Table2Graph: Transforming Tabular Data to Unified Weighted Graph

Authors: Kaixiong Zhou, Zirui Liu, Rui Chen, Li Li, Soo-Hyun Choi, Xia Hu

IJCAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The experimental results in a variety of real-world applications demonstrate the effectiveness and efficiency of our Table2Graph, in terms of the prediction accuracy and feature interaction detection.
Researcher Affiliation Collaboration Kaixiong Zhou1 , Zirui Liu1 , Rui Chen2 , Li Li2 , Soo-Hyun Choi3 and Xia Hu1 1Department of Computer Science, Rice University 2Samsung Research America 3Samsung Electronics
Pseudocode No The paper does not contain structured pseudocode or algorithm blocks.
Open Source Code No The paper does not provide concrete access to source code for the methodology described.
Open Datasets Yes we adopt two benchmark datasets from the financial fraud detection and online advertising: Creditcard [Dal Pozzolo et al., 2015] and Criteo1. ... For the large graph modeling, we adopt Movie Lens dataset commonly evaluated in the previous collaborative filtering work [He et al., 2017].
Dataset Splits Yes Following the common evaluation process, we use 10-fold cross validation and judge models with metrics AUC (Area Under the ROC curve) and Log Loss for real-world datasets Creditcard and Criteo.
Hardware Specification No The paper does not provide specific hardware details (e.g., exact GPU/CPU models, processor types) used for running its experiments.
Software Dependencies No The paper does not provide specific ancillary software details with version numbers (e.g., library or solver names with version numbers) needed to replicate the experiment.
Experiment Setup Yes The hyperparameters of λ1, λ2 and λ3 are determined based on the grid search, and their influences are empirically studied in the following experiments. ... We adopt two popular item-to-item collaborative filtering methods as the underlying recommendation frameworks, namely factored item similarity model (FISM) [Kabbur et al., 2013] and neural attentive item similarity model (NAIS) [He et al., 2018]. ... A three-layer GNN model is used to learn the feature interactions based upon the generated graph.