Narrow Results

In this paper, we propose a spreading activation approach for collaborative filtering (SA-CF). By using the opinion spreading process, the similarity between any users can be obtained. The algorithm has remarkably higher accuracy than the standard collaborative filtering using the Pearson correlation. Furthermore, we introduce a free parameter β to regulate the contributions of objects to user–user correlations. The numerical results indicate that decreasing the influence of popular objects can further improve the algorithmic accuracy and personality. We argue that a better algorithm should simultaneously require less computation and generate higher accuracy. Accordingly, we further propose an algorithm involving only the top-N similar neighbors for each target user, which has both less computational complexity and higher algorithmic accuracy.

In this paper, based on a weighted projection of the user-object bipartite network, we study the effects of user tastes on the mass-diffusion-based personalized recommendation algorithm, where a user's tastes or interests are defined by the average degree of the objects he has collected. We argue that the initial recommendation power located on the objects should be determined by both of their degree and the user's tastes. By introducing a tunable parameter, the user taste effects on the configuration of initial recommendation power distribution are investigated. The numerical results indicate that the presented algorithm could improve the accuracy, measured by the average ranking score. More importantly, we find that when the data is sparse, the algorithm should give more recommendation power to the objects whose degrees are close to the user's tastes, while when the data becomes dense, it should assign more power on the objects whose degrees are significantly different from user's tastes.

In this paper, the statistical property, namely degree correlation between users and objects, is taken into account and be embedded into the similarity index of collaborative filtering (CF) algorithm to improve the algorithmic performance. The numerical simulation on a benchmark data set shows that the algorithmic accuracy of the presented algorithm, measured by the average ranking score, is improved by 18.19% in the optimal case. The statistical analysis on the product distribution of the user and object degrees indicate that, in the optimal case, the distribution obeys the power-law and the exponential is equal to -2.33. Numerical results show that the presented algorithm can provide more diverse and less popular recommendations, for example, when the recommendation list contains 10 objects, the diversity, measured by the hamming distance, is improved by 21.90%. Since all of the real recommendation data evolving with time, this work may shed some light on the adaptive recommendation algorithm which could change its parameter automatically according to the statistical properties of the user-object bipartite network.

In this paper, the statistical property of the bipartite network, namely clustering coefficient C₄ is taken into account and be embedded into the collaborative filtering (CF) algorithm to improve the algorithmic accuracy and diversity. In the improved CF algorithm, the user similarity is defined by the mass diffusion process, and we argue that the object clustering C₄ of the bipartite network should be considered to improve the user similarity measurement. The statistical result shows that the clustering coefficient of the MovieLens data approximately has Poisson distribution. By considering the clustering effects of object nodes, the numerical simulation on a benchmark data set shows that the accuracy of the improved algorithm, measured by the average ranking score and precision, could be improved 15.3 and 13.0%, respectively, in the optimal case. In addition, numerical results show that the improved algorithm can provide more diverse recommendation results, for example, when the recommendation list contains 20 objects, the diversity, measured by the hamming distance, is improved by 28.7%. Since all of the real recommendation data are evolving with time, this work may shed some light on the adaptive recommendation algorithm according to the statistical properties of the user-object bipartite network.

The clustering coefficient of the bipartite network, C₄, has been widely used to investigate the statistical properties of the user-object systems. In this paper, we empirically analyze the evolution patterns of C₄ for a nine year MovieLens data set, where C₄ is used to describe the diversity of the user interest. First, we divide the MovieLens data set into fractions according to the time intervals and calculate C₄ of each fraction. The empirical results show that, the diversity of the user interest changes periodically with a round of one year, which reaches the smallest value in spring, then increases to the maximum value in autumn and begins to decrease in winter. Furthermore, a null model is proposed to compare with the empirical results, which is constructed in the following way. Each user selects each object with a turnable probability p, and the numbers of users and objects are equal to that of the real MovieLens data set. The comparison result indicates that the user activity has greatly influenced the structure of the user-object bipartite network, and users with the same degree information may have two totally different clustering coefficients. On the other hand, the same clustering coefficient also corresponds to different degrees. Therefore, we need to take the clustering coefficient into consideration together with the degree information when describing the user selection activity.

In this paper, two general bipartite network evolving models for energy supply-demand network with local-world are proposed. The node weight distribution, the "shifting coefficient" and the scaling exponent of two different kinds of nodes are presented by the mean-field theory. The numerical results of the node weight distribution and the edge weight distribution are also investigated. The production's shifted power law (SPL) distribution of coal enterprises and the installed capacity's distribution of power plants in the US are obtained from the empirical analysis. Numerical simulations and empirical results are given to verify the theoretical results.

