OpenNovelty.org

PLANETALIGN: A Comprehensive Python Library for Benchmarking Network Alignment

ICLR 2026 Conference SubmissionAnonymous Authors
OpenReview Score: 6.0Download Report PDF
Network AlignmentGraph Machine Learning
Abstract:

Network alignment (NA) aims to identify node correspondence across different networks and serves as a critical cornerstone behind various downstream multi-network learning tasks. Despite growing research in NA, there lacks a comprehensive library that facilitates the systematic development and benchmarking of NA methods. In this work, we introduce PLANETALIGN, a comprehensive Python library for network alignment that features a rich collection of built-in datasets, methods, and evaluation pipelines with easy-to-use APIs. Specifically, PLANETALIGN integrates 18 datasets and 14 NA methods with extensible APIs for easy use and development of NA methods. Our standardized evaluation pipeline encompasses a wide range of metrics, enabling a systematic assessment of the effectiveness, scalability, and robustness of NA methods. Through extensive comparative studies, we reveal practical insights into the strengths and limitations of existing NA methods. We hope that PLANETALIGN can foster a deeper understanding of the NA problem and facilitate the development and benchmarking of more effective, scalable, and robust methods in the future. The The source code of PLANETALIGN is available at https://anonymous.4open.science/r/PlanetAlign-E9BA

Disclaimer
This report is AI-GENERATED using Large Language Models and WisPaper (A scholar search engine). It analyzes academic papers' tasks and contributions against retrieved prior work. While this system identifies POTENTIAL overlaps and novel directions, ITS COVERAGE IS NOT EXHAUSTIVE AND JUDGMENTS ARE APPROXIMATE. These results are intended to assist human reviewers and SHOULD NOT be relied upon as a definitive verdict on novelty.
NOTE that some papers exist in multiple, slightly different versions (e.g., with different titles or URLs). The system may retrieve several versions of the same underlying work. The current automated pipeline does not reliably align or distinguish these cases, so human reviewers will need to disambiguate them manually.
If you have any questions, please contact: mingzhang23@m.fudan.edu.cn

Overview

Overall Novelty Assessment

The paper introduces PLANETALIGN, a comprehensive Python library designed to standardize network alignment research through integrated datasets, methods, and evaluation pipelines. Within the taxonomy, it resides in the 'Alignment Benchmarking and Survey' leaf alongside three sibling papers focused on surveys and methodological reviews. This leaf represents a small but critical segment of the field, containing only four papers out of fifty total across the taxonomy. The positioning reflects a sparse research direction dedicated to consolidation and evaluation infrastructure rather than algorithmic innovation, distinguishing it from the more crowded methodological branches like Embedding-Based or GNN-Based Alignment.

The taxonomy reveals that PLANETALIGN sits adjacent to nine major methodological branches, including Embedding-Based Methods, GNN-Based Alignment, and Optimal Transport approaches, which collectively house the algorithmic contributions the library aims to benchmark. The 'Alignment Benchmarking and Survey' branch explicitly excludes specific alignment algorithms, focusing instead on comprehensive frameworks for reviewing and comparing methods. This structural separation underscores the library's role as meta-research infrastructure, bridging diverse methodological paradigms rather than advancing a particular algorithmic direction. Neighboring leaves like Multi-Graph Learning and Specialized Alignment Contexts address orthogonal concerns such as pretraining strategies and domain-specific constraints.

Among thirty candidates examined across three contributions, none yielded clear refutations. The first contribution, the comprehensive library itself, examined ten candidates with zero refutable overlaps, suggesting limited prior work on unified Python libraries for network alignment. The standardized evaluation pipeline and extensible APIs each examined ten candidates with similar results, indicating that systematic benchmarking infrastructure remains underexplored. However, the limited search scope means these findings reflect top-thirty semantic matches rather than exhaustive coverage. The absence of refutations among examined candidates suggests the library addresses a gap in tooling and standardization, though broader literature may contain related efforts not captured in this analysis.

Based on the limited search scope, PLANETALIGN appears to occupy a relatively novel position within the benchmarking and infrastructure space, though the analysis covers only top-thirty semantic matches and does not guarantee exhaustive prior work identification. The sparse population of its taxonomy leaf and the absence of refutations among examined candidates suggest the work addresses an underserved need for standardized evaluation tools, though definitive novelty claims require broader literature coverage beyond the current search parameters.

