Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enhancing Cross-domain Link Prediction via Evolution Process Modeling

Xuanwen Huang, Wei Chow, Yize Zhu, Yang Wang, Ziwei Chai, Chunping Wang, Lei Chen, Yang Yang

TL;DR

This paper tackles cross-domain link prediction on dynamic graphs by learning and leveraging evolution patterns across multiple graphs. It introduces CrossLink, a framework that uses a conditioned link generation objective and a decode-only Transformer to model evolution sequences combined with target node representations, enabling parallel training and efficient zero-shot inference on unseen graphs. Empirical results show CrossLink achieves an average AP improvement of 11.40% across eight unseen graphs and can surpass fully supervised End2End baselines on most evaluated datasets, underscoring the value of learning evolution patterns for cross-domain transfer. The findings demonstrate the feasibility and scalability of evolution-aware, cross-domain LP on dynamic graphs, with implications for broader applications and future scaling to larger, more diverse datasets.

Abstract

This work proposes DyExpert, a dynamic graph model for cross-domain link prediction. It can explicitly model historical evolving processes to learn the evolution pattern of a specific downstream graph and subsequently make pattern-specific link predictions. DyExpert adopts a decode-only transformer and is capable of efficiently parallel training and inference by \textit{conditioned link generation} that integrates both evolution modeling and link prediction. DyExpert is trained by extensive dynamic graphs across diverse domains, comprising 6M dynamic edges. Extensive experiments on eight untrained graphs demonstrate that DyExpert achieves state-of-the-art performance in cross-domain link prediction. Compared to the advanced baseline under the same setting, DyExpert achieves an average of 11.40% improvement Average Precision across eight graphs. More impressive, it surpasses the fully supervised performance of 8 advanced baselines on 6 untrained graphs.

Enhancing Cross-domain Link Prediction via Evolution Process Modeling

TL;DR

This paper tackles cross-domain link prediction on dynamic graphs by learning and leveraging evolution patterns across multiple graphs. It introduces CrossLink, a framework that uses a conditioned link generation objective and a decode-only Transformer to model evolution sequences combined with target node representations, enabling parallel training and efficient zero-shot inference on unseen graphs. Empirical results show CrossLink achieves an average AP improvement of 11.40% across eight unseen graphs and can surpass fully supervised End2End baselines on most evaluated datasets, underscoring the value of learning evolution patterns for cross-domain transfer. The findings demonstrate the feasibility and scalability of evolution-aware, cross-domain LP on dynamic graphs, with implications for broader applications and future scaling to larger, more diverse datasets.

Abstract

This work proposes DyExpert, a dynamic graph model for cross-domain link prediction. It can explicitly model historical evolving processes to learn the evolution pattern of a specific downstream graph and subsequently make pattern-specific link predictions. DyExpert adopts a decode-only transformer and is capable of efficiently parallel training and inference by \textit{conditioned link generation} that integrates both evolution modeling and link prediction. DyExpert is trained by extensive dynamic graphs across diverse domains, comprising 6M dynamic edges. Extensive experiments on eight untrained graphs demonstrate that DyExpert achieves state-of-the-art performance in cross-domain link prediction. Compared to the advanced baseline under the same setting, DyExpert achieves an average of 11.40% improvement Average Precision across eight graphs. More impressive, it surpasses the fully supervised performance of 8 advanced baselines on 6 untrained graphs.
Paper Structure (24 sections, 7 equations, 4 figures, 20 tables, 2 algorithms)

This paper contains 24 sections, 7 equations, 4 figures, 20 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. Background
  4. Proposed Method: CrossLink
  5. Overview
  6. Represent Graph Evolution by Sequence
  7. Link Prediction based on Graph Evolution
  8. Training and Inferring
  9. Experiment
  10. Experimental Settings
  11. Experiment Result
  12. Analysis on Multi-domain Training
  13. Conclusion
  14. Dynamic Graph Models
  15. framework of dynamic graph
  16. ...and 9 more sections

Figures (4)

  • Figure 1: (a) shows a case of structure conflict. Graph A follow a triadic closure process, while Graph B exhibits a contrasting process. (b) shows current methods cannot address this conflict. and (c) shows how prediction via modeling evolution, and it can address structure conflict.
  • Figure 2: Framework of CrossLink. (a) Models the graph's evolution process via a sequence of link prediction tasks with ground truths; (b) Evolution-specific link prediction based on both nodes' representations and the evolution process.
  • Figure 3: Analysis result of CrossLink regarding multi-domain training. (a) shows the result of ablation studies, where "w/o" removes a certain component of our model. (b) shows the performance of CrossLink adopts different maximum sequence lengths (both training and inference). (c) indicates the performance on evaluated graphs that model solely trained by a specific graph. See more detailed results in Table \ref{['multidomain-a']}, Table \ref{['multidomain-b']}, and Table \ref{['multidomain-c']}, respectively.
  • Figure 4: Performance of CrossLink with diverse settings. (a) show the performance of the model improves with more training samples. (b) shows the performance of CrossLink is influenced by the number of training graphs. (c) shows the best-hidden size of the model under 6M training samples. (d) further shows the best-hidden size under different training samples. See more details in Appendix.