Source code of the paper "An Exploratory Study on Information Cocoon in Recommender Systems".
-
config.py -- basic configuration: dataset, model, cuda, etc.
-
rec_simulation.py -- interaction simulation between user and news recommender system.
Before running 'rec_simulation.py', the recommendation model needs to be trained in advance.
- train_prob_predict.py -- train news recommendation model: NRMS, NAML, DKN.
- train_fm_and_ncf.py -- train DeepFM and NCF models.
- train_ngcf.py -- train NGCF model.
And the model implementation code is in 'model/' directory.
You can get the preprocessed data through Baidu Cloud Disk:
Link: https://pan.baidu.com/s/1tsW6CqFbG8OMYT1aTHhEQQ
Extraction code: kj7a
recommendation models refer to the following: