维基百科的人类行为动力学探讨
发布时间：2019-03-25 21:46
【摘要】：一直以来,学者们认为人类在社会及经济活动中的行为发生时间间隔,可以简单地用泊松分布来描述。但自2005年以来,Albert Barabási等人进行大量的实证研究后发现:人类行为发生时间并不全部符合均匀的泊松分布,而大量是服从有胖尾的幂律分布,并提出了人类行为的两个普适类的幂指数标度。解释这方面的行为理论主要有优先排队论、记忆、习惯、自适应性模型等,但这些模型都只是人类行为的某一方面特性。 本文通过对维基百科进行了较全面的实证分析,从而得出维基人类行为的基本特点,然后提出一个综合的模型,进而全面的检验实证的结果,并从理论上分析这种行为的深层原因。本文得到了以下的主要研究结论: 维基百科中所有用户的登录次数、部分用户的登录次数和登录时间间隔都有近似的幂律分布拟合。实证所得到的幂指数主要可以划分为三个(由小大到分别是:1、1.5、2),其中两个幂指数(1和1.5)和Vázquez等提出的两个普适类是相一致的。但仅仅把人类行为特征简单地分成两个普适类幂指数是不够完整的,至少还应该存在着第三个幂指数(指数大小为2)。 人类行为综合模型仿真出来的结果可基本划分为四个幂指数,由小到大分别是0.6、1、1.5、2。后三个幂指数,与维基百科的实证所得结果是相一致的,但模型还得出了第四个幂指数——0.6。本文通过对维基百科的十四个重要贡献者的行为分析,揭示了人类什么样的行为,会使其最终行为分布的幂指数是多少的对应关系。 本文的研究意义在于:由于综合模型仿真出来的结果与实证相符合,这说明本文所归纳综合的人类行为模型是比较正确的,可以在公共设施资源的建设与分配中有比较大的实用价值。
[Abstract]:For a long time, scholars think that the time interval of human behavior in social and economic activities can be simply described by Poisson distribution. However, since 2005, Albert Barb 谩 si et al have conducted a large number of empirical studies and found that the occurrence time of human behavior does not all conform to the uniform Poisson distribution, but a large number of them follow the power-law distribution with fat tail. The power exponents of two universal classes of human behavior are also proposed. There are priority queuing theory, memory theory, habit model, self-adaptive model and so on to explain this aspect of behavior theory, but these models are only one aspect of human behavior characteristics. Through a comprehensive empirical analysis of Wikipedia, this paper obtains the basic characteristics of Wikipedia's human behavior, and then puts forward a comprehensive model, and then tests the results of empirical research in an all-round way. And from the theoretical analysis of the underlying causes of this behavior. In this paper, the following main conclusions are obtained: the number of login times of all users in Wikipedia, the number of login times of some users and the interval of login time have approximate power law distribution fitting. The power indices obtained by empirical analysis can be divided into three (from small to large: 1, 1.5, 2), in which two power indices (1 and 1.5) and two universal classes proposed by V 谩 zquez are consistent. But simply dividing the human behavior characteristics into two universal power indices is not complete enough, at least there should be a third power index (the size of the index is 2). The simulation results of the synthetic model of human behavior can be basically divided into four power exponents, from small to large, which are 0.6, 1, 1.5, 2. The last three power indices are consistent with the empirical results of Wikipedia, but the fourth power index-0.6 is obtained by the model. By analyzing the behavior of fourteen important contributors of Wikipedia, this paper reveals the relationship between what kind of human behavior and the power index of final behavior distribution. The research significance of this paper is as follows: because the simulation results of the synthetic model are consistent with the empirical results, this shows that the synthetic human behavior model in this paper is relatively correct. It has great practical value in the construction and distribution of public facilities resources.
【学位授予单位】：华南理工大学
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：O211.3;C912
本文编号：2447349
[Abstract]:For a long time, scholars think that the time interval of human behavior in social and economic activities can be simply described by Poisson distribution. However, since 2005, Albert Barb 谩 si et al have conducted a large number of empirical studies and found that the occurrence time of human behavior does not all conform to the uniform Poisson distribution, but a large number of them follow the power-law distribution with fat tail. The power exponents of two universal classes of human behavior are also proposed. There are priority queuing theory, memory theory, habit model, self-adaptive model and so on to explain this aspect of behavior theory, but these models are only one aspect of human behavior characteristics. Through a comprehensive empirical analysis of Wikipedia, this paper obtains the basic characteristics of Wikipedia's human behavior, and then puts forward a comprehensive model, and then tests the results of empirical research in an all-round way. And from the theoretical analysis of the underlying causes of this behavior. In this paper, the following main conclusions are obtained: the number of login times of all users in Wikipedia, the number of login times of some users and the interval of login time have approximate power law distribution fitting. The power indices obtained by empirical analysis can be divided into three (from small to large: 1, 1.5, 2), in which two power indices (1 and 1.5) and two universal classes proposed by V 谩 zquez are consistent. But simply dividing the human behavior characteristics into two universal power indices is not complete enough, at least there should be a third power index (the size of the index is 2). The simulation results of the synthetic model of human behavior can be basically divided into four power exponents, from small to large, which are 0.6, 1, 1.5, 2. The last three power indices are consistent with the empirical results of Wikipedia, but the fourth power index-0.6 is obtained by the model. By analyzing the behavior of fourteen important contributors of Wikipedia, this paper reveals the relationship between what kind of human behavior and the power index of final behavior distribution. The research significance of this paper is as follows: because the simulation results of the synthetic model are consistent with the empirical results, this shows that the synthetic human behavior model in this paper is relatively correct. It has great practical value in the construction and distribution of public facilities resources.
【学位授予单位】：华南理工大学
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：O211.3;C912
【参考文献】
相关期刊论文 前10条
1 赵金山,狄增如,王大辉;北京市公共汽车交通网络几何性质的实证研究[J];复杂系统与复杂性科学;2005年02期
2 俞桂杰;彭语冰;褚衍昌;;复杂网络理论及其在航空网络中的应用[J];复杂系统与复杂性科学;2006年01期
3 韩筱璞;周涛;汪秉宏;;基于自适应调节的人类动力学模型[J];复杂系统与复杂性科学;2007年04期
4 杨春学;利他主义经济学的追求[J];经济研究;2001年04期
5 曹毅;张明智;傅凝;罗批;;基于SWARM的虚拟社会经济系统建模与仿真[J];计算机仿真;2008年05期
6 余柏峰;张辉;;基于swarm仿真平台的合同网协议的应用分析[J];计算机仿真;2008年09期
7 陆履平;杨建梅;;广州软件产业网络特征分析[J];科技管理研究;2006年06期
8 焦玉;刘衍珩;王健;王静;;基于习惯的人类动力学建模[J];科学通报;2010年11期
9 廖飞;;关于人类行为原因的整合[J];学理论;2009年32期
10 刘宏鲲;周涛;;中国城市航空网络的实证研究与分析[J];物理学报;2007年01期
相关硕士学位论文 前1条
1 张雯雯;复杂网络理论在航空网络中的应用研究[D];中国民航大学;2009年
，本文编号：2447349
本文链接：https://www.wllwen.com/shekelunwen/shgj/2447349.html
