摘要
通用型情感分析技术较为成熟,却有一定的领域局限性。针对时下热门的网络订餐平台评论进行挖掘,引入搜狗细胞词库,修正错别字,并对网络术语和颜文字进行转换,在通用HowNet情感词库的基础上,自建网络订餐关联逻辑词库,使用TF-IDF加权算法,最终构建出专属网络订餐情感词库,提升了情感评分准确度,同时分析出网络订餐的主要属性,并对此进行更深层次的意见挖掘。
The universal sentiment analysis technology is more mature,but there are certain limitations of the field,this paper reviews the popular nowadays network ordering platform for mining,the introduction of Sogou cell thesaurus,correcting typos,and the network terminology and color text conversion,based on the general emotion lexicon HowNet,self built network ordering logic Association Thesaurus the use of Term frequency–inverse document frequency（TF-IDF）weighted algorithm,and finally construct the exclusive network ordering emotion lexicon,enhance the emotion score accuracy,at the same time analysis of the main attributes of the network order,and make deeper digging opinions.
出处
《软件导刊》
2017年第12期33-35,共3页
Software Guide
关键词
网络订餐
情感分析
专属词库
意见挖掘
network ordering
sentiment analysis
thesaurus
opinion mining
