Simple and fantastic network, complex and predictable behavior----Read < link > Thoughts

Do you know why you're here? ----Preface

More than a month ago, has read < burst, and re-read < link, the heart can not help yearning to deep, feeling that scientists Barabasi to the real world deep analysis, but also very grateful to the translators of the two books (Shen Weihua teacher and Zhou Tao teacher) Hard work. But panic dare not to write, today decided to stop laughing ears. Although the body temperature of the book has dissipated, the glowing thought endures. < links > A book was written around 00, the problem is the network formation mechanism and network-oriented structure of the relevant research hotspots. The version read is the 10 Anniversary edition, so I do not know the original, only the version. Although the book is divided into 3 parts, but I read, divided into 4 parts is better, the essence of the second part, I was as a two parts to read.

Before the first part, the author tells the importance of the network, how the network makes the world so wonderful, and why the network is so fragile, Yahoo website in the attack of the little boy is so fragile, let people sigh unceasingly, however, the existing reductionism-based research methods may no longer be suitable for explaining the complexity of the network structure. This chapter is the leader of this book, the key issue of this book, the basic rules of the evolution of complex network structure is what.

The first part introduces the complex network research context. Before the 70 's, the study of complex networks was focused on random networks, based on the research conducted by many teachers such as Euler and Hamilton, but stochastic networks could not effectively portray online network structure, although random networks had graceful mathematical expression, but Small World Network, the discovery of six degree theory let random network become the past, In the existing research, only the specific research background will use the stochastic network structure, such as the study of complex network characteristics under different network structure, still some people use random network, rules network research, from the physics, it is still very meaningful, because in physics, a large number of networks is still random network and rule network , and I engaged in the Internet in the background of the network research, it is not too concerned about the characteristics of the rules network, especially for the formation of community under the complex network, from the physics to become "seepage" phenomenon. The six-degree separation theory points out an important feature of the social network structure, which is a significant advance in the research of stochastic networks, with a small maximum path and a logarithmic relationship between the maximum path and the number of nodes. When granovetter a weak connection to the deep observation of employment Opportunities (Granovetter, 1973), it formally introduces the important issues of modern research on network structure, the convergence of social network relations and the phenomenon of clustering, and the academic Erdös number of the proposed also reflects the social network of high-poly phenomenon, the Small World Network from this (Watts and Strogatz, 1998), the random world view of the network research in this dead.

The second part is the core of the book, I am divided into two parts to read, first of all, the complex network structure of the interpretation of the mechanism, and then the complex network of important research issues to comb. The discovery of a Small World network opens a window into the study of social networks, but the research of the Small World Network is a disconnected reconnection to the regular network, but in the real world, the mechanism of complex networks remains unclear until Barabasi's masterpiece was born (Barabási and Albert, 1999; Palla and Barabási et al.,2007; Barabási, 2013). Through the analysis of the structure of the Internet Web page, found that a few pages occupy a large number of connections, and the vast majority of pages only a few connections, network structure distribution density and degrees into a power-rate relationship, where information is passed in, social network structure presents no scale characteristics. The scale-free feature effectively depicts the existing network results, although there is a certain gap between the results of numerical simulation and the actual results, such as the existence of "power saturation" phenomenon in the network structure, and the power rate value is not necessarily 3, but these can be explained by the complex behavior of the node. So here are two questions to ask: First, what are the core elements of a complex network? And then what is the complex network formation mechanism?

For the first problem, focus on the hub node, the hub node plays the role of Information Bridge, so from the network structure, the hub node is the backbone of the network structure, which also leads to the important characteristics of complex networks, the robustness of complex networks and the vulnerability of the face of deliberate attacks, The attack on Yahoo, mentioned in the first part, is proof. At the same time, it also raises two important research questions, one is how to identify the hub node, the research is very common in management science, specifically for the social network structure of opinion Leader identification, specific research, combined with specific research background, more from the behavior, knowledge and other aspects of the node description, extract node attributes, improve accuracy rate The other is to maximize the research of WOM in social networks or to understand WOM communication, and its common research methods are simulated or empirical, in the study of maximizing WOM, the general strategy is to select information Initial node, based on linear threshold model (Watts, 2002; Watts and Dodds, 2007) or independent cascade model to simulate the interaction behavior of nodes, so as to test the initial node selection efficiency, and in the empirical study, based on the bass model (bass, 1969), the use of time series empirical method of the propagation process to study the dissemination of the law in a specific context. The core of the above research is based on the complex network of the hub node.

