Research On Identification And Classification Of Customer Requirements Considering Online Comments With Time Efficiency

In the Web2.0 environment,the relationship between enterprises and users is increas-ingly close.Reasonable use of users' online comments can effectively help enterprises identify users' requirements,promoting product improvement,and further improving the resource allocation efficiency and user satisfaction.Traditional customer requirement re-search is given priority to user interviews and expert interviews,which is much qualitative,such as Kano model.In order to make up for these shortages,this thesis proposes the con-cept of time efficiency and ways to screen useful comments from the product perspective,performing product feature extraction and opinion mining based on these useful comments,thus constructing the model of identifying and classifying customer requirements.The main research contents include the following aspects:·From the product perspective,the concept of comment time efficiency is proposed and the model of identifying useful comments considering this characteristic is con-structed.The content of comments is taken as the usefulness evaluation index,based on which the comments are divided into two stages to construct a useful comments screening model respectively,furthermore,index evaluation and SKC clustering al-gorithm are comprehensively used to determine useful comments.In this thesis,the time distribution equalization rate and accuracy are treated as evaluation index,and by comparison,the number of useful comments is distributed balanced in time and the accuracy of selected useful comments is much higher for the previous staged model·Based on the useful comments,the existing product feature set and sentimental dic-tionary are expanded and the customer requirements are exploded based on these dic-tionaries.By training Word2Vec model based on online comments corpus as well as related documents associated with products,the existing dictionaries are further devel-oped.On the basis of improving the existing product feature database and sentimental dictionary,this thesis makes use of dependency syntax analysis as well as feature-sentiment rules to extract feature-sentiment pairs,refines and classifies the product features to identify specific customer requirements and acquire attention and satisfac-tion of corresponding requirements·PF-Kano model is proposed to classify the identified requirements.PF-Kano mod-el is put forward based on the attention and satisfaction of requirements to sort the requirements.According to the classification results in view of staged and unstaged comments,it is found that the staged modeling of the comments can effectively dig out the attractive requirements and one dimensional requirements.In the light of the classification of requirements,the improved engineering characteristic value can be solved by combining with QFD with maximizing customer satisfaction and minimiz-ing improvement cost,so as to provide guidance for enterprises to improve products.