基于文本挖掘的社交网络用户画像建模与应用

发布时间：2018-05-12 07:17

本文选题：用户画像 + 主题模型　；参考：《北京邮电大学》2016年硕士论文

【摘要】：互联网的发展使得人们日常生活产生了翻天地覆的变化,随着智能手机的普及,网络环境的不断完善,如今已经全面进入移动互联网时代。无论是通信运营商,还是网络应用提供商,都认识到了用户的重要性。如何更好的了解用户,提升用户体验成为了互联网服务的最重要的课题。海量的网络用户数据为我们更好的了解用户提供了丰富的素材,如何利用这些数据在如今成为了研究者的热点,网络用户画像就是其中之一。现有的社交网络用户画像研究关注的特征维度很多集中于用户基本信息作为特征,而这些基本信息一般需要通过特定的API来获取,越来越多的社交媒体平台开始保护用户的隐私使得这些信息越来越难于获取。而由于社交网络平台具有丰富的用户生成内容,这些内容容易获取,本文主要研究了用户的文本内容。本文将用户的人格维度引入到了用户画像中来,研究了与分析了主题特征与用户人格之间的相关性,并建立了基于主题模型为特征的社交网络用户人格预测模型。目前的研究并没有将人格因素与主题相结合的概率模型,本文在主题与用户人格具有相关性结论的基础上,提出了用户人格主题的概率生成模型及其改进模型,并在真实数据集上进行实验。用户人格主题模型能够同时生成用户的主题-词分布以及人格-词分布,从而可以直接观测出人格与词的对应关系。
[Abstract]:With the development of the Internet, people's daily life has changed. With the popularity of smart phones and the continuous improvement of the network environment, it has entered the era of mobile Internet. Both communication operators and network application providers recognize the importance of users. How to better understand the user and improve the user experience has become the most important issue of Internet services. The massive data of network users provide us with abundant materials to understand users better. How to make use of these data has become a hot spot of researchers nowadays, among which the portrait of network users is one of them. Many of the feature dimensions concerned in the current social network user portrait research focus on the user's basic information, which usually needs to be obtained through a specific API. More and more social media platforms are protecting users' privacy, making this information increasingly difficult to access. Because the social network platform has rich user-generated content, which is easy to obtain, this paper mainly studies the text content of users. In this paper, the personality dimension of the user is introduced into the user portrait, the correlation between the subject feature and the user personality is studied and analyzed, and a social network user personality prediction model based on the theme model is established. At present, there is no probabilistic model combining personality factors with themes. Based on the conclusion of correlation between subject and user personality, this paper puts forward the probabilistic generating model of user personality theme and its improved model. Experiments were carried out on real data sets. The user personality thematic model can generate the user's topic-word distribution and the personal-word distribution at the same time, so that the corresponding relationship between personality and words can be observed directly.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.1

【相似文献】