Towards Fine-Grained Temporal Network Representation via Time-Reinforced Random Walk
Authors: Zhining Liu, Dawei Zhou, Yada Zhu, Jinjie Gu, Jingrui He4973-4980
AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In this section, we demonstrate the performance of our proposed Fi GTNE algorithm on quantitative evaluations and parameter sensitivity analysis. Quantitative results regarding node classification and link prediction are shown in Fig. 3 and Fig. 4, respectively. |
| Researcher Affiliation | Collaboration | Zhining Liu,1 Dawei Zhou,2 Yada Zhu,3 Jinjie Gu,4 Jingrui He2 1University of Electronic Science and Technology of China, 2University of Illinois at Urbana-Champaign, 3IBM T. J. Watson Research Center, 4Ant Financial Services Group |
| Pseudocode | Yes | Algorithm 1 Time-Reinforced Random Walk; Algorithm 2 Fine-Grained Temporal Network Embedding |
| Open Source Code | No | The paper provides links to datasets but no explicit statement or link indicating that the source code for the proposed FiGTNE methodology is publicly available. |
| Open Datasets | Yes | Taobao1 is a bipartite online shopping network... 1https://tianchi.aliyun.com/datalab/data Set.html?spm= 5176.100/073.0.0.4703ea7EOGsc Idata Id=649; Epinion2 is a consumer review network... 2https://www.cse.msu.edu/ tangjili/trust.html; Criteo3 is an action network... 3http://ailab.criteo.com/criteo-sponsored-search-conversionlog-dataset/ |
| Dataset Splits | No | The paper mentions "different sizes of training data" (e.g., 30%, 50%, 70%) for node classification, and specific ratios of edges for link prediction, but does not provide explicit train/validation/test dataset splits needed for full reproduction, nor does it explicitly mention a validation set. |
| Hardware Specification | No | The paper does not provide any specific hardware details such as CPU/GPU models, memory, or other computing specifications used for experiments. |
| Software Dependencies | No | The paper does not provide specific software dependencies with version numbers. |
| Experiment Setup | Yes | The dimension of the node embedding is set to 64. For Deep Walk, we tune hyperparameters: nw {10, 30, 50}, l {10, 30, 50} and w {3, 5, 7}. For node2vec, we test p {0.5, 1, 1.5} and q {0.5, 1, 1.5}. For TRRW, we study K {nr, 2nr, ..., 10nr} and Δ {1, 2, 4, ..., 128}. |