Sentiment analysis of comments of wooden furniture based on naive Bayesian model

Abstract

In this paper, the sentiment analysis method based on machine learning is used to analyse the commodity reviews of wooden furniture which are sold well on MeiLeLe Furniture Website, and then extract the potential and valuable information generated in the interactive process by users and businesses, such as user attitudes and emotional tendencies, user preferences for the commodity attributes of wooden furniture to achieve the goal of improving the industrial design process of wooden furniture manufacturers, providing effective suggestions for merchants, improving the quality of wooden furniture commodities, and then improving the export level of our country. In order to obtain the information, the "Octopus Collector" is used to crawl the commentary information of wooden furniture commodities. The original data are needed to be cleaned and labelled manually before training the emotional tendency. In this paper, naive Bayesian algorithm is used to calculate the prior probability by using the training data with manual labelling, and a text classifier is constructed to classify the emotional tendency of the experimental data. Afterwards, it visualizes the positive emotional tendency of the commentary data to understand the commodity attributes that consumers are more concerned about when buying wooden furniture online. The experimental results show that consumers pay more attention to the quality factors, price factors and appearance factors of wooden furniture. Effective suggestions can be put forward to the merchants in these aspects.