The Design And Implementation Of The Spam Review Detection System

Contents of the abstract. Times New Roman. Online reviews plays a very important role in today’s e-commerce sites. It provides reviews made by other customers for potential customer. So, potential customers can have a comprehensive understanding of the product. However, no matter domestic e-commerce sites, or foreign e-commerce sites, the spam reviews are widespread. This often misleads customers to purchase the product of poor quality.According to our investigation about spam detecting system all over the world, there is no such a system that can widely apply to domestic and foreign each big e-commerce sites of the system to detect fake reviews. Through summing up experience about spam detection and using some text mining methods, we build our spam detecting system. Our system is mainly composed of data preprocessing module, data analysis module, the classification of clustering module modules and data evaluation, etc. in order to make our system to adapt to as many commerce sites as possible, our pre-process module mainly designed by the dataset from some big commerce sites. Meanwhile, in order to analyze the ubiquitous properties, we designed different data analysis module for the different attributes. The analysis module design for many aspects of the text feature, rating characteristics, emotional features, user background characteristics, review time, so we make the system be able to deal with more complex data, and enhance the robustness of the system. In classification and clustering module, we use some Data Mining algorithms and several improved algorithms to analysis the reviews. Finally, we were able to use the evaluation module to evaluate the analysis result.