Research On Commodity Evaluation Based On Emotion Analysis

In the rapid development of the Internet era, Jingdong, Tmall and Amazon and other online shopping playing an increasingly important role in people's lives,online shopping has become an important way to buy. In the online shopping, people often through three ways to obtain commodity information, pictures, product parameters and comments. Commodity information hidden in the picture for the seller has been beautified. Product parameters may be too specialized, not everyone can understand. Comment data is readable and rich, so many sellers will refer to the review data before buying. According to the survey only about 18%of the buyers do not look at other comments before the buying. But the number of comments is huge, therefore, the focus of this paper is how to put these comments effectively organize the establishment of commodity evaluation model to help customers select goods to help sellers to improve products. There are two main types of commodity evaluation models, one is based on product parameters, this method think the product quality is entirely determined by the hardware. But a good or bad products in addition to consider the hardware should pay attention to the use of the customer, of course, saving time and effort is the advantage of the method. The other is based on a questionnaire survey,this method puts the customer's sense in the first place, but the process of design, distribution and finishing of the questionnaire is time consuming and labor intensive. Establishment of the product evaluation model based on emotional analysis is saving time and effort, and the evaluation results are also very fit customer experience.This article mainly in the establishment of the evaluation model of goods to complete the following work: 1. Data acquisition and cleaning. Using python to crawl the comment data on the e-commerce site, Customize the corresponding crawl rules. In the process of crawling the data may be repeated crawling,there is a similarity between the comments of false comments, meaningless comments are the same. In order to reduce the impact of the above three cases on the final evaluation model,the author used the text similarity calculation to clean up the review data. 2. Emotional unit extraction. In this paper, the use of dictionary matching based on the emotional unit to extract the model In order to irregularly commented data into standardized questionnaire data. The author also uses the Apriori model to extend the positive and negative evaluation dictionary provided by the HowNet. The final evaluation finds that the emotional model has reached 90% accuracy for the emotional unit extraction of the short sentences. And finally found that the emotional model for the short sentences of the emotional unit extraction accuracy rate has reached 90%. 3. The establishment of commodity evaluation model. First, we use the LDA model to analyze the comments to identify potential topics in the comments to establish the index system. Then in order to make high-quality comments have a greater impact on the final evaluation results,the validity model of the evaluation was established. Finally, the fuzzy evaluation model is used to evaluate the goods, The construction of fuzzy matrices completely dependent on the results of the validity model.In the fourth chapter of the paper, the author uses three MIUI phones to comment on the establishment of evaluation models based on emotional analysis. Through the evaluation results can be known in the battery capacity and mobile phone screen MIUI max slightly better,consistent with the results of the parameters. In the camera function,simply consider the phone parameters, MIUI 5s should be the first millet, But MIUI 5s lost to MIUI 5 by the evaluation results,Through the analysis of the comments found that MIUI 5s camera will appear unable to focus, slightly shake will make photo is not clear and the pixel is not enough.By analyzing the evaluation results can be found,the author combined with web spider,emotional analysis techniques and statistical knowledge, establishment of the product evaluation model based on emotional analysis is saving time and effort, and the evaluation results are also very fit customer experience.