Measuring the preferential attachment mechanism in citation networks

In this paper, we investigated the preferential attachment mechanism (PAM) by considering the dynamic property in papers’ in-degree k for three citation networks. We found that the past citations obtained in different years will have different influences on papers’ attachment rate Π(k,t). We proposed two methods to consider these different influences. One is the Gradually-vanishing Memory Preferential Attachment Mechanism (GMPAM) based on weighted past citations. The other is the Short-term Memory Preferential Attachment Mechanism (SMPAM) based on citations obtained in the recent one-year period. Experiments showed that SMPAM is simpler and more universal in practice. We can just calculate the citations to papers in the recent one-year period to study the papers’ attachment property.     

Effect of the age of papers on the preferential attachment in citation networks

In this paper, we investigated the influences of the age of papers on the preferential attachment on the basis of three actual citation networks. We found that the time dependence of the attachment rate follows a uniform exponentially decreasing function, T(t)∼exp(−λt), in different citation networks. Younger papers are more likely to be cited by new ones than older papers. On the basis of the aging influences, we modified the expression for the preferential attachment, to . Our results show that the modified preferential attachment works well for citation networks.     

Maximum entropy networks are more controllable than preferential attachment networks

A maximum entropy (ME) method to generate typical scale-free networks has been recently introduced. We investigate the controllability of ME networks and Barabási–Albert preferential attachment networks. Our experimental results show that ME networks are significantly more easily controlled than BA networks of the same size and the same degree distribution. Moreover, the control profiles are used to provide insight into control properties of both classes of network. We identify and classify the driver nodes and analyze the connectivity of their neighbors. We find that driver nodes in ME networks have fewer mutual neighbors and that their neighbors have lower average degree. We conclude that the properties of the neighbors of driver node sensitively affect the network controllability. Hence, subtle and important structural differences exist between BA networks and typical scale-free networks of the same degree distribution.     

Nonlinear growth in weighted networks with neighborhood preferential attachment

We propose a nonlinear growing model for weighted networks with two significant characteristics: (i) the new weights triggered by new edges at each time step grow nonlinearly with time; and (ii) a neighborhood local-world exists for local preferential attachment, which is defined as one selected node and its neighbors. Global strength-driven and local weight-driven preferential attachment mechanisms are involved in our model. We study the evolution process through both mathematical analysis and numerical simulation, and find that the model exhibits a wide-range power-law distribution for node degree, strength, and weight. In particular, a nonlinear degree–strength relationship is obtained. This nonlinearity implies that accelerating growth of new weights plays a nontrivial role compared with accelerating growth of edges. Because of the specific local-world model, a small-world property emerges, and a significant hierarchical organization, independent of the parameters, is observed.     

Measuring preferential attachment in growing networks with missing-timelines using Markov chain Monte Carlo

Preferential attachment is widely recognised as the principal driving force behind the evolution of many growing networks, and measuring the extent to which it occurs during the growth of a network is important for explaining its overall structure. Conventional methods require that the timeline of a growing network is known, that is, the order in which the nodes of the network appeared in time is available. But growing network datasets are commonly accompanied by missing-timelines, in which instance the order of the nodes in time cannot be readily ascertained from the data. To address this shortcoming, we propose a Markov chain Monte Carlo algorithm for measuring preferential attachment in growing networks with missing-timelines. Key to our approach is that any growing network model gives rise to a probability distribution over the space of networks. This enables a growing network model to be fitted to a growing network dataset with missing-timeline, allowing not only for the prevalence of preferential attachment to be estimated as a model parameter, but the timeline also. Parameter estimation is achieved by implementing a novel Metropolis–Hastings sampling scheme for updating both the preferential attachment parameter and timeline. A simulation study demonstrates that our method accurately measures the occurrence of preferential attachment in networks generated according to the underlying model. What is more, our approach is illustrated on a small sub-network of the United States patent citation network. Since the timeline for this example is in fact known, we are able to validate our approach against the conventional methods, showing that they give mutually consistent estimates.     

