Facet Based Estimation Polling From Customer Reviews

Reputation-based belief systems are broadly used in e-Trade applications, and response ratings are aggregated to figure out traders’ reputation grades. The “all good reputation” problem, however, is prevalent in recent reputation systems. Reputation grades across the web are commonly high for traders and it is difficult for potential customers to choose accurate traders. This work is based on the observation that customers often express their viewpoints explicitly in free text response comments. We propose a system that figure out Comm-Trust for trust evaluation by drilling response comments. We propose multidimensional belief model for estimating reputation grades from user response comments. We propose an algorithm for mining response comments for dimension ratings and weights, combing techniques of natural language processing, opinion mining and topic modeling. This research work is mainly based on the first piece of work on trust evaluation by mining response comments