On Multi-Relational Link Prediction With Bilinear Models

Authors: Yanjie Wang, Rainer Gemulla, Hui Li

AAAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We report results of an independent experimental study that evaluates recent bilinear models in a common experimental setup.
Researcher Affiliation Academia Yanjie Wang University of Mannheim, Germany ywang@uni-mannheim.de Rainer Gemulla University of Mannheim, Germany rgemulla@uni-mannheim.de Hui Li The University of Hong Kong hli2@cs.hku.hk
Pseudocode No The paper does not contain any explicit pseudocode blocks or algorithms.
Open Source Code Yes All datasets, experimental results, and source code will be made publicly available.4 http://dws.informatik.uni-mannheim.de/en/resources/software/tf/
Open Datasets Yes We used the WN18 (Bordes et al. 2014) and FB15K (Bordes et al. 2013) datasets, which were extracted from Word Net (Miller 1995) and Freebase (Bollacker et al. 2008), respectively.
Dataset Splits Yes The two datasets are presplit into a training set, a validation set, and a test set. Table 2 summarizes the key statistics.
Hardware Specification No The paper does not provide specific details about the hardware used for experiments, such as GPU/CPU models or memory.
Software Dependencies No The paper mentions C++, Intel Math Kernel Library, Adagrad, and LIBLINEAR, but does not provide specific version numbers for these software components.
Experiment Setup Yes We considered the following hyperparameter settings: r {100, 200}, learning rate η {0.01, 0.1, 1}, weight of L2-regularization λe, λr {0, 0.1, 0.01} for entity and relation parameters, resp., margin hyperparameter γ {1, 2, 4, 8} for RESCAL, γ {0.2, 0.5, 0.7} for Hol E, and γ {0.2, 0.5, 0.7, 1.0, 1.5} for Trans E. ... Tab. 3 reports the hyperparameters ultimately selected.