In this paper, we propose weighted symmetric binary matrix factorization (wSBMF) framework to detect overlapping communities in bipartite networks, which describes the relationships between two types of nodes. Our method improves performance by recognizing the distinction between two types of missing edges — ones among the nodes in each node type and the others between two node types. Our method can also explicitly assign community membership and distinguish outliers from overlapping nodes, as well as incorporating existing knowledge on the network. We propose a generalized partition density for bipartite networks as a quality function, which identifies the most appropriate number of communities. The experimental results on both synthetic and real-world networks demonstrate the effectiveness of our method.

The mobility pattern of a population plays a key role in the spread of epidemics. Despite extensive work on epidemic spreading, little attention has been paid to the impact of subpopulation size. This paper investigates the spread of epidemics on a bipartite metapopulation network considering recurrent mobility patterns and different sizes of subpopulations. With the Markovian process approach, the epidemic threshold can be predicted as a function of subpopulation size and epidemic parameters. Simulation and theoretical results indicate that there exists a critical mobility intensity below which the epidemic will be eliminated, while limiting the size of the subpopulation can suppress the epidemic. Additionally, the epidemic threshold will approach zero when the size of the public area is large. The results can help the prevention of epidemic spreading under recurrent crowd mobility.

Recommender systems have proven to be an effective method to deal with the problem of information overload in finding interesting products. It is still a challenge to increase the accuracy and diversity of recommendation algorithms to fulfill users' preferences. To provide a better solution, in this paper, we propose a novel recommendation algorithm based on heterogeneous diffusion process on a user-object bipartite network. This algorithm generates personalized recommendation results on the basis of the physical dynamic feature of resources diffusion which is influenced by objects' degrees and users' interest degrees. Detailed numerical analysis on two benchmark datasets shows that the presented algorithm is of high accuracy, and also generates more diversity.

Real-world relations are often represented as bipartite networks, such as paper-author networks and event-attendee networks. Extracting dense subnetworks (communities) from bipartite networks and evaluating their qualities are practically important research topics. As the attempts for evaluating divisions of bipartite networks, Guimera and Barber propose bipartite modularities. This paper discusses the properties of these bipartite modularities and proposes another bipartite modularity that allows one-to-many correspondence of communities of different vertex types.

Distribution of producer size in a globalized market is a complex market phenomena, which is affected by the market behavior of consumers such as the loyalty of consumers to producers and the purchasing power of consumers, as well as the trade barriers among countries. In the present paper, in order to study the distribution of producer size in the globalized market, we construct a bipartite network that consists of consumers and producers with community structure. We find that the distribution of producer size in each community in a multi-community network can be projected to that in one-community bipartite network by mapping the globalized market behavior of consumers to an isolated market behavior. The mapped market behavior is dependent on the trade barriers among communities. The distribution of producer size in globalized market is thereby dependent on the mapped loyalty of consumers and the mapped growing rate of purchasing power. Furthermore, simulation results show that the distribution of producer size differs community by community. It follows the power-law distribution if both the mapped loyalty of consumers and growing rate are high.

The relations between agents of complex networks are generally determined by their attributes, so we can instead study the corresponding bipartite network formed by agents and their attributes to gain a higher-dimensional perspective. General bipartite community detecting algorithms implicitly contain a fixed generation step to determine the intra-correlations of the two separate vertex sets (denoted as instance set and attribute set), thus ignoring problem-related heuristics. Inspired by this, we propose a bi-community detection framework concerning the problem-related features that directly takes such intra-correlations into account, and can be freely combined with different objective functions and optimization algorithms to cope with various network structures such as directed graphs with negative edge weights. The framework is adopted to analyze international relations on the dispute and alliance datasets, whose results contain the relevant events that support the establishment of each community and are highly consistent with Huntington’s theory. In addition, we analyze the impact of the instance–instance, instance–attribute, and attribute–attribute relations on the detection result through control experiments, and conclude that for the general community searching algorithms (including the bi-community case), appropriately taking these three relations together into account can help obtain different reasonable detection results.

Environmental exposure is a key factor of understanding health and diseases. Beyond genetic propensities, many disorders are, in part, caused by human interaction with harmful substances in the water, the soil, or the air. Limited data is available on a disease or substance basis. However, we compile a global repository from literature surveys matching environmental chemical substances exposure with human disorders. We build a bipartite network linking 60 substances to over 150 disease phenotypes. We quantitatively and qualitatively analyze the network and its projections as simple networks. We identify mercury, lead and cadmium as associated with the largest number of disorders. Symmetrically, we show that breast cancer, harm to the fetus and non-Hodgkin’s lymphoma are associated with the most environmental chemicals. We conduct statistical analysis of how vertices with similar characteristics form the network interactions. This dyadicity and heterophilicity measures the tendencies of vertices with similar properties to either connect to one-another. We study the dyadic distribution of the substance classes in the networks show that, for instance, tobacco smoke compounds, parabens and heavy metals tend to be connected, which hint at common disease causing factors, whereas fungicides and phytoestrogens do not. We build an exposure network at the systems level. The information gathered in this study is meant to be complementary to the genome and help us understand complex diseases, their commonalities, their causes, and how to prevent and treat them.