{"paper":{"title":"SP-GCRL: Influence Maximization on Incomplete Social Graphs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"SP-GCRL learns end-to-end seed selection policies for influence maximization on incomplete social graphs using contrastive representations and a nonlinear diffusion model.","cross_cats":["cs.AI"],"primary_cat":"cs.SI","authors_text":"Haohua Niu, Hao Li, Jiao Liang, Lingfeng Zhang, Luca Rossi, Yuxuan Yang, Zongfu Luo","submitted_at":"2026-03-31T09:44:20Z","abstract_excerpt":"Influence maximization (IM) in real platforms is challenged by incomplete, noisy social graphs and non-stationary diffusion dynamics. We propose SP-GCRL, a social-propagation-aware graph contrastive reinforcement learning framework that learns end-to-end seed selection under partial observability.We first introduce a social-propagation-aware nonlinear diffusion function to model reinforcement/diminishing effects and probability drift under repeated exposure; we then construct dual structural views and perform contrastive learning to obtain node representations robust to missing edges and weak "},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Experiments on multiple real-world networks show that SP-GCRL achieves significant gains over heuristic and learning-based baselines across budgets and topologies, while maintaining strong large-scale scalability.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The proposed social-propagation-aware nonlinear diffusion function correctly models reinforcement, diminishing returns, and probability drift under repeated exposure in real incomplete graphs.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"SP-GCRL combines a nonlinear social diffusion model, dual-view contrastive learning for robust node embeddings, a GAT surrogate, and DDQN to learn end-to-end seed selection policies for influence maximization under partial graph observability.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"SP-GCRL learns end-to-end seed selection policies for influence maximization on incomplete social graphs using contrastive representations and a nonlinear diffusion model.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"4fbc9d99e03a12d81dfd47d4791e1c5548b881e3fa5fa17173cc9fd54a039041"},"source":{"id":"2605.12513","kind":"arxiv","version":1},"verdict":{"id":"fe9e45d0-561b-46ea-8720-3e4253c7994a","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T21:06:38.102110Z","strongest_claim":"Experiments on multiple real-world networks show that SP-GCRL achieves significant gains over heuristic and learning-based baselines across budgets and topologies, while maintaining strong large-scale scalability.","one_line_summary":"SP-GCRL combines a nonlinear social diffusion model, dual-view contrastive learning for robust node embeddings, a GAT surrogate, and DDQN to learn end-to-end seed selection policies for influence maximization under partial graph observability.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The proposed social-propagation-aware nonlinear diffusion function correctly models reinforcement, diminishing returns, and probability drift under repeated exposure in real incomplete graphs.","pith_extraction_headline":"SP-GCRL learns end-to-end seed selection policies for influence maximization on incomplete social graphs using contrastive representations and a nonlinear diffusion model."},"references":{"count":36,"sample":[{"doi":"","year":2024,"title":"Science advances10(15), eadh4439 (2024)","work_id":"eb6d3968-250d-46f1-a0dc-b0b06de81c4a","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2018,"title":"In: 2018 IEEE/WIC/ACM Interna- tional Conference on Web Intelligence (WI)","work_id":"77cfe480-4c45-481a-9e6b-3fdf66e5b490","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2018,"title":"Proceedings of the National Academy of Sciences115(37), 9216–9221 (2018)","work_id":"4a291fda-160e-4e6e-a795-6846f02118ab","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2021,"title":"In: Uncertainty in Artificial In- telligence","work_id":"2ba968f8-4fc7-4978-b64f-1a2eb7e957fe","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"IEEE Transactions on Compu- tational Social Systems11(2), 2210–2221 (2023)","work_id":"f47ebf0c-63de-4345-9de4-fb6090ecffec","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":36,"snapshot_sha256":"a8e71eff1121b08217523c3ab84acbad2df7deab5dd9e8254ed670e82f055f89","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"5e5ff1114b1ebce4c20c4a31daa2e76cef175d0ea7de765e075de0f62a5c67bd"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}