Methods of Customer Requirements Feature Extraction on Product Reviews

摘要

This paper regards product reviews on e-commerce sites as the research object. Firstly, to divide the basic expression of the customer requirements into Parameters type, Boolean type and Emotion type. Then through the method of events extraction technology, it puts forward a customer requirements feature extraction method. This method is based on the reviews text pretreatment, and then through product feature word identification, requirements trigger word recognition and basic needs recognition, and ultimately get the structured customer requirements information. Experiments show that this method is effective.