文献类型：期刊文献

中文题名：基于主题增强的递归自编码情感分类研究

英文题名：Study on Recursive Auto-encoding Sentiment Classification Based on Topic Enhancement

作者：朱引[1];黄海燕[1]

机构：[1]华东理工大学信息科学与工程学院,上海200237

年份：2018

卷号：45

期号：12

起止页码：142

中文期刊名：计算机科学

外文期刊名：Computer Science

收录：CSTPCD;;北大核心:【北大核心2017】；CSCD:【CSCD_E2017_2018】；

语种：中文

中文关键词：递归自编码;主题模型;情感分类;数据挖掘

外文关键词：Recursive auto-encoder;Topic model;Sentiment classification;Data mining

摘要：中文文本情感分析旨在发现用户对事物、事件的情感倾向,然而现有研究往往忽视了文本之间的相互联系。提出一种基于主题增强的递归自编码情感分类模型,通过将文本的主题信息融入到递归自编码模型中,使得该模型可以更深层次地考虑文本的内容信息,提高其对文本情感的理解和泛化能力。在COAE2014数据集上的实验结果表明,将所提分类模型用于情感分类任务时可获得更优的分类效果,证实了其在实际问题中的适用性与可行性。
The emotional analysis of Chinese text aims to discover the emotional tendencies of users to things and events,however,the existing studies often neglect the interrelationships between texts.In light of this,this paper proposed a recursive auto-encoding classification model based on topic enhancement.By incorporating the subject information of the text into the recursive auto-encoding model,this model can further consider the content information of the text and improve the capability to understand the text emotion and generaliza ability.The experimental results on the COAE2014 dataset show that the proposed classification model can achieve better classification performance when used for tasks of sentiment classification,thus verifying its applicability and feasibility in practical problems.