文章基本信息

标题：A Novel Cluster based Over-sampling Approach for Classifying Imbalanced Sentiment Data
本地全文：下载
作者：Jing-Rong Chang ; Long-Sheng Chen ; Li-Wei Lin 等
期刊名称：IAENG International Journal of Computer Science
印刷版ISSN：1819-656X
电子版ISSN：1819-9224
出版年度：2021
卷号：48
期号：4
语种：English
出版社：IAENG - International Association of Engineers
摘要：In social media, sentiments or online reviews are important information sources for product purchase decisionmaking. Usually, both favorable reviews and negative reviews from social media users may significantly impact companies' trade. Therefore, effective and efficient methods for sentiment classification have currently become the most concerning issues for companies. One of the best useful methods is machine learning. However, when employing these sentiment classification methods, the class imbalance problems which are caused by imbalanced data need to be considered since the performance accuracy of the majority class is often higher than that of the minority class. Therefore, this study aims to propose the Modified Cluster based over-Sampling (MCS) method which is expected to be a novel cluster based over-sampling method, for imbalanced sentiment classification. Some UCI data sets and three sentiment classification cases including two actual cases of imbalanced text comments regarding MP3 products and electronic commerce services were our research data which were employed to illustrate the usefulness of our proposed method. The experimental results indicate that our proposed method is superior to conventional re-sampling methods, such as over-sampling, cluster-based sampling, and decision tree algorithm.
关键词：Class imbalance problems;Sentiment classification;Cluster based sampling;Social media;Over-sampling;decision trees