Method For Product Selection Based On Online Reviews

People's life style has been deeply influenced by rapid development of E-commerce.The channel of getting information and the way of shopping are all turning to online environment gradually.The scale of e-commerce platforms are developing rapidly in the age of big data.Consumers need more efficient product selection methods which would help making decisions quickly for the massive kinds and amount of products.Some of the e-commerce webs already began providing some supports for the consumers selecting products such as generalized products ranking and individualized products recommendation currently.But these tentative exploration need further perfection for didn't integrating and providing the online reviews to help consumers making decisions.Consumers' opinions and shopping experience of product are included in online reviews,which make online reviews being one of the very important references of making decisions when consumers shopping online.Thus it can be seen that products selection methods based on online reviews in e-commerce environment is a theoretically foundational and practically valuable research topic.On the basis of a comprehensive review of the research status and development of consumer products selection model,online reviews usefulness,attribute extraction and sentiment analysis,this thesis made thorough research on the problems about products selection based on online reviews.The purpose and significance of the research are in the theory aspect,method aspect and practice aspect.In the theory aspect,products selection problem based on online reviews has been concluded and summarized and the generalized problem description and research framework has been proposed.A solid theoretical foundation has been laid for further research of products selection problem in e-commerce.In the method aspect,thinking of products attribute extraction based on online reviews has been proposed.The online reviews has been presented in the form of decision making initial information.In addition,the products ranking and selection method has been proposed,which could be not only a scientific guidance and necessary support of solving the problem of consumer's products selection in e-commerce environment but also a model reference of products selection supporting system.In the practice aspect,a specific example analysis of laptop selection in JD.com has been given.It provides a beneficial reference method to the products selection problem concerning consumers' online reviews.This paper has deeply researched of products selection problem based on online reviews and conducted the following three aspects.(1)The problem description and research framework of products selection problem based on online reviews has been proposed.This paper has given the specific mathematical description and proposed the specific research problem.Moreover,the main research framework of the problem has been proposed.The works mentioned above has laid a foundation to extended issues and applications.(2)The attribute extraction method and weight value determination method has been proposed.First,on the basis of data acquisition and pre-processing of products and online reviews,calculating the utility value of each reviews according to reviewers' credit,reviews'number of supporters,and reviews' timeliness.Secondly,extracting the products' dominant attributes included in the reviews which the consumers are care about by Apriori algorithm and extracting the products' recessive attributes by semantic analysis.Determining the final attribute set and calculating the weights of the attributes in the event that concerning the utility value of reviews.(3)The products selection method based on stochastic dominance has been proposed.Quantizing the consumer's sentiment preference contained in the online reviews as the score matrix of sentiment preference.Transforming the score matrix into random information to calculate the stochastic dominance relationship between each two of the products and using PROMETHEE II to provide the ranking in each attribute and overall ranking,which would be effective references for the consumers' selection.The method mentioned above could retain the consumers' sentiment attitude contained in the text information as far as possible.The products selection case of selecting notebook computers in JD.com has been provided,which would indicating the feasibility and practicability of the method proposed in this paper.