Open AccessSentiment Analysis For Product Reviews Based on Deep Learning
Author(s) -
Shaozhang Xiao,
Hexiang Wang,
Zhi Ling,
Lanfang Wang,
Zhaoxia Tang
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1651/1/012103
Subject(s) - computer science , naive bayes classifier , sentiment analysis , artificial intelligence , product (mathematics) , word2vec , machine learning , popularity , word (group theory) , logistic regression , feature (linguistics) , the internet , stop words , data mining , information retrieval , support vector machine , world wide web , preprocessor , mathematics , psychology , social psychology , linguistics , philosophy , geometry , embedding
With the popularity of the Internet and e-commerce, the sentiment analysis of text can help users to quickly and accurately obtain effective information they are interested in from massive product reviews to purchase satisfactory products. In this paper, a sentiment analysis system for product reviews was designed based on deep learning, and the digital and electronic products on Jingdong Mall with at least 100,000 reviews were crawled as the training data set. After data pre-processing operations such as word segmentation and removal of stop words for product reviews to remove useless features, the feature vectors were constructed based on the bag of words model and word2vec method, and then three classification algorithms, namely LSTM, Naive Bayes and logistic regression were used to model reviews. The LSTM algorithm is significantly superior to Naive Bayes and logistic regression algorithm in the training stage, and provides a reliable reference for the analysis of product reviews.