arXiv subjects

An Zeng

Publications and source records attributed to An Zeng.

At least 55 records · Page 3Linked to original sources

arXiv2014

Uncovering the information core in recommender systems

With the rapid growth of the Internet and overwhelming amount of information that people are confronted with, recommender systems have been developed to effiectively support users' decision-making process in online systems. So far, much attention has been paid to designing new recommendation algorithms and improving existent ones. However, few works considered the different contributions from different users to the performance of a recommender system. Such studies can help us improve the recommendation efficiency by excluding irrelevant users. In this paper, we argue that in each online system there exists a group of core users who carry most of the information for recommendation. With them, the recommender systems can already generate satisfactory recommendation. Our core user extraction method enables the recommender systems to achieve 90% of the accuracy by taking only 20% of the data into account.

cs.IR↗

arXiv2013

Inferring network topology via the propagation process

Inferring the network topology from the dynamics is a fundamental problem with wide applications in geology, biology and even counter-terrorism. Based on the propagation process, we present a simple method to uncover the network topology. The numerical simulation on artificial networks shows that our method enjoys a high accuracy in inferring the network topology. We find the infection rate in the propagation process significantly influences the accuracy, and each network is corresponding to an optimal infection rate. Moreover, the method generally works better in large networks. These finding are confirmed in both real social and nonsocial networks. Finally, the method is extended to directed networks and a similarity measure specific for directed networks is designed.

physics.soc-ph↗

arXiv2013

Information filtering via hybridization of similarity preferential diffusion processes

The recommender system is one of the most promising ways to address the information overload problem in online systems. Based on the personal historical record, the recommender system can find interesting and relevant objects for the user within a huge information space. Many physical processes such as the mass diffusion and heat conduction have been applied to design the recommendation algorithms. The hybridization of these two algorithms has been shown to provide both accurate and diverse recommendation results. In this paper, we proposed two similarity preferential diffusion processes. Extensive experimental analyses on two benchmark data sets demonstrate that both recommendation and accuracy and diversity are improved duet to the similarity preference in the diffusion. The hybridization of the similarity preferential diffusion processes is shown to significantly outperform the state-of-art recommendation algorithm. Finally, our analysis on network sparsity show that there is significant difference between dense and sparse system, indicating that all the former conclusions on recommendation in the literature should be reexamined in sparse system.

cs.IR↗

arXiv2013

Membership in social networks and the application in information filtering

During the past a few years, users' membership in the online system (i.e. the social groups that online users joined) are wildly investigated. Most of these works focus on the detection, formulation and growth of online communities. In this paper, we study users' membership in a coupled system which contains user-group and user-object bipartite networks. By linking users' membership information and their object selection, we find that the users who have collected only a few objects are more likely to be "influenced" by the membership when choosing objects. Moreover, we observe that some users may join many online communities though they collected few objects. Based on these findings, we design a social diffusion recommendation algorithm which can effectively solve the user cold-start problem. Finally, we propose a personalized combination of our method and the hybrid method in [PNAS 107, 4511 (2010)], which leads to a further improvement in the overall recommendation performance.

cs.IR↗

arXiv2013

Information filtering in sparse online systems: recommendation via semi-local diffusion

With the rapid growth of the Internet and overwhelming amount of information and choices that people are confronted with, recommender systems have been developed to effectively support users' decision-making process in the online systems. However, many recommendation algorithms suffer from the data sparsity problem, i.e. the user-object bipartite networks are so sparse that algorithms cannot accurately recommend objects for users. This data sparsity problem makes many well-known recommendation algorithms perform poorly. To solve the problem, we propose a recommendation algorithm based on the semi-local diffusion process on a user-object bipartite network. The numerical simulation on two sparse datasets, Amazon and Bookcross, show that our method significantly outperforms the state-of-the-art methods especially for those small-degree users. Two personalized semi-local diffusion methods are proposed which further improve the recommendation accuracy. Finally, our work indicates that sparse online systems are essentially different from the dense online systems, all the algorithms and conclusions based on dense data should be rechecked again in sparse data.

cs.IR↗

arXiv2013

Path diversity improves the identification of influential spreaders

Identifying influential spreaders in complex networks is a crucial problem which relates to wide applications. Many methods based on the global information such as $k$-shell and PageRank have been applied to rank spreaders. However, most of related previous works overwhelmingly focus on the number of paths for propagation, while whether the paths are diverse enough is usually overlooked. Generally, the spreading ability of a node might not be strong if its propagation depends on one or two paths while the other paths are dead ends. In this Letter, we introduced the concept of path diversity and find that it can largely improve the ranking accuracy. We further propose a local method combining the information of path number and path diversity to identify influential nodes in complex networks. This method is shown to outperform many well-known methods in both undirected and directed networks. Moreover, the efficiency of our method makes it possible to be applied to very large systems.

