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 [1].

Single-View Graph Contrastive Learning with Soft Neighborhood Awareness

Authors: Qingqiang Sun, Chaoqi Chen, Ziyue Qiao, Xubin Zheng, Kai Wang

AAAI 2025 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on diverse node-level tasks demonstrate that our simple single-view GCL framework consistently outperforms existing methods by margins of up to 21.74% (PPI). We evaluate SIGNA on three kinds of node-level tasks, including transductive node classification, inductive node classification, and node clustering.
Researcher Affiliation Academia Qingqiang Sun1, Chaoqi Chen2, Ziyue Qiao1*, Xubin Zheng1, Kai Wang3, 1Great Bay University 2Shenzhen University 3Central South University EMAIL, EMAIL, EMAIL
Pseudocode No The paper describes the methodology using mathematical formulas and descriptive text in the 'Methodology' section, but it does not include a clearly labeled pseudocode block or algorithm.
Open Source Code Yes Code https://github.com/sunisfighting/SIGNA
Open Datasets Yes Datasets. We comprehensively evaluate SIGNA on three kinds of node-level tasks across 7 datasets with various scales and properties (Velickovic et al. 2019; Jiao et al. 2020; Thakoor et al. 2021; Lee, Lee, and Park 2022). Wiki CS, Amazon Photo, Amazon Computers, Coauthor CS, and Coauthor Physics are used for transductive node classification and node clustering tasks. Two larger-scale datasets, Flickr and PPI, are used for inductive node classification on a single graph and multiple graphs, respectively.
Dataset Splits Yes Datasets. We comprehensively evaluate SIGNA on three kinds of node-level tasks across 7 datasets with various scales and properties (Velickovic et al. 2019; Jiao et al. 2020; Thakoor et al. 2021; Lee, Lee, and Park 2022). Wiki CS, Amazon Photo, Amazon Computers, Coauthor CS, and Coauthor Physics are used for transductive node classification and node clustering tasks. Two larger-scale datasets, Flickr and PPI, are used for inductive node classification on a single graph and multiple graphs, respectively. Statistics of these datasets are presented in the Appendix.
Hardware Specification No Implementation Details. Details about encoder implementation, hyperparameter selection, and computing infrastructure are provided in appendix due to the space limitation.
Software Dependencies No The paper mentions implementation details are in the appendix but does not specify any software names with version numbers in the main text.
Experiment Setup No Implementation Details. Details about encoder implementation, hyperparameter selection, and computing infrastructure are provided in appendix due to the space limitation.