摘要
情感分析是得到文本正负中性情感的过程。该论文以电子商务评论为例,对其进行了情感分析。与电子商务中五星打分制相一致,评论最终被划分为:强烈贬意、一般贬意、中性、一般褒扬、强烈褒扬5级的情感强度。实现分为以下两个步骤:(1)句子级情感分析:通过复杂句句法模式和基于词典的算法得到评论中每个句子的情感倾向;(2)篇章级情感分析:即得到评论的总体情感倾向,采用了两种方法进行比较:一种为通过统计和比较评论中正面,负面和中性的句子的数量来得到评论,另一种则是通过SVM监督学习的方法,结果显示SVM方法性能更好。
Sentiment analysis is the process of detecting a piece of writing for positive, negative, or neutral feelings bound to it. This paper does the sentiment classification research on e-commerce product review. In order to be consistent with the five stars-grading in e-commerce, product reviews are classified into five-level： extremely negative, negative, neutral, positive and extremely positive.Thus, multi-level sentiment classification is added into the sentiment classification. There are two steps in general in the implementation： （1） sentiment analysis on sentences level, get the type and polarity of the sentence with our algorithm, which is based on words dictionary, patterns of complex sentences. （2）On document level, to classify the polarity-level, this paper makes use of two approaches： one is to compare the number of negative, neutral and positive sentences; another is using SVM supervised learning with the sentence type and position, which shows better performance.
出处
《软件》
2013年第9期65-67,94,共4页
Software
关键词
情感分析
多级
SVM
Sentiment Analysis
multi-level
SVM