physics.soc-ph↗

arXiv2013

Extracting the information backbone in online system

Information overload is a serious problem in modern society and many solutions such as recommender system have been proposed to filter out irrelevant information. In the literature, researchers mainly dedicated to improve the recommendation performance (accuracy and diversity) of the algorithms while overlooked the influence of topology of the online user-object bipartite networks. In this paper, we find that some information provided by the bipartite networks is not only redundant but also misleading. With such "less can be more" feature, we design some algorithms to improve the recommendation performance by eliminating some links from the original networks. Moreover, we propose a hybrid method combining the time-aware and topology-aware link removal algorithms to extract the backbone which contains the essential information for the recommender systems. From the practical point of view, our method can improve the performance and reduce the computational time of the recommendation system, thus improve both of their effectiveness and efficiency.

cs.IR↗

arXiv2013

Trend prediction in temporal bipartite networks: the case of Movielens, Netflix, and Digg

Online systems where users purchase or collect items of some kind can be effectively represented by temporal bipartite networks where both nodes and links are added with time. We use this representation to predict which items might become popular in the near future. Various prediction methods are evaluated on three distinct datasets originating from popular online services (Movielens, Netflix, and Digg). We show that the prediction performance can be further enhanced if the user social network is known and centrality of individual users in this network is used to weight their actions.

cs.SI↗

arXiv2013

The role of taste affinity in agent-based models for social recommendation

In the Internet era, online social media emerged as the main tool for sharing opinions and information among individuals. In this work we study an adaptive model of a social network where directed links connect users with similar tastes, and over which information propagates through social recommendation. Agent-based simulations of two different artificial settings for modeling user tastes are compared with patterns seen in real data, suggesting that users differing in their scope of interests is a more realistic assumption than users differing only in their particular interests. We further introduce an extensive set of similarity metrics based on users' past assessments, and evaluate their use in the given social recommendation model with both artificial simulations and real data. Superior recommendation performance is observed for similarity metrics that give preference to users with small scope---who thus act as selective filters in social recommendation.

physics.soc-ph↗

arXiv2012

Improving information filtering via network manipulation

Recommender system is a very promising way to address the problem of overabundant information for online users. Though the information filtering for the online commercial systems received much attention recently, almost all of the previous works are dedicated to design new algorithms and consider the user-item bipartite networks as given and constant information. However, many problems for recommender systems such as the cold-start problem (i.e. low recommendation accuracy for the small degree items) are actually due to the limitation of the underlying user-item bipartite networks. In this letter, we propose a strategy to enhance the performance of the already existing recommendation algorithms by directly manipulating the user-item bipartite networks, namely adding some virtual connections to the networks. Numerical analyses on two benchmark data sets, MovieLens and Netflix, show that our method can remarkably improve the recommendation performance. Specifically, it not only improve the recommendations accuracy (especially for the small degree items), but also help the recommender systems generate more diverse and novel recommendations.

physics.soc-ph↗

arXiv2012

Adaptive social recommendation in a multiple category landscape

People in the Internet era have to cope with the information overload, striving to find what they are interested in, and usually face this situation by following a limited number of sources or friends that best match their interests. A recent line of research, namely adaptive social recommendation, has therefore emerged to optimize the information propagation in social networks and provide users with personalized recommendations. Validation of these methods by agent-based simulations often assumes that the tastes of users and can be represented by binary vectors, with entries denoting users' preferences. In this work we introduce a more realistic assumption that users' tastes are modeled by multiple vectors. We show that within this framework the social recommendation process has a poor outcome. Accordingly, we design novel measures of users' taste similarity that can substantially improve the precision of the recommender system. Finally, we discuss the issue of enhancing the recommendations' diversity while preserving their accuracy.

physics.soc-ph↗

arXiv2012

Spectral coarse graining for random walk in bipartite networks

Many real-world networks display a natural bipartite structure, while analyzing or visualizing large bipartite networks is one of the most challenges. As a result, it is necessary to reduce the complexity of large bipartite systems and preserve the functionality at the same time. We observe, however, the existing coarse graining methods for binary networks fail to work in the bipartite networks. In this paper, we use the spectral analysis to design a coarse graining scheme specifically for bipartite networks and keep their random walk properties unchanged. Numerical analysis on artificial and real-world bipartite networks indicates that our coarse graining scheme could obtain much smaller networks from large ones, keeping most of the relevant spectral properties. Finally, we further validate the coarse graining method by directly comparing the mean first passage time between the original network and the reduced one.

physics.soc-ph↗

arXiv2012

Ranking spreaders by decomposing complex networks

Ranking the nodes' ability for spreading in networks is a fundamental problem which relates to many real applications such as information and disease control. In the previous literatures, a network decomposition procedure called k-shell method has been shown to effectively identify the most influential spreaders. In this paper, we find that the k-shell method have some limitations when it is used to rank all the nodes in the network. We also find that these limitations are due to considering only the links between the remaining nodes (residual degree) while entirely ignoring all the links connecting to the removed nodes (exhausted degree) when decomposing the networks. Accordingly, we propose a mixed degree decomposition (MDD) procedure in which both the residual degree and the exhausted degree are considered. By simulating the epidemic process on the real networks, we show that the MDD method can outperform the k-shell and the degree methods in ranking spreaders. Finally, the influence of the network structure on the performance of the MDD method is discussed.

