基于Hadoop的微博用户情感分类研究与实现
发布时间：2018-10-20 08:00
【摘要】：随着微博等新型社交网络服务的发展与普及，人们借助此类媒介表达观点和情感变得更加灵活、自由、快速。因此，，针对微博的情感分类也显得越来越重要，通过微博情感分类，了解用户对政策、产品、舆论热点等的反应，更好的对用户自身、企业、政府等提供决策支持具有重要的意义。 在微博海量数据集上执行情感分类任务时，传统的情感分类算法的扩展性成为系统的瓶颈。因而，本文首先研究云计算平台-Hadoop的主要技术，分析了在Hadoop上实施情感分类的可行性。在此基础上，本文针对微博文本情感特点，通过自动构建和人工构建相结合的情感语料库，改进基于微博情感元素和语义的特征抽取算法，并采用Hadoop技术，设计了一种分布式、可扩展、自治的微博情感分类模型。针对该模型中的情感分类问题，设计并实现了基于Hadoop的朴素贝叶斯情感分类算法。测试结果表明，采用基于Hadoop的朴素贝叶斯情感分类模型对海量微博数据进行情感分类，具有良好的执行效率和较高的扩展性。
[Abstract]:With the development and popularity of new social networking services such as Weibo, people have become more flexible, free and quick to express their views and feelings through such media. Therefore, according to Weibo's emotional classification, it is becoming more and more important to understand the reaction of users to policies, products, public opinion hot spots, and so on, and better to the users themselves and the enterprises, through Weibo emotional classification. It is of great significance for the government to provide decision support. When carrying out emotion classification task on Weibo's massive data set, the expansibility of traditional emotion classification algorithm becomes the bottleneck of the system. Therefore, this paper first studies the main technology of cloud computing platform-Hadoop, and analyzes the feasibility of implementing emotion classification on Hadoop. On this basis, according to the emotional characteristics of Weibo text, this paper improves the feature extraction algorithm based on Weibo emotion elements and semantics through the combination of automatic and artificial construction of emotional corpus, and designs a distributed feature extraction algorithm based on Hadoop technology. Extensible, autonomous Weibo emotional classification model. Aiming at the emotion classification problem in this model, a naive Bayesian emotion classification algorithm based on Hadoop is designed and implemented. The test results show that using naive Bayesian emotion classification model based on Hadoop to classify massive Weibo data has good performance efficiency and high scalability.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.092;TP391.1
本文编号：2282534
[Abstract]:With the development and popularity of new social networking services such as Weibo, people have become more flexible, free and quick to express their views and feelings through such media. Therefore, according to Weibo's emotional classification, it is becoming more and more important to understand the reaction of users to policies, products, public opinion hot spots, and so on, and better to the users themselves and the enterprises, through Weibo emotional classification. It is of great significance for the government to provide decision support. When carrying out emotion classification task on Weibo's massive data set, the expansibility of traditional emotion classification algorithm becomes the bottleneck of the system. Therefore, this paper first studies the main technology of cloud computing platform-Hadoop, and analyzes the feasibility of implementing emotion classification on Hadoop. On this basis, according to the emotional characteristics of Weibo text, this paper improves the feature extraction algorithm based on Weibo emotion elements and semantics through the combination of automatic and artificial construction of emotional corpus, and designs a distributed feature extraction algorithm based on Hadoop technology. Extensible, autonomous Weibo emotional classification model. Aiming at the emotion classification problem in this model, a naive Bayesian emotion classification algorithm based on Hadoop is designed and implemented. The test results show that using naive Bayesian emotion classification model based on Hadoop to classify massive Weibo data has good performance efficiency and high scalability.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.092;TP391.1
【参考文献】
相关期刊论文 前9条
1 胡光民;周亮;柯立新;;基于Hadoop的网络日志分析系统研究[J];电脑知识与技术;2010年22期
2 吴维;肖诗斌;;基于多特征与复合分类法的中文微博情感分析[J];北京信息科技大学学报(自然科学版);2013年04期
3 刘志明;刘鲁;;基于机器学习的中文微博情感分类实证研究[J];计算机工程与应用;2012年01期
4 张玉芳;彭时名;吕佳;;基于文本分类TFIDF方法的改进与应用[J];计算机工程;2006年19期
5 谢丽星;周明;孙茂松;;基于层次结构的多策略中文微博情感分析和特征抽取[J];中文信息学报;2012年01期
6 庞磊;李寿山;周国栋;;基于情绪知识的中文微博情感分类方法[J];计算机工程;2012年13期
7 周胜臣;瞿文婷;石英子;施询之;孙韵辰;;中文微博情感分析研究综述[J];计算机应用与软件;2013年03期
8 张浩;尚进;;微博时代的电子政务建设与创新[J];中国信息界;2011年09期
9 陈彦舟;曹金璇;;基于Hadoop的微博舆情监控系统[J];计算机系统应用;2013年04期
本文编号：2282534
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2282534.html
