Evaluation Of Consumer’s Perceived Quality Based On Text Mining

Consumer’s perceived quality is the subjective judgment of consumers on the pros and cons of products through subjective perception methods such as seeing,hearing,hearing,which has a direct impact on consumers’ purchase and evaluation behaviors.Scientific evaluation of consumer perceived quality will provide decision-making reference for enterprise product quality planning and improvement,thereby enhancing product competitiveness.Existing consumer perceived quality evaluation methods have defects such as high cost,few samples,long feedback period,and low analysis efficiency,they are not suitable for today’s rapidly changing consumer demand,nor can they meet the needs of enterprises for rapid iterative product development.In this regard,this article considers the basic condition that the Internet community comment text contains a wealth of consumer opinions,and comprehensively uses text mining technology to obtain and analyze the emotional information in the massive unstructured text on the Internet,the efficient,automated,and accurate analysis of consumer opinions can be carried out,and then consumer’s perceived quality is evaluated.Specifically,the research of this article mainly includes the following three parts:(1)In order to solve the problem of extracting effective texts containing consumer opinions from massive multi-source texts,a method of opinion sentence recognition based on support vector machines is proposed.First,high-dimensional candidate feature vectors are constructed through the joint extraction of product attributes and opinion words and the design of multiple feature templates such as 2-POS and indicative words.Secondly,the features are filtered by chi-square test and then put into the support vector machine classification.Finally,the method is verified on the experimental corpus,and the model shows excellent classification performance by adjusting the screening threshold.Experimental results show that this method can accurately and efficiently identify the difference between opinion sentences and non-opinion sentences,and lay a foundation for extracting emotional information in opinion sentences.(2)Aiming at the small-granularity,multi-label,and imbalanced attribute-level sentiment analysis problems in consumer perceived quality evaluation,an end-to-end sentiment analysis method based on transfer learning is proposed.First,a feature extraction method is designed that integrates three pre-trained semantic models of BERT,ERNIE,and Ro BERTa to achieve a comprehensive characterization of the text features of Chinese comments.Secondly,a multi-layer perceptron with two hidden layers is used and two loss functions are designed.Finally Experimental corpus is used to verify the effectiveness of the method.The experimental results show that the method has strong generalization ability,and the product attributes and emotional tendency information obtained through sentiment analysis are used as the input of the perceived quality evaluation method.(3)In order to realize the scientific evaluation of consumer perceived quality in big data scenarios,a consumer perceived quality evaluation method based on emotional information is proposed.First,through the inductive analysis of the concept and connotation of the perceived quality,a mathematical definition of the level of perceived quality is proposed,the mathematical relationship between the value rate difference-the perceived quality-the emotional intensity is established.Secondly,the indicator weight design scheme that considers the difference of opinions,the mention ratio and the negative ratio is proposed.The evaluation results under the experimental corpus are obtained.Finally,the evaluation method is applied in the three links of building the house of quality.The analysis shows that the evaluation results of this method are in line with the actual product quality performance and the general opinions in the market,evaluation methods can help companies identify quality advantages and shortcomings,and provide decision-making basis for companies to improve product quality and product design.