Robust Negative Sampling for Network Embedding
Authors: Mohammadreza Armandpour, Patrick Ding, Jianhua Huang, Xia Hu3191-3198
AAAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | R-NS is scalable to large-scale networks, and we empirically demonstrate the superiority of R-NS over NS for multi-label classification on a variety of real-world networks including social networks and language networks. |
| Researcher Affiliation | Academia | 1Department of Statistics, Texas A&M University 2Department of Computer Science and Engineering, Texas A&M University {armand, patrickding, jianhua}@stat.tamu.edu, hu@cse.tamu.edu |
| Pseudocode | No | The paper does not contain any structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | Source code for running experiments and data are available online at https://github.com/delimited0/network embedding. |
| Open Datasets | Yes | The Blog Catalog dataset (Zafarani and Liu 2009) is a friendship network of bloggers on the Blog Catalog website. The Flickr dataset (Huang, Li, and Hu 2017a) is a network of interactions between Flickr users. The Wikipedia dataset (Mahoney 2011) is the word co-occurrence network of the Wikipedia dataset, which used a 2-word window to determine co-occurrence edges. |
| Dataset Splits | No | The paper mentions 'randomly divide the data into train and test splits' and varying the 'proportion of training examples', but does not explicitly describe a separate validation split. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used for running the experiments (e.g., GPU models, CPU types, or memory). |
| Software Dependencies | No | The paper mentions using the 'Lib Linear library' but does not specify its version number or any other software dependencies with their versions. |
| Experiment Setup | Yes | The penalty coefficient λ is calculated based on the formula given in the paper. The degree power β is chosen to be 3/4, which is a widespread default in the literature. Exact settings are presented in the supplementary file. |