Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
Where Graph Meets Heterogeneity: Multi-View Collaborative Graph Experts
Authors: Zhihao Wu, Jinyu Cai, Yunhe Zhang, Jielong Lu, Zhaoliang Chen, Shuman Zhuang, Haishuai Wang
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments on diverse datasets demonstrate Mv CGE s superiority. Experiments on both single and multi-view graph datasets demonstrate that Mv CGE outperforms competitors, showcasing impressive performance and generalization capabilities. |
| Researcher Affiliation | Academia | 1Zhejiang Key Laboratory of Accessible Perception and Intelligent Systems, College of Computer Science and Technology, Zhejiang University 2Institute of Data Science, National University of Singapore 3Department of Computer and Information Science, University of Macau 4Department of Computer Science, Hong Kong Baptist University 5College of Computer and Data Science, Fuzhou University |
| Pseudocode | No | The paper describes the methodology in prose and mathematical formulations, and illustrates the architecture with figures, but does not present any structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | Question: Does the paper provide open access to the data and code, with sufficient instructions to faithfully reproduce the main experimental results, as described in supplemental material? Answer: [Yes] Justification: The code is available in an anonymous link. |
| Open Datasets | Yes | To evaluate the effectiveness of Mv CGE, we conducted experiments on ten datasets, including five multi-view graph datasets (ACM, DBLP, IMDB, YELP, AMINER) and five typical single-view graph datasets (ACM, Citeseer, Cora Full, Flickr, UAI) plus one large-scale multi-view graph Freebase and one large-scale single-view graph OGBN-ar Xiv. |
| Dataset Splits | Yes | Here, the training ratio is set to 20% and the Macro F1 and Micro F1 scores are recorded in Table 1. Following the commonly used semi-supervised node classification settings, we randomly select 20 samples per class for training, 500 samples for validation, and 1,000 samples for testing, with the detailed results presented in Table 2. |
| Hardware Specification | Yes | Question: For each experiment, does the paper provide sufficient information on the computer resources (type of compute workers, memory, time of execution) needed to reproduce the experiments? Answer: [Yes] Justification: The computer resources are detailed in the Appendix. |
| Software Dependencies | No | The paper does not explicitly provide specific software dependencies with version numbers (e.g., library or solver names with version numbers like Python 3.8, CPLEX 12.4) in the main text. While the NeurIPS checklist indicates that information needed to reproduce results is available in the appendix and source code, these specific details are not directly accessible or stated in the main body provided. |
| Experiment Setup | Yes | To adapt Mv CGE to single-view graphs, we generate a supplementary view for each single-view dataset based on the k-Nearest Neighbor algorithm, where we set k = 10. It can be seen that model performance exhibits similar stable fluctuation patterns for both α and β across ACM and YELP datasets, with the optimal performance achieved at α = 1e 1 or 1e 2 and β = 1e 1 or 1e 3, demonstrating that the integrated two losses effectively capture complementary information and consistently enhance performance when appropriate parameter values are selected. Question: Does the paper specify all the training and test details (e.g., data splits, hyperparameters, how they were chosen, type of optimizer, etc.) necessary to understand the results? Answer: [Yes] Justification: The details are included in Appendix. |