Dual stream fusion link prediction for sparse graph based on variational graph autoencoder and pairwise learning

Information Processing & Management, Journal Year: 2025, Volume and Issue: 62(3), P. 104073 - 104073

Published: Jan. 24, 2025

Language: Английский

---

Adaptive node similarity for DropEdge

Neurocomputing, Journal Year: 2025, Volume and Issue: unknown, P. 129574 - 129574

Published: Feb. 1, 2025

Language: Английский

---

Revealing spatiotemporal connections in container hub ports under adverse events through link prediction

Journal of Transport Geography, Journal Year: 2025, Volume and Issue: 125, P. 104198 - 104198

Published: March 19, 2025

Language: Английский

---

Joint Structural Balance Feature Representation on Graph Convolution Networks for Link Prediction

Cognitive Computation, Journal Year: 2025, Volume and Issue: 17(3)

Published: April 29, 2025

Language: Английский

---

Multilayer network link prediction considering multiple correlation features

Expert Systems with Applications, Journal Year: 2025, Volume and Issue: unknown, P. 127700 - 127700

Published: May 1, 2025

Language: Английский

---

Reliable multiplex semi-local random walk based on influential nodes to improve link prediction in complex networks

Artificial Intelligence Review, Journal Year: 2024, Volume and Issue: 57(6)

Published: May 27, 2024

Abstract In recent years, the exponential growth of online social networks as complex has presented challenges in expanding and forging new connections. Link prediction emerges a crucial technique to anticipate future relationships among users, leveraging current network state address this challenge effectively. While link models on monoplex have well-established history, exploration similar tasks multilayer garnered considerable attention. Extracting topological multimodal features for weighting links can improve weighted networks. Meanwhile, establishing reliable trustworthy paths between users is useful way create metrics that convert unweighted similarity. The local random walk widely used predicting aim paper develop semi-local over network, which denoted Reliable Multiplex semi-Local Random Walk (RMLRW). RMLRW leverages paths, integrating intra-layer inter-layer information from multiplex conduct biased within target layer order make scalable, we walk-based embedding represent lower-dimensional space while preserving its original characteristics. Extensive experimental studies several real-world demonstrate performance assurance compared equivalent methods. Specifically, improves average f-measure by 3.2% 2.5% SEM-Path MLRW, respectively.

Language: Английский

---

TLFSL: link prediction in multilayer social networks using trustworthy Lévy-flight semi-local random walk

Journal of Complex Networks, Journal Year: 2024, Volume and Issue: 12(4)

Published: June 24, 2024

Abstract As the landscape of online social networks continues to evolve, task expanding connections and uncovering novel relationships presents a growing complexity. Link prediction emerges as crucial strategy, harnessing current network dynamics forecast future interactions among users. While traditional single-layer link models boast storied legacy, recent attention has shifted towards tackling analogous challenges within realm multilayer networks. This paradigm shift underscores critical role extracting topological multimodal features effectively evaluate weights, thereby enriching weighted Furthermore, establishment trustworthy pathways between users pivotal tactic for translating unweighted similarities into meaningful metrics. Leveraging foundational principles local random walk techniques, this paper introduces Lévy-flight semi-local (TLFSL) framework in By seamlessly integrating intralayer interlayer information, TLFSL harnesses dependable mechanism anticipate new links target layers Traditional techniques often overlook global relationships, they confine path exploration immediate neighbours. However, absence direct edge nodes does not necessarily imply lack relationship; with semantic affinity may be spatially distant network. To overcome limitation, we introduce concept walk, which enables walker hopping wider perspective. Meanwhile, includes distributed community detection strategy improve performance dealing large-scale Rigorous experimentation across diverse real-world consistently demonstrates TLFSL’s superior compared equivalent methods.

Language: Английский

---

Introducing new Link Prediction Measures and an Evaluation Metric

Research Square (Research Square), Journal Year: 2024, Volume and Issue: unknown

Published: Aug. 5, 2024

Link prediction plays a critical role in network analysis as it tackles the task of predicting missing or future connections within given network. A wide array link measures has been proposed to estimate likelihood existence between nodes, aiming uncover hidden relationships and anticipate formation new connections. This holds particular significance various domains, including social networks, biological transportation recommender systems. Nonetheless, evaluating effectiveness these poses challenges, encompassing both themselves employed evaluation metrics. article offers comprehensive overview existing measures, shedding light on their limitations addressing associated challenges efficacy. Moreover, we propose enhancements overcome entailing modifications introduction novel By tackling issues head-on, our objective is enhance accuracy reliability analysis.

Language: Английский

---

A novel recommendation-based framework for reconnecting and selecting the efficient friendship path in the heterogeneous social IoT network

Computer Networks, Journal Year: 2024, Volume and Issue: unknown, P. 111016 - 111016

Published: Dec. 1, 2024

Language: Английский

---