Local preferential attachment model for hierarchical networks

In real-life networks, incomers may only connect to a few others in a local area for their limited information, and individuals in a local area are likely to have close relations. Accordingly, we propose a local preferential attachment model. Here, a local-area-network stands for a node and all its neighbors, and the new nodes perform nonlinear preferential attachment, , in local areas. The stable degree distribution and clustering-degree correlations are analytically obtained. With the increasing of α, the clustering coefficient increases, while assortativity decreases from positive to negative. In addition, by adjusting the parameter α, the model can generate different kinds of degree distribution, from exponential to power-law. The hierarchical organization, independent of α, is the most significant character of this model.     

A self-consistent approach to measure preferential attachment in networks and its application to an inherent structure network

Preferential attachment is one possible way to obtain a scale-free network. We develop a self-consistent method to determine whether preferential attachment occurs during the growth of a network, and to extract the preferential attachment rule using time-dependent data. Model networks are grown with known preferential attachment rules to test the method, which is seen to be robust. The method is then applied to a scale-free inherent structure (IS) network, which represents the connections between minima via transition states on a potential energy landscape. Even though this network is static, we can examine the growth of the network as a function of a threshold energy (rather than time), where only those transition states with energies lower than the threshold energy contribute to the network. For these networks we are able to detect the presence of preferential attachment, and this helps to explain the ubiquity of funnels on potential energy landscapes. However, the scale-free degree distribution shows some differences from that of a model network grown using the obtained preferential attachment rules, implying that other factors are also important in the growth process.     

Weighted self-similar networks under preferential attachment

We introduce an abstract evolutionary formalism that generates weighted networks whose growth under stochastic preferential attachment triggers unrestricted weight rearrangements in existing links. The class of resulting algorithms for different parameter values includes the Barabási-Albert and Barrat-Barthélemy-Vespignani models as special cases. We solve the recursions that describe the average growth to derive exact solutions for the expected degree and strength distribution, the individual strength and weight development and the joint distribution of neighboring degrees. We find that the network exhibits a particular form of self-similarity, namely every sufficiently interconnected node has on average the same constitution of small-degree neighbors as any other node of large degree. Finally we suggest potential applications in several fields of interest.     

Evolving model of weighted networks inspired by scientific collaboration networks

Inspired by scientific collaboration networks (SCN), especially our empirical analysis of econophysicists network, an evolutionary model for weighted networks is proposed. Besides a new vertex added in at every time step, old vertices can also attempt to build up new links, or to reconnect the existing links. The number of connections repeated between two nodes is converted into the weight of the link. This provides a natural way for the evolution of link weight. The path-dependent preferential attachment mechanism with local information is also introduced. It increases the clustering coefficient of the network significantly. The model shows the scale-free phenomena in degree and vertex weight distribution. It also gives well qualitatively consistent behavior with the empirical results.     

Weighted scale-free networks induced by group preferential mechanism

Based on the BBV model [A. Barrat, M. Barthelemy, A. Vespignani, Phys. Rev. Lett. 92 (22) (2004)], we propose a weighted group preferential model, which is generated by the group preferential mechanism. We derive analytically the various statistical properties, such as the distribution of degree, strength and weight, the degree-strength relationship. Finally, we provide a contrast with the BBV model on the synchronization robustness and fragility through numerical simulation.     

Epidemic spreading on weighted complex networks

Nowadays, the emergence of online services provides various multi-relation information to support the comprehensive understanding of the epidemic spreading process. In this Letter, we consider the edge weights to represent such multi-role relations. In addition, we perform detailed analysis of two representative metrics, outbreak threshold and epidemic prevalence, on SIS and SIR models. Both theoretical and simulation results find good agreements with each other. Furthermore, experiments show that, on fully mixed networks, the weight distribution on edges would not affect the epidemic results once the average weight of whole network is fixed. This work may shed some light on the in-depth understanding of epidemic spreading on multi-relation and weighted networks.     

