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..
Opinion Maximization in Social Networks by Modifying Internal Opinions
Authors: Gengyu Wang, Runze Zhang, Zhongzhi Zhang
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments on real-world datasets demonstrate that our methods outperform baseline approaches. Notably, our asynchronous algorithm delivers exceptional efficiency and accuracy across all scenarios, even in networks with tens of millions of nodes. |
| Researcher Affiliation | Academia | Gengyu Wang, Runze Zhang, Zhongzhi Zhang College of Computer Science and Artificial Intelligence Fudan University EMAIL, EMAIL |
| Pseudocode | Yes | Algorithm 1: GLOBALINFAPPROX(G, R, s, ϵ) Algorithm 2: TARGETEDNODEREFINE(G, r a, r 0, ϵ) Algorithm 3: MAXINFLUENCESELECTOR(G, R, s, ϵ, k) Algorithm 4: RANDOMFOREST(G, R) Algorithm 5: Forest(G, R, s, l) |
| Open Source Code | Yes | Justification: The experimental code has been packaged into ZIP files and submitted as supplementary materials with the paper. Sufficient instructions are provided in the supplemental material to faithfully reproduce the main experimental results, including details on dataset preparation, parameter settings, and execution steps. |
| Open Datasets | Yes | We use 8 benchmark datasets that are obtained from the Koblenz Network Collection [45], SNAP [46] and Network Repository [47]. Table 1 summarizes the key characteristics of the networks used in our experiments, including network name, number of nodes, number of edges, maximum out-degree, and network type. |
| Dataset Splits | No | The paper does not explicitly provide training/test/validation dataset splits. It describes evaluating algorithms on full datasets by varying a parameter 'k' for node selection, which is not equivalent to standard dataset splits for model training and evaluation. |
| Hardware Specification | Yes | Our extensive experiments were conducted on a Linux server equipped with 28-core 2.0GHz Intel(R) Xeon(R) Gold 6330 CPU and 1TB of main memory. |
| Software Dependencies | Yes | All the algorithms we proposed are implemented in Julia v1.10.7 using single-threaded execution. |
| Experiment Setup | Yes | In all experiments, we set the absolute error parameters of RWB to 10 2. For algorithm FOREST, we set the number of samplings l = 4000. We set the initial error parameter of algorithm MIS to 10 3. To further demonstrate the effectiveness of our approach, we compare against five widely-used benchmark algorithms: TOPRANDOM, TOPDEGREE, TOPCLOSENESS, TOPBETWEENNESS, and TOPPAGERANK [49] across all networks. |