基于改进LDA算法的微博用户兴趣偏好分析系统的设计与实现

发布时间：2019-07-06 21:41

【摘要】：随着互联网技术以及社交网络的发展,微博作为一种知名的社交网络,以其优秀的用户体验吸引了海量用户。大量用户在微博这个平台抒发每日的情感,进行社交活动,结交新的好友。随着微博上数据量与信息量的日益庞大,其自然而然地形成了规模可观的语料库。本文旨在设计并实现一种系统,用于根据微博用户的文本信息分析用户的潜在兴趣,并使用情感分类算法判断微博用户对于各兴趣主题的情感倾向(积极或消极)。通过实现本文中的系统,可以为个性化推荐服务提供一种更加可靠可行的实现方案。商家可以根据分析得到的用户兴趣对每个用户进行较为精准的消息推送,实现更好的宣传效果。本文的研究内容主要在于通过数据抓取、兴趣分析、情感分类三个模块的实现,构成一个完整的系统,进而实现针对微博用户的个性化兴趣偏好分析。系统通过实现分布式爬虫模块,完成大规模兴趣语料的抓取;通过实现基于Labeled-LDA模型的主题分析模块,完成兴趣主题提取与用户的兴趣预测;最后,通过实现基于朴素贝叶斯分类器的情感分类模块,完成用户兴趣的情感判定。
文内图片：

图片说明：图３－２分布式爬虫架构图逡逑分布式架构由一台Ｂｒｏｋｅｒ以及多个Ｗｏｒｋｅｒ组成，每个Ｗｏｒｋｅｒ拥有独立的ＩＰ地址
[Abstract]:With the development of Internet technology and social network, Weibo, as a well-known social network, attracts a large number of users with its excellent user experience. A large number of users express their daily feelings, socialize and make new friends on Weibo. With the increasing amount of data and information on Weibo, it naturally forms a considerable corpus. The purpose of this paper is to design and implement a system to analyze the potential interests of Weibo users according to the text information of Weibo users, and to use emotional classification algorithm to judge the emotional tendencies (positive or negative) of Weibo users to each topic of interest. By implementing the system in this paper, we can provide a more reliable and feasible implementation scheme for personalized recommendation service. Merchants can push more accurate messages to each user according to the user interest obtained from the analysis, so as to achieve better publicity effect. The main research content of this paper is to form a complete system through the realization of three modules: data capture, interest analysis and emotion classification, and then realize the analysis of personalized interest preference for Weibo users. The system completes the capture of large-scale interest corpus by realizing the distributed crawler module; completes the interest topic extraction and user interest prediction by realizing the topic analysis module based on Labeled-LDA model; finally, completes the emotion judgment of user interest by realizing the emotion classification module based on naive Bays classifiers.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.1;TP393.092

【相似文献】