed widely, there are few results available in the literature for manipulation detection. This study aims to improve the performance for manipulation detection through feature selection. The study is divided into three parts. In the first part, we use conventional feature vectors obtained directly from the text files, and show that these feature vectors are in fact not useful for manipulation detection. In the second part, we adopt eight features recommended in the literature, and show that these features can improve the detection rate significantly. In the third part, we add three new features that can improve the detection accuracy further. A real case study of smart phone is used to illustrate the effectiveness of the proposed features.