A Boosting Algorithm for Item Recommendation with Implicit Feedback

Authors: Yong Liu, Peilin Zhao, Aixin Sun, Chunyan Miao

IJCAI 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Ada BPR demonstrates its effectiveness on three datasets compared with strong baseline algorithms.
Researcher Affiliation Academia Yong Liu1,2, Peilin Zhao3, Aixin Sun2, Chunyan Miao1,2 1Joint NTU-UBC Research Centre of Excellence in Active Living for the Elderly (LILY), Nanyang Technological University, Singapore 2School of Computer Engineering, Nanyang Technological University, Singapore 3Institute for Infocomm Research, A*STAR, Singapore
Pseudocode Yes Algorithm 1: The Ada BPR Algorithm; Algorithm 2: Component Recommender Construction
Open Source Code No The paper does not provide an explicit statement or link to open-source code for the described methodology.
Open Datasets Yes The experiments are conducted on three public datasets: the Movie Lens-100K, Movie Lens-1M, and the Taste Profile Subset of the Million Song Dataset (TPMSD) [Bertin Mahieux et al., 2011].
Dataset Splits No The paper mentions validation data construction for parameter tuning but does not specify a distinct validation dataset split percentage or size as part of the main dataset splits.
Hardware Specification No The paper does not provide specific details about the hardware used for running the experiments.
Software Dependencies No The paper does not provide specific software dependencies or version numbers.
Experiment Setup Yes For Ada BPR, the dimensionality of the latent space d is selected from {10, 30, 50} (see Section 3.2), and for the other models, d is chosen from {10, 30, 50, 70, 90}. The regularization parameters are chosen from 10[ 6 : 1 : 0] (see Eq. 11), and the optimal learning rates are selected from 2[ 10 : 1 : 0] (see Alg. 2). For GBPR, the group size is chosen from {2, 3, 4, 5} and the parameter ρ is chosen from {0.2, 0.4, 0.6, 0.8, 1.0}. The number of component models in Ada MFexp, Ada MFimp, and Ada BPR are all set to 30. All experiments are repeated for 5 times, each with a different random seed. The results reported are average of the 5 runs.