基于属性特征的评论文本情感极性量化分析

发布时间：2018-06-22 15:20

本文选题：评论文本 + 属性因子　；参考：《数据分析与知识发现》2017年10期

【摘要】：【目的】从评论对象的属性特征出发解决情感极性量化问题。【方法】将在线评论文本分解构建三层评论体系,即评论对象对象属性评论描述,从属性层级抽取属性词集和对应的评论集,考虑评论对象属性特征的不同影响,引入属性因子,并对TFIDF进行改进用以计算属性因子;结合评论模式和评论语境提出基于属性特征的评论情感量化分析算法并采用Python语言予以实现。【结果】相较于传统机器学习分类算法(NB、SVM)、属性因子设置为等权重时,本文算法在评论文本情感分类准确性方面有显著提高。【局限】评论集领域选择方面具有局限性,量化算法在系数设定方面存在主观性。【结论】本文算法能有效解决情感极性量化问题,进一步提高了情感分类准确性。
[Abstract]:[objective] to solve the problem of quantification of emotional polarity based on the attribute characteristics of comment objects. [methods] the online comment text is decomposed into three layers of comment system. The attribute word set and the corresponding comment set are extracted from the attribute level, considering the different effects of the attribute characteristics of the comment object, the attribute factor is introduced, and the TFIDF is improved to calculate the attribute factor. Combining the comment pattern with the comment context, a quantitative analysis algorithm of comment emotion based on attribute feature is proposed and implemented in Python language. [results] compared with the traditional machine learning classification algorithm (NBN SVM), when the attribute factor is set to equal weight, The algorithm in this paper has significantly improved the accuracy of emotional classification of comment text. [limitations] comment set domain selection has limitations, Quantization algorithm has subjectivity in coefficient setting. [conclusion] this algorithm can effectively solve the problem of emotional polarity quantization and further improve the accuracy of emotion classification.
【作者单位】：西安电子科技大学经济与管理学院;
【基金】：国家自然科学基金项目“基于可信语义Wiki的知识库构建方法与研究应用”(项目编号:71203173) 中央高校基本科研业务费专项资金资助项目“大数据环境下基于主题模型的信息服务研究”(项目编号:JB160606) 国家自然科学青年基金项目“大规模动态社交网络社团检测算法研究”(项目编号:71401130)的研究成果之一
【分类号】：TP391.1

【相似文献】