Low-Rank Linear Cold-Start Recommendation from Social Data

Authors: Suvash Sedhain, Aditya Menon, Scott Sanner, Lexing Xie, Darius Braziunas

AAAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on four realworld datasets show that Lo Co yields significant improvements over state-of-the-art cold-start recommenders that exploit high-dimensional social network metadata.
Researcher Affiliation Collaboration Suvash Sedhain, Aditya Krishna Menon, Scott Sanner, Lexing Xie, Darius Braziunas S { Australian National University, Data61}, Canberra, ACT, Australia University of Toronto, Toronto, Canada S Rakuten Kobo Inc., Toronto, Canada
Pseudocode No The paper describes methods mathematically but 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 create 10 temporal train-test splits for the Ebook dataset. We create 10 train-test folds on the other datasets by including random 10% of the users in the test set and remaining 90% users in the training set.
Dataset Splits Yes We used cross-validation with grid-search to tune all hyperparameters.
Hardware Specification No The paper does not provide specific hardware details used for running its experiments.
Software Dependencies No The paper describes algorithms used (e.g., randomized SVD) but does not provide specific ancillary software details like library names with version numbers.
Experiment Setup Yes We used cross-validation with grid-search to tune all hyperparameters. For the latent factor methods, we tuned the latent dimension K from {5, 10, 50, 100, 500, 1000}. For the methods relying on ℓ2 regularisation, we tuned all regularisation strengths from {10-3, 10-2, ..., 103}.