Common neighbours and the local-community-paradigm for topological link prediction in bipartite networks

Abstract

Bipartite networks are powerful descriptions of complex systems characterized by two different classes of nodes and connections allowed only across but not within the two classes. Unveiling physical principles, building theories and suggesting physical models to predict bipartite links such as productconsumer connections in recommendation systems or drug-target interactions in molecular networks can provide priceless information to improve e-commerce or to accelerate pharmaceutical research. The prediction of nonobserved connections starting from those already present in the topology of a network is known as the link-prediction problem. It represents an important subject both in many-body interaction theory in physics and in new algorithms for applied tools in computer science. The rationale is that the existing connectivity structure of a network can suggest where new connections can appear with higher likelihood in an evolving network, or where nonobserved connections are missing in a partially known network. Surprisingly, current complex network theory presents a theoretical bottle-neck: a general framework for local-based link prediction directly in the bipartite domain is missing. Here, we overcome this theoretical obstacle and present a formal definition of common neighbour index and local-community-paradigm (LCP) for bipartite networks. As a consequence, we are able to introduce the first node-neighbourhood-based and LCP-based models for topological link prediction that utilize the bipartite domain. We performed link prediction evaluations in several networks of different size and of disparate origin, including technological, social and biological systems. Our models significantly improve topological prediction in many bipartite networks because they exploit local physical driving-forces that participate in the formation and organization of many real-world bipartite networks. Furthermore, we present a local-based formalism that allows to intuitively implement neighbourhood-based link prediction entirely in the bipartite domain.

A perturbation-based framework for link prediction via non-negative matrix factorization

“…Model-based methods are based on an explicit deterministic model that simulates physical mechanism behind the network organization. Model-learning methods are based on implicit model-learning: providing at every step a different solution that can converge to hidden the network evolution by many times of iterations12. NMF 1, NMF 2, NMF  −  D 1, NMF  −  A 1, NMF  −  D 2, NMF  −  A 2, SPM are model-learning methods and the other methods are model-based methods.…”

“…For example, Common Neighbours (CN) index is defined as the number of common neighbours of the two nodes in the networks9, Jaccard index is defined as the number of common neighbours of two nodes divided by interaction set of their degrees10, Katz index is based on the ensemble of all paths between each node pair. Cannistraci-resource-allocation (CRA) is a powerful local and parameter-free similarity-based index for link prediction in both monopartite network and bipartite network, and it is based on the local-community-paradigm1112, which is a theory recently proposed to model local-topology-dependent link-growth in complex networks. In brief, similarity-based indices can be local or global, parameter-free or parameter-dependent, simple or complex.…”

Mutual information model for link prediction in heterogeneous complex networks

“…It has been demonstrated through extensive experimental evaluations that LCP based indices could provide better performance predictions compared to other conventional indices. This approach has been also successfully extended on the bipartite complex networks8.…”

Mining user behavior in location-based social networks

“…Because the community structure encloses itself different structural and social properties as social ties, triadic closure, structural holes, and others, much of researchers focused on the use of social theory on link prediction have put their attention on better use the community information to improve the link prediction accuracy. Therefore, based on the primary works of Zheleva et al (2008), Soundarajan and Hopcroft (2012), and Valverde-Rebaza and Lopes (2012a), a considerable amount of works using community information to enhance the link prediction have been proposed (HOSEINI; HASHEMI; HAMZEH, 2012; CANNISTRACI; ALANIS-LOBATO; RAVASI, 2013;KEMAL;TSUYOSHI et al, 2014;MALLEK et al, 2015;DAMINELLI et al, 2015;DING et al, 2016;MA et al, 2016;KUANG;YU, 2016;CAIYAN;BISWAS;BISWAS, 2017). Considering the fact that social grouping is a natural behavior of users in OSNs and that social groups can be associated directly with community structure (YANG; LESKOVEC, 2015), we will use the concepts of social group and community interchangeably throughout this thesis.…”

“…Several link prediction methods use a variety of information sources available in OSNs. Some of the information sources commonly used are: i) homophily (MCPHERSON;SMITH-LOVIN;COOK, 2001;CHANG et al, 2014;FARALLI;STILO;VELARDI, 2015), ii) social ties (LÜ; ZHOU, 2010; SOCIEVOLE; RANGO; MARANO, 2013;KLEINBERG, 2014;XU et al, 2017) and iii) communities (ZHELEVA et al, 2008;SOUNDARAJAN;HOPCROFT, 2012;LOPES, 2012a;HOSEINI;HASHEMI;HAMZEH, 2012; CANNIS-TRACI; ALANIS-LOBATO; RAVASI, 2013;KEMAL;TSUYOSHI et al, 2014;MALLEK et al, 2015;DAMINELLI et al, 2015;DING et al, 2016;MA et al, 2016;KUANG;YU, 2016;CAIYAN;BISWAS;BISWAS, 2017). Because some researches have shown that community information can drastically improve the link prediction accuracy XU, 2012;, most of the research efforts have been directed to better explore this information source.…”