Link Prediction in Heterogeneous Collaboration Networks

Abstract

Abstract. Traditional link prediction techniques primarily focus on the effect of potential linkages on the local network neighborhood or the paths between nodes. In this article, we study both supervised and unsupervised link prediction in networks where instances can simultaneously belong to multiple communities, engendering different types of collaborations. Links in these networks arise from heterogeneous causes, limiting the performance of predictors that treat all links homogeneously. To solve this problem, we introduce a new supervised link prediction framework, Link Prediction using Social Features (LPSF ), which incorporates a reweighting scheme for the network based on nodes' features extracted from patterns of prominent interactions across the network.Experiments on coauthorship networks demonstrate that the choice for measuring link weights can be critical for the link prediction task. Our proposed reweighting method in LPSF better expresses the intrinsic relationship between nodes and improves prediction accuracy for supervised link prediction techniques. We also compare the unsupervised performance of the individual features used within LPSF with two new diffusion-based methods: LPDP (Link Prediction using Diffusion Process) and LPDM (Link Prediction using Diffusion Maps). Experiments demonstrate that LPDP is able to identify similar node pairs, even far away ones, that are connected by weak ties in the coauthorship network using the diffusion process; however, reweighting the network has little impact on prediction performance.

Job Insecurity in Nursing: A Bibliometric Analysis

“…We used the author’s affiliation available in the WoS database, and an additional Google search was performed in its absence [ 33 ]. The removal of these ambiguities is necessary since they impair bibliometric analyses such as co-authorial link predictions [ 34 ], collaborative network analysis [ 35 ], and citation network analysis [ 36 ].…”

Research on Weighted Directed Dynamic Multiplexing Network of World Grain Trade Based on Improved MLP Framework

“…The principle of PA is that the more links one already has, the greater the likelihood of new links being generated between that node pair [25]. Redefined as:…”

Leveraging Network Dynamics for Improved Link Prediction

“…In our work, we only use network-based features, since those are the easiest to generalize across different types of networks; both proximity and aggregated features require more feature engineering to transfer to different datasets. Wang and Sukthankar [11] promoted the importance of social features in both supervised and unsupervised link prediction; social features aim to express the community membership of nodes and can be used to construct alternate distance metrics. However we believe that rate generalizes better across different types of dynamic networks; moreover it can be easily combined with dataset-specific feature sets.…”

“…Approaches to the link prediction problem are commonly categorized as being unsupervised [4,7,[11][12][13] or supervised [8,[14][15][16]. In unsupervised approaches, pairs of non connected nodes are initially ranked according to a chosen similarity metric (for instance, the number of common neighbors) [17,18].…”