CAJ | 학술논문

针对消费者网上购物时选择商品的用户体验不足等问题,设计了产品评论情感倾向性分类系统.系统首先对评论进行了分词,然后根据停用词表去停用词,分别采用CHI、IG进行特征选择,最后比较了使用不同的特征选择算法对文本情感分类结果产生的影响. 系统采用了Java Web相关技术实现了可视化,并对产品评论的分类过程进行了展示. 实验结果表明,有效的特征选择方法有助于提升推荐系统的性能.
침대소비자망상구물시선택상품적용호체험불족등문제,설계료산품평론정감경향성분류계통.계통수선대평론진행료분사,연후근거정용사표거정용사,분별채용CHI、IG진행특정선택,최후비교료사용불동적특정선택산법대문본정감분류결과산생적영향. 계통채용료Java Web상관기술실현료가시화,병대산품평론적분류과정진행료전시. 실험결과표명,유효적특정선택방법유조우제승추천계통적성능.
For the insufficient of the customers experience when shopping on-line, this paper designs a product review sentiment classification system. The system segments the reviews first-ly. Then according to the stop word list, the system removes the stop words. The system selects the features by using CHI method and IG method respectively. At last, the paper compares the effects produced by the text sentiment classification results using the different feature selection algorithms. The system uses Java Web related technologies to realize the visualization, and pres-ents the classification process on product reviews. Experiment results show that the effective fea-ture selection methods can help to enhance the performance of the recommendation system.