physics.soc-ph↗

arXiv2012

Enhancing network robustness for malicious attacks

In a recent work [Proc. Natl. Acad. Sci. USA 108, 3838 (2011)], the authors proposed a simple measure for network robustness under malicious attacks on nodes. With a greedy algorithm, they found the optimal structure with respect to this quantity is an onion structure in which high-degree nodes form a core surrounded by rings of nodes with decreasing degree. However, in real networks the failure can also occur in links such as dysfunctional power cables and blocked airlines. Accordingly, complementary to the node-robustness measurement ($R_{n}$), we propose a link-robustness index ($R_{l}$). We show that solely enhancing $R_{n}$ cannot guarantee the improvement of $R_{l}$. Moreover, the structure of $R_{l}$-optimized network is found to be entirely different from that of onion network. In order to design robust networks resistant to more realistic attack condition, we propose a hybrid greedy algorithm which takes both the $R_{n}$ and $R_{l}$ into account. We validate the robustness of our generated networks against malicious attacks mixed with both nodes and links failure. Finally, some economical constraints for swapping the links in real networks are considered and significant improvement in both aspects of robustness are still achieved.

physics.soc-ph↗

arXiv2011

Degree heterogeneity in spatial networks with total cost constraint

Recently, In [Phys. Rev. Lett. 104, 018701 (2010)] the authors studied a spatial network which is constructed from a regular lattice by adding long-range edges (shortcuts) with probability $P_{ij}\sim r_{ij}^{-α}$, where $r_{ij}$ is the Manhattan length of the long-range edges. The total length of the additional edges is subject to a cost constraint ($\sum r=C$). These networks have fixed optimal exponent $α$ for transportation (measured by the average shortest-path length). However, we observe that the degree in such spatial networks is homogenously distributed, which is far different from real networks such as airline systems. In this paper, we propose a method to introduce degree heterogeneity in spatial networks with total cost constraint. Results show that with degree heterogeneity the optimal exponent shifts to a smaller value and the average shortest-path length can further decrease. Moreover, we consider the synchronization on the spatial networks and related results are discussed. Our new model may better reproduce the features of many real transportation systems.

physics.soc-ph↗

arXiv2011

Removing spurious interactions in complex networks

Identifying and removing spurious links in complex networks is a meaningful problem for many real applications and is crucial for improving the reliability of network data, which in turn can lead to a better understanding of the highly interconnected nature of various social, biological and communication systems. In this work we study the features of different simple spurious link elimination methods, revealing that they may lead to the distortion of networks' structural and dynamical properties. Accordingly, we propose a hybrid method which combines similarity-based index and edge-betweenness centrality. We show that our method can effectively eliminate the spurious interactions while leaving the network connected and preserving the network's functionalities.

physics.soc-ph↗

arXiv2011

Behavior patterns of online users and the effect on information filtering

Understanding the structure and evolution of web-based user-object bipartite networks is an important task since they play a fundamental role in online information filtering. In this paper, we focus on investigating the patterns of online users' behavior and the effect on recommendation process. Empirical analysis on the e-commercial systems show that users have significant taste diversity and their interests for niche items highly overlap. Additionally, recommendation process are investigated on both the real networks and the reshuffled networks in which real users' behavior patterns can be gradually destroyed. Our results shows that the performance of personalized recommendation methods is strongly related to the real network structure. Detail study on each item shows that recommendation accuracy for hot items is almost maximum and quite robust to the reshuffling process. However, niche items cannot be accurately recommended after removing users' behavior patterns. Our work also is meaningful in practical sense since it reveals an effective direction to improve the accuracy and the robustness of the existing recommender systems.

physics.soc-ph↗

arXiv2011

Reconstructing directed networks for better synchronization

In this paper, we studied the strategies to enhance synchronization on directed networks by manipulating a fixed number of links. We proposed a centrality-based reconstructing (CBR) method, where the node centrality is measured by the well-known PageRank algorithm. Extensive numerical simulation on many modeled networks demonstrated that the CBR method is more effective in facilitating synchronization than the degree-based reconstructing method and random reconstructing method for adding or removing links. The reason is that CBR method can effectively narrow the incoming degree distribution and reinforce the hierarchical structure of the network. Furthermore, we apply the CBR method to links rewiring procedure where at each step one link is removed and one new link is added. The CBR method helps to decide which links should be removed or added. After several steps, the resulted networks are very close to the optimal structure from the evolutionary optimization algorithm. The numerical simulations on the Kuramoto model further demonstrate that our method has advantage in shortening the convergence time to synchronization on directed networks.

physics.soc-ph↗