Taxonomy

Core-task Taxonomy Papers
50
3
Claimed Contributions
30
Contribution Candidate Papers Compared
0
Refutable Paper

Research Landscape Overview

Core task: network alignment across different networks. The field addresses the challenge of identifying corresponding nodes or structures across distinct networks, a problem that arises in domains ranging from social network integration to biological network comparison. The taxonomy reveals a rich landscape organized around several methodological pillars. Embedding-Based Network Alignment Methods and Graph Neural Network-Based Alignment form the backbone of modern approaches, leveraging learned representations to capture node similarities across networks. Optimal Transport and Graph Matching Methods provide a principled mathematical framework for alignment, while Probabilistic and Statistical Alignment and Topological and Spectral Alignment offer complementary perspectives grounded in uncertainty modeling and structural invariants. Specialized Alignment Contexts address domain-specific challenges such as biological networks or user alignment across platforms, and Multi-Graph Learning and Pretraining explore how shared knowledge can improve alignment quality. The Alignment Benchmarking and Survey branch consolidates evaluation practices and synthesizes progress across these diverse directions. Recent work highlights tensions between scalability, accuracy, and the need for supervision. Many studies explore embedding techniques that balance expressiveness with computational efficiency, as seen in works like Multiple Embedding Alignment[5] and Cross-network Embedding[13], while others investigate how graph neural networks can refine alignment through message-passing architectures. PLANETALIGN[0] situates itself within the Alignment Benchmarking and Survey branch alongside Network Alignment Survey[19] and Network Alignment Advances[27], contributing to the consolidation of evaluation standards and methodological insights. Unlike purely algorithmic contributions, this line of work emphasizes systematic comparison and the identification of open challenges, helping researchers navigate the trade-offs between different alignment paradigms. The interplay between benchmarking efforts and algorithmic innovation remains central to advancing the field, as comprehensive surveys inform the design of next-generation methods that address gaps in existing approaches.

Claimed Contributions

PLANETALIGN: A comprehensive Python library for network alignment

The authors present PLANETALIGN, an open-source PyTorch-based library designed for unified evaluation and streamlined development of network alignment methods. It integrates 18 datasets across 6 domains and 14 NA methods spanning consistency-based, embedding-based, and optimal transport-based approaches, with standardized evaluation pipelines and extensible APIs.

10 retrieved papers
Standardized evaluation pipeline with diverse benchmarking tools

The library provides a comprehensive evaluation framework that includes standard metrics (Hits@K, MRR), consistent dataset splits for reproducibility, and benchmarking tools to assess NA methods across multiple dimensions including effectiveness, scalability, and robustness under various graph noises and supervision levels.

10 retrieved papers
Extensible APIs and utility functions for method development

The library offers carefully designed base classes and unified APIs that enable users to implement custom NA methods and integrate custom datasets with minimal code. It includes commonly used utility functions such as random walk with restart embedding and anchor-based embedding to facilitate development.

10 retrieved papers

Core Task Comparisons

Comparisons with papers in the same taxonomy category

Contribution Analysis

Detailed comparisons for each claimed contribution

Contribution

PLANETALIGN: A comprehensive Python library for network alignment

The authors present PLANETALIGN, an open-source PyTorch-based library designed for unified evaluation and streamlined development of network alignment methods. It integrates 18 datasets across 6 domains and 14 NA methods spanning consistency-based, embedding-based, and optimal transport-based approaches, with standardized evaluation pipelines and extensible APIs.

Contribution

Standardized evaluation pipeline with diverse benchmarking tools

The library provides a comprehensive evaluation framework that includes standard metrics (Hits@K, MRR), consistent dataset splits for reproducibility, and benchmarking tools to assess NA methods across multiple dimensions including effectiveness, scalability, and robustness under various graph noises and supervision levels.

Contribution

Extensible APIs and utility functions for method development

The library offers carefully designed base classes and unified APIs that enable users to implement custom NA methods and integrate custom datasets with minimal code. It includes commonly used utility functions such as random walk with restart embedding and anchor-based embedding to facilitate development.

Table of Contents