| Whatever in the fields of investment,management,or design,there are lots of decision making problems that include various complex factors for the pursuit of optimal product alternative.Similarly,it is significant for product development to recognize customer requirement and design the best product.Customers express their requirement through product reviews,which contain customers' likes,dislikes and exceptions.However,the process of evaluation is uncertain as the external environment factors and internal human factors.In spite of the interference of complex environment factors,the fuzziness of perception and expression,the difference of preference and the dynamic of reviews all have effects on the accuracy of decision making towards customers.Therefore,the customer reviews obtained by enterprise are not real enough.Based on the problems,this thesis put forward a product reviews decision making mechanism considering the unbalance and dynamic evolution of reviews to acquire steadier customer reviews.The main contents were as follows.(1)Customer reviews measurement combing with preference and unbalanced linguistics was proposed.First off,the linguistic term set was designed by preference comparison that included the comparison with exception products provided by customers,so the perception and expression of customers can be strengthened to decline the vagueness of their reviews.Then the unbalanced linguistic scale was designed to quantify customer reviews and present the unbalanced sentiment changes of customers.In addition,binary uncertain linguistic terms were used to help customer evaluate product more accurately and avoid the integration problems,such as distortion and overflow.At last,a customer review measurement structure was established,which was verified its feasibility and advantage compared with balanced scale through case study.(2)Customer reviews evolution model was set up.Due to the bounded rationality of customers,they would contact with others and amend individual opinion to make a relatively stable assessment.Therefore,three influence factors of interaction and decision making: confidence threshold,the helpfulness of reviews and negativity bias were proposed in light of opinion dynamics,which formed the dynamically interactive customer reviews mechanism and customer reviews evolution model.Finally,a numerical simulation proved the function of the model that could predict the changing of reviews and the tendency of group customer reviews,as well as the superiority compared to static reviews.(3)Product multi-attribute decision making mechanism grounded on dynamic evolution and unbalanced preference was presented.To start with,the corresponding normalization methods were provided according to the comparative standards of attributes.Product attributes could be divided into objective or subjective attribute,where the former could be estimated directly,while the latter was estimated by customers and the comparative review of real products would evolve.Then the attributes preference matrices were composed of every customer reviews,which would be integrated and analyzed the optimal alternative.Since there was a compatibility problem between binary uncertain linguistic terms and customer reviews evolution model in the interaction,the standardization method was utilized,and the model was modified to calculate the interval number.At last,the case study revealed the practicality of the decision making mechanism and its advantages through two comparison analyses.Faced with inaccurate quantification and the lack of timeliness,the customer decision making mechanism towards product reviews based on dynamic evolution was proposed,which presented unbalanced perception and expression of customers,as well as their behavior of dynamic decision making and review updates.Consequently,the mechanism can analyze more real and reasonable product reviews,which is practically significant for company to decide on the best alternative that truly meets the customer requirement,and develop most popular product in the market.There are 16 figures,9 tables,and 100 references in the thesis. |
PDF Full Text Request