A Complex Network Model Based On Domain-specific Knowledge

In recent years, as the rapid development of science and technology represented by information technology make human society entered the Internet age. A large number of complex networks such as WWW,Twitter arise at this historic moment. The impact of these complex network systems all the time in our life, it has become more and more urgent for our humans to study of complex network structure characteristics and the rule of these evolutionary. So the research of complex networks has been as a research hotspot in the last few years.Because of the traditional network model has no specific areas of the applicable scope,therefore in the process of modeling has failed to effectively use the domain knowledge which is closely related to the complex network structure characteristics and evolution rule of the network. And model in describing the real world network structure characteristics has certain limitation. On the basis of these introduce, we propose a new way to build network model more effective by taking advantage of domain related knowledge. Puts forward the concept of domain cluster, and put forward using the statistical characteristics of complex network classification method. And we figure out two important theories in actor cooperation domain:First in this field the growing of network's nodes is based on the completely graph unit continuously joining the network which contains old and new nodes; Second the possibility of one node is being chosen is based on the times it being chosen before when some new nodes adding to the whole network. Thus we build a network model appropriate for actor collaboration graph by these two theories, which are proved to be more effective by simulation results.At the end of the paper, we discuss the model parameters' influence on the simulation of network structure characteristics through the model simulation comparison and analysis between the networks by using the correlation of network simulation software. On the other hand in the part of our experimental simulation the article also make a comparison between the model simulation network and real world data network, and the simulation results show that the constructed model is very consistent with the real world network structure. The degree distribution, velocity of average pathway and Watts-Strogatz clustering coefficient of the mode are both bring into correspondence with the dataset from real world.