Research On The Usefulness Of Online Product Evaluation

This paper,starting with two aspects of consumer and commodity evaluation,uses the beautiful soup crawler tool to crawl all the evaluations of the Amazon Web Product trisomy 1 and all the comments of each commentator.This article crawled to1474 users,20853 evaluations,and all evaluations of each user.Then we use the Jieba word segmentation tool and the Kazakhstan disable word dictionary to process the crawling data,and calculate the text similarity by using the LSI model for the processed data.The results show good effect.By calculating the average value of the similarity between each evaluation document of the commentator and the total evaluation document of the commentator,it is supplemented by the same day.The number of comments is large enough to determine whether the user is a water force.After eliminating the water army evaluation,we model all the remaining three body 1.This paper adopts linear regression,logistic regression and multiple classifier methods in machine learning.By contrast,the root mean square error of the random forest regression model and the linear regression model is lower and the effect is better,but the linear regression model is tested.It is not as good as the random forest model,so this paper uses random forest model to classify the commodity evaluation.The result shows that the model has a good effect and the product evaluation in the front row is longer.At the same time,it provides information that other goods are not mentioned or mentioned in a small amount.It has a short content and a small amount of evaluation.Most of the information provided by them has been mentioned by other evaluations,so the usefulness of the users is very low,which is in line with our expectations.