The BA model with finite-precision preferential attachment

Preferential attachment is an indispensable ingredient of the BA model and its variants. In this paper, we modify the BA model by considering the effect of finite-precision preferential attachment, which exists in many real networks. Finite-precision preferential attachment refers to existing nodes with preferential probability Π varying within a certain interval, which is determined by the value of a given precision, being considered to have an equal chance of capturing a new link. The new model reveals a transition from exponential scaling to a power-law distribution along with the increase of the precision. Epidemic dynamics and immunization on the new network are investigated and it is found that the finite-precision effect should be considered in tasks such as infection rate prediction or immunization policy making.     

Generating multi-scaling networks with two types of vertices

A variety of scale-free networks have been created since the pioneer work by Barabási and Albert [Science 286 (1999) 509]. Most of these models are homogeneous since they are composed of the same kind of nodes. In the realistic world, however, elements (nodes or vertices) in the network may play different roles or have different functions. In this work, we develop an alternative way of vertex classification other than the ordinary modularity method by introducing two types of vertices. The interaction between two neighbor vertices is dependent on their types. It is found that the vertex degree exhibits a multi-scaling law distribution with the scaling exponent of each types of vertex adjustable. This network model may exhibit some interesting properties concerning the dynamical processes on it.     

Scaling of average receiving time and average weighted shortest path on weighted Koch networks

In this paper we present weighted Koch networks based on classic Koch networks. A new method is used to determine the average receiving time (ART), whose key step is to write the sum of mean first-passage times (MFPTs) for all nodes to absorption at the trap located at a hub node as a recursive relation. We show that the ART exhibits a sublinear or linear dependence on network order. Thus, the weighted Koch networks are more efficient than classic Koch networks in receiving information. Moreover, average weighted shortest path (AWSP) is calculated. In the infinite network order limit, the AWSP depends on the scaling factor. The weighted Koch network grows unbounded but with the logarithm of the network size, while the weighted shortest paths stay bounded.     

Modeling online social networks based on preferential linking

We study the phenomena of preferential linking in a large-scale evolving online social network and find that the linear preference holds for preferential creation,preferential acceptance,and preferential attachment.Based on the linear preference,we propose an analyzable model,which illustrates the mechanism of network growth and reproduces the process of network evolution.Our simulations demonstrate that the degree distribution of the network produced by the model is in good agreement with that of the real network.This work provides a possible bridge between the micro-mechanisms of network growth and the macrostructures of online social networks.     

Accuracy and precision of methods for community identification in weighted networks

Different algorithms, which take both links and link weights into account for the community structure of weighted networks, have been reported recently. Based on the measure of similarity among community structures introduced in our previous work, in this paper, accuracy and precision of three algorithms are investigated. Results show that Potts model based algorithm and weighted extremal optimization (WEO) algorithm work well on both dense or sparse weighted networks, while weighted Girvan–Newman (WGN) algorithm works well only for relatively sparse networks.     

A power law distribution in patients’ lengths of stay in hospital

The distribution of patients’ lengths of stay in English hospitals is measured by using routinely collected data from 11 years. It is found to be well approximated by a power law distribution spanning over more than three decades. To explain this observation, a theoretical resource allocation model is presented. It is based on iterative long-term scheduling of hospital beds, and its main assumption is that future beds are allocated preferentially. This represents a situation where different parts of the health care system compete for resources, with bargaining powers proportional to current resource levels.     

Network growth with preferential attachment for high indegree and low outdegree

We study the growth of a directed transportation network, such as a food web, in which links carry resources. We propose a growth process in which new nodes (or species) preferentially attach to existing nodes with high indegree (in food-web language, number of prey) and low outdegree (or number of predators). This scheme, which we call inverse preferential attachment, is intended to maximize the amount of resources available to each new node. We show that the outdegree (predator) distribution decays at least exponentially fast for large outdegree and is continuously tunable between an exponential distribution and a delta function. The indegree (prey) distribution is poissonian in the large-network limit.