CAJ | 학술논문

随着国内电子商务的蓬勃发展，各大电商网站评论量飞速增长，如何从Web大量的商品评论中挖掘出价值信息并进行有效分类对消费者和生产厂商决策有重要的影响。传统分类方法能有效地抽取商品评论中的特征及观点，但对中文评论分类仍存在一些不足。为了进一步提高商品评论分类有效性，首先，综合前人研究提出一种基于评论长度的特征提取方法，提高分类准确率；然后，设计了评论样本自动标注方法，并构建评论的有效性分类模型，改善分类效率；最后，以京东商城上爬取的1710条商品评论为例，对提出的特征提取与自动标注方法进行验证。实验结果表明，根据该方法，评论分类准确率得到明显提高。
수착국내전자상무적봉발발전，각대전상망참평론량비속증장，여하종Web대량적상품평론중알굴출개치신식병진행유효분류대소비자화생산엄상결책유중요적영향。전통분류방법능유효지추취상품평론중적특정급관점，단대중문평론분류잉존재일사불족。위료진일보제고상품평론분류유효성，수선，종합전인연구제출일충기우평론장도적특정제취방법，제고분류준학솔；연후，설계료평론양본자동표주방법，병구건평론적유효성분류모형，개선분류효솔；최후，이경동상성상파취적1710조상품평론위례，대제출적특정제취여자동표주방법진행험증。실험결과표명，근거해방법，평론분류준학솔득도명현제고。
With development of e-commerce,the major e-commerce website product reviews increased rapidly. How to utilize the large number of reviews and classify it efficiently make an important impact on manufacturers’decisions. Traditional classification method can effectively extract the product opinions,but not for Chinese reviews. In order to improve the effectiveness of product reviews classification,firstly we present a kind of feature extraction method based on the length of feature to improve classification accuracy; secondly we design a comment sample automatic annotation methods and construct the classification model;and finally,take 1710 product reviews from Jingdong Mall and proposed this methods. The results show that this method could improve the classification accuracy significantly.