Network Creation with Homophilic Agents
Authors: Martin Bullinger, Pascal Lenzner, Anna Melnichenko
IJCAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Despite their different initial conditions, both our theoretical and experimental analysis show that both the composition and segregation strength of the resulting stable networks are almost identical, indicating a robust structure of social networks under homophily. |
| Researcher Affiliation | Academia | Martin Bullinger1 , Pascal Lenzner2 and Anna Melnichenko2 1 Technical University of Munich 2 Hasso Plattner Institute, University of Potsdam bullinge@in.tum.de, {pascal.lenzner, anna.melnichenko}@hpi.de |
| Pseudocode | No | The paper does not contain any structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | The source code for the experiments can be found at https://github.com/melnan/Homophilic NCG |
| Open Datasets | No | The paper describes simulation experiments that start with 'sparse initial networks (spanning tree or grid)' rather than using a pre-existing, publicly available dataset with concrete access information. |
| Dataset Splits | No | The paper describes a simulation-based experimental setup and does not mention traditional train/validation/test dataset splits. |
| Hardware Specification | No | The paper mentions 'detailed simulations' but does not specify any hardware details like GPU models, CPU types, or memory used for running the experiments. |
| Software Dependencies | No | The paper does not provide specific version numbers for any software dependencies or libraries used in the experimental setup. |
| Experiment Setup | Yes | To this end, we simulate a simple dynamic process based on distributed and strategic edge creation and deletion over time, incentivized by optimizing the cost functions of our two models. The dynamics start with sparse initial networks (spanning tree or grid) and distribute agents of two equally-sized types such that the segregation of the initial network is very low or high. In each step, one agent is activated uniformly at random and can either create or delete an edge, performing a best response with respect to the cost function under consideration. The dynamics proceed until the consideration of no agent changes the network. |