For the second problem, Barabasi the three evolutionary mechanisms of the complex network: growth mechanism, preference mechanism and adaptation mechanism, the first two describe the generation of scale-free network, emphasize the first advantage and Matthew effect in the process of network evolution, and the inductive adaptation mechanism explains it. The growth mechanism from the perspective of dynamic evolution, the network is a successive addition of nodes resulting in a large network size, and the preference mechanism emphasizes the node's first mechanism, that the node is biased to connect with more nodes, and based on the network retrieval method, such as PageRank (Brin and Page, 1998; Page and Brin et al, 1999), HITS (kleinberg,1999) are on this assumption to identify the pivot node; Although the first two nodes explain the key to the evolution of the network structure, it is impossible to explain why Google can defeat Yahoo and other behind the phenomenon , and the adaptability model and adaptation mechanism are introduced to improve the attraction of the competitive node to the link. These three mechanisms are proposed from the interpretation of the existing network generation process, and the interpretation process more derived from the physical level.

Then, the research on the complex network is more concerned about the characteristics of the network structure, such as the network phase change behavior, network fragmentation, network community recognition, which based on the management science background, I think it is particularly interesting is three, one is the network of anti-attack, and the other is the impact of the problem; Although Watts has proposed a variety of community measurement and identification methods, but still need to further develop and apply, of course, and so on. This article does not do in-depth development. However, this paper still needs to put forward an interesting question: what is the dimension of scale-free network, and then based on the exposition of scale-free network, which should have fractal features, how to extract the network structure from similar networks in the scale-free network, in the concrete use, the results can be automatically extended to the whole network structure?

The third part is the expansion of the existing network, from the biology, epidemiology, Internet and other fields to analyze the rationality of complex network applications, to understand the "network effect", such as why the Internet market competition, the eldest brother is far greater than the second, old three, why AIDS, such as the spread of such characteristics and so on, this article is no longer described.

For the link, write it and thank the author of the book and the Translator (Shen Weihua) for their work. After a while, then tidy up < burst >.

Look at the objective light pat.

References:

Barabási, A. (2013). "Network Science." Philosophicaltransactions of the Royal Society a:mathematical,physical and Engineeringsciences371 (1987).

Barabási, A. and R. Albert (1999). "Emergence ofscaling in the random networks." Science 286 (5439): 509-512.

Bass, F. M. (1969). "A New Product growth for modelconsumer durables." Management Science (5): 215-227.

Brin, S. and L. Page (1998). "The Anatomy of Alarge-scale hypertextual Web search engine." Computer Networks and Isdnsystems(1–7): 107-117.

Granovetter, M. S. (1973). "The strength of weakties." American Journal of Sociology (6): 1360-1380.

Kleinberg, J. M. (1999). "Authoritative sources in ahyperlinked environment." J. ACM ( 5): 604-632.

Page, L. and S. Brin, et al. (1999). The pagerankcitation ranking:bringing Order to the Web., Stanford Infolab.

Palla, G. and A. Barabási, et al. (2007). " Quantifying social group evolution. " Nature 446 (7136): 664-667.

Watts, D. J. (2002). "A simple model of globalcascades on random networks." Proceedings of the National Academy ofsciences (9): 5766-5771.

Watts, D. J and P. S. Dodds (2007). "Influentials,networks, and public opinion formation." Journal of Consumer ( 4): 441-458.

Watts, D. J. and S. H. Strogatz (1998). "Collectivedynamics of ' small-world ' networks." Nature 393 (6684): 440-442.

Simple and fantastic network, complex and predictable behavior----Read < link > Thoughts