| According to the "42nd Statistical Report on Internet Development in China"issued by China Internet Network Information Center,in the first half of 2018,the national online retail sales reached ￥4.08 trillion,with a year-on-year increase of more than 30%.Unlike physical shopping,consumers need to get product information online and make shopping decisions when making online purchases.The information conveyed by online reviews is also particularly important.However,along with the continuous expansion of e-commerce scale,the number of online reviews is huge and the quality of the content is uneven.Therefore,how to quickly extract effective information from online reviews has become an urgent problem.To fix this problem,we propose a method for review feature extraction and a review ranking model.Firstly,we propose two feature extraction methods:review attribute features and emotional features.In view of how to easily and correctly mine effective information from massive reviews,this paper starts with the e-commerce online review mining,and optimizes the traditional review feature extraction method by using Word2vec model and HowNet sentiment dictionary.Secondly,we propose a ranking model for the usefulness of e-commerce reviews,From the perspective of the source of review and the quality of review that affect the usefulness of online reviews,we propose an indicator system with seven indicators,combining the concerns of consumers in browsing reviews and the characteristics of e-commerce reviews.Based on this,we use the fuzzy analytic hierarchy process to calculate index weights and construct our review ranking model.For both search-based products and experiential products,the research results show that the model can quickly screen out high-quality reviews,and the readers'perceptions are better than traditional e-commerce platform ranking.In summary,our review ranking model is very helpful to enhance the consumer's shopping experience. |
PDF Full Text Request