This is an implement of Collective Factor Model (CFM) for Multi-Criteria Recommender Systems. Predicting ratings in multi-criteria recommender systems via a collective factor model and Improving Rating Prediction in Multi-Criteria Recommender Systems Via a Collective Factor Model

• Clone this repo.
• enter the directory where you clone it, and run the following code

  pip install -r requirements.txt
  python -m CFM --method BMF

You can check out the other options available to use with CFM using:

python -m CFM --help

• -d, --dataset, The dataset name; the default is ta;
• --K, Number of latent vectors; the default is 50;
• --criteriaNum, Number of criteria ratings; the default is 6;
• --cv, ,The cv in datasets; the default is 10;
• --percent, The percent in trainning; the default is 02;
• -e, --epochs, The training epochs; the default is 1;
• --batchSize, The batch size of training; the default is 1;
• --lr, The learning rate; the default is 0.1;
• --learner, the optimazation algorithms; the default is adam;
• --maxR, The maximum rating of datasets; the default is 5.0;
• --minR, The minimum rating of datasets; the default is 1.0;
• --method, The learning method (BMF, CFM);
• --biasR, The regularization of biasd in BMF et. al; the default is 0.01;
• --uR, The regularization of users\ latent vector; the default is 0.01;
• --iR, The regularization of items\ latent vector; the default is 0.01;
• --regressionR, The regularization of regression weights; the default is 0.01;
• --reg, set regularization for all weights ; the default is 0.01;
• --lam, The effect of criteria rating in CFM et. al; the default is 0.01;
• --share, sharing users' or items' for CFM model (user ,item ,ind); the default is ind;
• --saveThreshold, The Threshold for saving model,the default is 0.89;
• --CPU, The numbers of CPU cores, the default is 1;
• --GPU, The numbers of GPU cores, the default is 0;

If you find CFM is useful for your research, please consider citing the following papers:

    @inproceedings{fan2021predicting,
    title={Predicting ratings in multi-criteria recommender systems via a collective factor model},
    author={Fan, Ge and Zhang, Chaoyun and Chen, Junyang and Wu, Kaishun},
    booktitle={DeMal@ The Web Conference},
    pages={1--6},
    year={2021}
    }
    
    @ARTICLE{fan2023improving,
      author={Fan, Ge and Zhang, Chaoyun and Chen, Junyang and Li, Paul and Li, Yingjie and Leung, Victor C. M.},
      journal={IEEE Transactions on Network Science and Engineering}, 
      title={Improving Rating Prediction in Multi-Criteria Recommender Systems Via a Collective Factor Model}, 
      year={2023},
      volume={},
      number={},
      pages={1-11},
      doi={10.1109/TNSE.2023.3270910}}