柬埔寨语新闻评论文本情感分类研究

发布时间：2018-04-10 08:54

本文选题：柬埔寨新闻　切入点：评论文本　出处：《昆明理工大学》2017年硕士论文

【摘要】：本文主要进行了柬埔寨语新闻评论文本的情感分类研究,本文的最终目标是研究柬埔寨新闻评论在文档层面的情感倾向。首先,我们研究句子层面的情感分类,并产生三级特定的句子情感分类模型。在句子层面情感分类研究的基础上,通过对句子感情的分类,我们从各个方面深入研究文档层面的情感分类算法。通过研究,运用几个主要算法的改进的基础,从研究中得出了三个具体的与研究内容相关的文档级情感分类模型。然后,针对不同的分类模型给出每个模型的具体实现,在研究的基础上并对每个模型的实验结果进行多方的深入分析。首先介绍了柬埔寨语新闻评论文本情感分类国内外研究背景、意义和现状研究现状及存在的问题。其次基于句子水平的情感分类研究,总结三种特定的情感分类模型,为词库的建立和实验平台的构建提供一些思路。在文档级的基础上多方研究推理,分别介绍和研究了文档级别的分类过程和算法,在对文档级的基本算法的优化改进基础上,给出了三种特定情感分类模型。对柬埔寨新闻语料库的建设,实验平台的建设,和实验结果的分析,以及几个模型的性能进行了比较。描述了情感分类系统的设计,情感分类系统分析的结果,介绍了性能指标,给出了实验模型的情感分类实验结果,对每个模型的性能给出了比较,讨论影响每个模型的准确性的因素。最后总结了本文的工作,并期待与本文主题相关的后续研究。
[Abstract]:This paper mainly studies the emotional classification of Cambodian news reviews. The ultimate goal of this paper is to study the emotional tendency of Cambodian news reviews at the document level.First of all, we study the affective classification at sentence level, and produce a three-level specific sentence affective classification model.On the basis of the research of sentence level emotion classification, through the sentence emotion classification, we deeply study the document level emotion classification algorithm from all aspects.Based on the improvement of several main algorithms, three specific document-level affective classification models are obtained from the research.Then, the realization of each model is given for different classification models, and the experimental results of each model are analyzed in detail on the basis of the research.Firstly, this paper introduces the research background, significance, status quo and existing problems of the emotional classification of Cambodian news commentary texts at home and abroad.Secondly, based on the research of emotion classification at sentence level, three kinds of specific emotion classification models are summarized, which provide some ideas for the establishment of lexicon and the construction of experimental platform.On the basis of document level, this paper introduces and studies the classification process and algorithm of document level. On the basis of optimizing and improving the basic algorithm of document level, three kinds of specific emotion classification models are given.The construction of Cambodian news corpus, the construction of experimental platform, the analysis of experimental results, and the performance of several models are compared.This paper describes the design of the emotion classification system, the results of the analysis of the emotion classification system, introduces the performance indexes, gives the experimental results of the emotional classification of the experimental model, and compares the performance of each model.The factors that affect the accuracy of each model are discussed.Finally, this paper summarizes the work of this paper, and looks forward to the follow-up study related to the theme of this paper.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.1

【参考文献】