虛擬社會網(wǎng)絡(luò)中用戶行為特性與模式研究.pdf

1、南京航空航天大學(xué)碩士學(xué)位論文虛擬社會網(wǎng)絡(luò)中用戶行為特性與模式研究姓名：朱永真申請學(xué)位級別：碩士專業(yè)：計算機科學(xué)與技術(shù)指導(dǎo)教師：夏正友2010-12虛擬社會網(wǎng)絡(luò)中用戶行為特性與模式研究 II ABSTRACT Network virtual community is a mapping and extension of real social network on the Internet, with unique properties

2、and characteristics. Because of the openness, virtualization, interaction of networked virtual community, mining network virtual community and its member behavior has become hot, while the study on the potential of commu

3、nity organizations is also subject to more and more researchers. With the forum of “TianYa on international” as research subjects, base on the network data in five years, the structural characteristics of community, the

4、Member website and the relationship between the potential of community organizations is analyzed, the main works of the thesis are summarized as follows: Firstly, Network crawler is developed to automatically collect the

5、 data set on “international observation “. The data set covers the information on the topic during the five years from 2005 to 2009. According to the needs of specific research, they are divided into three categories: 1)

6、 321027 user information, including user ID, nick name; 2) 96073 article topic information, containing topic ID, subject name, subject, author, subject, time and so on; 3) 5241338 replies relationship information that co

7、ntains the reply is ID, respondents nickname, the topic that is replied, the time, the content, etc. Which provides rich data sources for later analysis directly, therefore, the reliability of the real is guaranteed. Sec

8、ondly, According to the various complex situations presented when the users taking part in the subjects, an algorithm of user interest based on the common response relationship was proposed. Based on an existing Data Set

9、, the regulation that different type of theme has a different number of respondents is discovered. In General, the users only participate in the discussion of some topics, suppose a type of the topics corresponds to one

10、of interest, users participate in the discussion of different types of topics can be viewed as having different hobbies. Thus, it is of great significance to discover user's interest according to the relationship bet

11、ween the same reply. Finally, Societies partitioning algorithm is put forward base on time and space according to user interest. The user will gather into a “group“ based on the same or similar hobbies, therefore, it is