[毕业论文] 动态社会网络的用户行为建模与预测 [复制链接]

资料简介：
毕业论文-动态社会网络的用户行为建模与预测，共79页，31500字

中文摘要

随着互联网的飞速发展以及Web 2.0时代的到来，网络呈现出越来越真实的

社会性。大量的社会网络资源充斥于互联网的每个角落。而随着对于社会网络

研究的深入，研究动态社会网络中的用户行为显得更加重要。用户的动态行为

会被各种各样的因素影响，诸如个人兴趣，社会影响以及社会中整体的趋势。

然而，目前并没有相关的论文对用户的动态行为进行系统的研究，包括动态社

会网络中用户行为如何演化，以及不同的因素如何影响用户的行为。

本文提出了动态抗噪音因子图模型（Noise Tolerant Time-varying Factor

Graph Models （简称NTT-FGM）），来对用户的动态行为进行建模和预测。为了

更准确地建模，这个模型同时考虑了社会网络的结构，用户的属性以及用户行

为的历史记录。具体地说，某个用户在 t 时刻的行为是由她在 t 时刻的隐行为

状态决定的。而这个隐行为状态受到多种因素的影响，包括她自己的属性，她

自己在 t ? 1 时刻的隐行为状态以及她朋友在 t 和 t ? 1 时刻的隐行为状态。本文

通过对动态社会网络中的用户行为进行数据分析来验证这些想法，然后形式化

地定义了动态社会网络中用户行为建模的问题，并且提出了动态抗噪音因子图

模型（NTT-FGM）。针对模型的求解，本文结合线性连续系统和马尔科夫随机

场中的理论提出了高效的算法。同时为了满足社会网络的复杂性和处理大规模

数据的要求，本文提出并实现了基于MPI的算法并行化。

最后，本文通过在三个真实的数据集（Twitter，Flickr，Arnetminer）上的

实验验证了动态抗噪音因子图模型的有效性。实验结果表明，就F1-Measure而

言，提出的模型比基线算法（SVM，wvRN）高出10%-20%。同时，在5台4核

机器上运行的并行算法可以达到13 倍到17倍的加速比。

关键词：动态社会网络中用户行为建模力分析

ABSTRACT

Users’ behaviors (actions) in a social network are in?uenced by various factors

such as personal interests, social in?uence, and global trends. However, few publica-

tions systematically study how social actions evolve in a dynamic social network and

to what extent di?erent factors a?ect the user actions.

In this paper, we propose a Noise Tolerant Time-varying Factor Graph Model

(NTT-FGM) for modeling and predicting social actions. NTT-FGM simultaneously

models social network structure, user attributes and user action history for better pre-

diction of the users’ future actions. More speci?cally, a user’s action at time t is gen-

erated by her latent state at t, which is in?uenced by her attributes, her own latent state

at time t ? 1 and her neighbors’ states at time t and t ? 1. Based on this intuition, we

formalize the social action tracking problem using the NTT-FGM model; then present

an e?cient algorithm to learn the model, by combining the ideas from both continuous

linear system and Markov random ?eld.

Finally, we present a case study of our model on predicting future social actions.

We validate the model on three di?erent types of real-world data sets. Qualitatively, our

model can discover interesting patterns of the social dynamics. Quantitatively, exper-

imental results show that the proposed method outperforms several baseline methods

for social action prediction.

Key words: Social action tracking uence analysis

目 录

第 1 章 绪论 ............................................................................1

1.1 引言 ...............................................................................1

1.2 动态社会网络的用户行为问题定义 ...........................................2

1.3 论文的主要工作..................................................................3

1.4 论文的组织 .......................................................................5

第 2 章 社会网络的研究现状 .........................................................6

2.1 动态社会网络分析 ...............................................................6

2.1.1

2.1.2

2.1.3

动态社会网络模型...........................................................6

动态随机模型 ................................................................7

行为选择与影响力分析 .....................................................8

2.2 社会影响力分析..................................................................9

2.2.1

2.2.2

定性分析......................................................................9

定量分析.................................................................... 10

2.3 群组行为分析 .................................................................. 12

2.4 社会网络的研究现状小结 .................................................... 14

第 3 章 社会网络行为动态建模 .................................................... 15

3.1 数据分析 ........................................................................ 15

3.2 问题形式化定义................................................................ 17

3.3 动态抗噪音因子图模型 ....................................................... 19

3.4 模型的学习 ..................................................................... 22

3.4.1

3.4.2

学习算法.................................................................... 23

社会行为的预测 ........................................................... 26

3.5 算法并行实现 .................................................................. 26

3.6 社会网络行为动态建模小结.................................................. 28

III

第 4 章 实验结果分析 ............................................................... 29

4.1 实验数据 ........................................................................ 29

4.2 基线算法 ........................................................................ 30

4.3 结果分析 ........................................................................ 31

4.3.1

4.3.2

4.3.3

4.3.4

评价方法.................................................................... 31

预测效果.................................................................... 31

算法效率比较 .............................................................. 33

定性的样例研究 ........................................................... 34

4.4 实验结果分析小结 ............................................................. 36

第 5 章 结论和进一步的工作 ....................................................... 37

5.1 论文总结 ........................................................................ 37

5.2 进一步的工作 .................................................................. 37

插图索引 ................................................................................ 39

表格索引 ................................................................................ 40

公式索引 ................................................................................ 41

参考文献 ................................................................................ 43

致 谢 ................................................................................... 47

声 明 ................................................................................... 48

附录 A

外文资料的调研阅读报告或书面翻译 ................................... 49

A.1 Identify In?uence and Qualitative Analysis ................................... 49

A.1.1 In?uence and Correlation .................................................. 50

A.1.2 Three Degree of In?uence.................................................. 50

A.1.3 Strong tie and Weak tie ..................................................... 51

A.2 Quantitative Analysis and Application ........................................ 52

A.2.1 Quantifying In?uence ...................................................... 53

A.2.2 Selection and In?uence ..................................................... 54

A.2.3 In?uence and Action........................................................ 56

IV

A.2.4 In?uence and Interaction ................................................... 57

A.2.5 In?uence and Link Analysis ............................................... 57

A.3 In?uence Maximization and Viral Marketing ................................. 58

A.3.1 Di?usion in?uence Model.................................................. 59

A.3.1.1 General threshold model ............................................... 59

A.3.1.2 General cascade model................................................. 59

A.3.2 Maximizing the Spread of In?uence ....................................... 60

A.3.2.1 High-degree Heuristic.................................................. 60

A.3.2.2 Low-distance Heuristic ................................................ 60

A.3.2.3 Greedy Algorithm ...................................................... 60

A.3.2.4 Fast In?uence Maximization........................................... 61

A.3.3 Learning to Predict the Customers ......................................... 62

A.4 Conclusion ...................................................................... 63

参考文献 ................................................................................ 64

在学期间参加课题的研究成果 ....................................................... 67

资料文件预览：
共1文件夹，1个文件，文件总大小：2.90MB，压缩后大小：503.92KB

• 毕业论文-动态社会网络的用户行为建模与预测
• 毕业论文-动态社会网络的用户行为建模与预测.doc  [2.90MB]