Calibration Fairness
- Downloading datasets raw file (
datasets/DS_NAME/raw
) - Collecting required features and mapping IDs to range
0
-N
(the number of users or items). To do this we provide each dataset a specific notebook (datasets/DS_NAME/DS_Name_dataset.ipynb
)- Output: Here we create two file, one for the
rating
data which show a user's rating on an item, and another one iscat
file. Thecat
file indicate the category of each item (datasets/DS_NAME/DS_NAME_data_map.txt
anddatasets/DS_NAME/DS_NAME_cat_map.txt
).
- Output: Here we create two file, one for the
dataset.ipynb
--->datasets/DS_NAME/DS_NAME_data.txt
anddatasets/DS_NAME/DS_NAME_cat.txt
GoogleDrive/0_dataset_in_use.ipnb
--->Train
,Test
,Category
,Inters
- User Grouping (
user-groups
): 5% - Item Grouping (
item-groups
): 20%
- ClothingFit: 5-core
- MovieLens1M: 10-core