基于word2vec的酒店评论情感分类研究  

Research on Emotional Classification of Hotel Comments Based on Word2vec

作　　者：谢宗彦 黎巎[2] 周纯洁 Xie Zongyan;Li Nao;Zhou Chunjie(Beijing Key Laboratory of Information Service Engineering,Beijing Union University,Beijing 100101,China;Beijing Collaborative Innovation Center of eTourism,Beijing Union University,Beijing 100101,China)

机构地区：[1]北京联合大学北京市信息服务工程重点实验室,北京100101 [2]北京联合大学北京市旅游信息化协同创新中心,北京100101

出　　处：《北京联合大学学报》2018年第4期34-39,共6页Journal of Beijing Union University

基　　金：国家自然科学基金青年项目(41101111);北京联合大学人才强校优选计划项目(BPHR2017CS15)

摘　　要：酒店评论客观真实地反映了客户对于酒店和服务的感受,基于酒店在线评论构建了一个情感分析的模型。首先根据设计提取评价对象-情感词的句法规则,得到细粒度的分类特征信息;然后采用word2vec和TF-IDF算法进行评价对象-情感词的向量转换;同时利用主成分分析法降低特征的维度灾难。在酒店评论情感语料中进行验证试验,实验结果表明,与传统方法相比,该方法在酒店评论的情感分类任务中具有良好的准确率和召回率。Hotel comments objectively and truly reflect the customer＇ s feelings about the hotel and stow-ice. An emotional analysis model based on hotel online comments has been constructed. Firstly, the model extracts the syntax rules of the evaluation object - emotional word according to the design, and gets the fine-grained classification feature information. Then, vector representation of the evaluation object - emotional word is gotten by Word2vec and TF-IDF algorithm. At the same time, the dimensionality of the feature in the text classification is reduced by using PCA dimensionality reduction. The results show that compared with the traditional method, this method has good accuracy and recall rate in the emotional classification task of hotel comments.

关 键 词：情感分类 词向量 依存句法分析 支撑向量机 

分 类 号：F719.2[经济管理—产业经济]