Semantic Data Representation for Improving Tensor Factorization

Authors: Makoto Nakatsuji, Yasuhiro Fujiwara, Hiroyuki Toda, Hiroshi Sawada, Jin Zheng, James Hendler

AAAI 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments show that SRTF achieves up to 10% higher accuracy than state-of-the-art methods.
Researcher Affiliation Collaboration Makoto Nakatsuji1, Yasuhiro Fujiwara2, Hiroyuki Toda3, Hiroshi Sawada4, Jin Zheng5, James A. Hendler6 1,3,4NTT Service Evolution Laboratories, NTT Corporation, 1-1 Hikarinooka, Yokosuka-Shi, Kanagawa 239-0847 Japan 2NTT Software Innovation Center, NTT Corporation, 3-9-11 Midori-Cho, Musashino-Shi, Tokyo 180-8585 Japan 5,6Tetherless World Constellation, Rensselaer Polytechnic Institute, 110 8th Street, Troy, NY 12180-3590 USA
Pseudocode No No clearly labeled pseudocode or algorithm blocks were found. The paper describes the MCMC procedure in numbered prose steps.
Open Source Code No The paper does not provide concrete access to source code for the methodology described.
Open Datasets Yes Available at http://www.grouplens.org/node/73
Dataset Splits Yes We divided the dataset into three parts and performed three-fold cross validation.
Hardware Specification No Since GCTF requires much more computation than BPTF or SRTF (see Preliminary), we could not apply GCTF to the whole evaluation dataset on our computer.
Software Dependencies No The paper mentions tools like 'Stanford-parser' but does not specify version numbers for any software libraries or dependencies used in the implementation.
Experiment Setup Yes Following (Xiong et al. 2010), the parameters are µ0=0, ν0=0, β0=0, W0=I, ν0=1, and W0=1. L is 500.