Topic Analysis Model for Online Social Network
Yan Jia
College of Computer
National University of Defense
Technology
Changsha, China
jiayanjy@vip.sina.com
Lu Deng*
College of Computer
National University of Defense
Technology
Changsha, China
denglu@nudt.edu.cn
Bin Zhou
College of Computer
National University of Defense
Technology
Changsha, China
binzhou@nudt.edu.cn
Binxing Fang
Institute of Information
Engineering
Chinese Academy of Sciences
Beijing, China
fangbx@bupt.edu.cn
Weihong Han
Cyberspace Institute of Advanced
Technology
Guangzhou University
Guangzhou, China
hanweihongnudt@139.com
Aiping Li
College of Computer
National University of Defense
Technology
Changsha, China
13017395458@163.com
Qiang Liu
College of Computer
National University of Defense
Technology
Changsha, China
liuqiang1981@nudt.edu.cn
Yong Quan
College of Computer
National University of Defense
Technology
Changsha, China
quanyong8801@126.com
Abstract— With the rapid development of social network
platform and the widespread use of mobile terminals, the topic
analysis technology based on data of social network gets more and
more attention. Due to the characteristics as short length of data and
the seriousness of content fragmentation, the topic analysis
technology on social network has evolved into three main parts:
information recovery, topic detection and topic evolution. Among
them, information recovery is the basis of topic detection and topic
evolution relies on topic detection. Based on the characteristics
summary of related works of the three parts, a generalized model of
topic analysis on social networks is proposed with the consideration
of factors such as community, influence and user. It expounds the
inner process of topic analysis technology and is an effective
summary of topic analysis technology on social networks.
Keywords—Topic analysis, Information recovering, Topic
detection, Topic evolution
I. INTRODUCTION
With the rapid development of the Internet and information
technology, online social networks have become an important
way for people to obtain information, express views and
opinions. Online social networks map the associations between
individuals and individuals, individuals and groups, groups and
groups in the real world to the Internet, which forms a virtual
network society. Events in the real world will be published on
social networking platforms as information, thus forming a topic
discussed by public. As time goes on and the interaction between
users increases, the public opinion formed by topic evolution
will affect events in the real world. How to quickly and
accurately discover new content, new events and new topics that
people are interested in and track the trend of these information
has got more and more attention.
In the early stage, topic analysis [2] mainly regards the news
information published by the official media as the research
object. By monitoring the topics described by the news, topic
analysis discovers new information that the user is interested in,
analyzes its dissemination and presents collecting news
information of a topic to users in some way. The research texts
are long and rich, and contains detailed information that
describes the whole process of event and topic. The correlation
between documents is strong and the content is issued by the
official media which means it has high credibility and its source
is single. Therefore, the situations as content overlaps and
inconsistent content will not happen. Due to the rapid
development of the Internet and the widespread use of mobile
terminals, various social media platforms get more and more
attention. As a result, researchers put their focus into new media
such as mail, forums, communities, and blogs. The research
priority is text with long content. Different from news reports,
the content in social media is more arbitrary and along with
noise, and the logical connection between documents is weak.
What’s more, any user can publish information instead of only
official media and a topic is created by any user, which means
participation sense of users is stronger. All these characteristics
have raised new requirements for the topic detection and
evolution technology. Following the popularity of social media
with short text such as Twitter, Facebook, Sina Weibo, Tencent
Weibo and so on, the focus of researchers has shifted to new
types of social media. Users can publish same or similar
information on multiple social platforms. Therefore, the analysis
of content requires considerations on multiple platforms and
multi-source instead of limited to a single social platform. In
addition, In addition, there are other challenges in the study of
topic detection and evolution for short texts, such as short text,
severe fragmentation, huge data volume, arbitrary and
unregulated content.
The spread of events and topics on the Internet has shaped
and influenced online public opinion. The evolution of online
public opinion will affect the development of events in the real
world in turn. Therefore, researches on topic analysis
technology have strong practical significance and social value.