Implementation and dataset of the paper
Szubartowicz, Michael, "GroupFM: Enabling Context-Aware Group Recommendations with Factorization Machines" (2021). ICIS 2021 Proceedings. 5. https://aisel.aisnet.org/icis2021/data_analytics/data_analytics/5
The dataset consists of four csv files found in the subfolder GroupFM.
group_assignments.csv lists which group each user belongs to.
group_types.csv denotes, whether the group members are familiar or collected by random selection.
group_ratings.csv contains the base ratings as well as the indicated impact of each contextual variable for each group.
user_ratings.csv contains the base ratings as well as the indicated impact of each contextual variable for each user.
GroupFM is realized by a python script which is running the original libFM implementation in the background. Therefore you need to follow these steps:
- compile libFM (Windows compilation with cygwin is supported; see libFM documentation and forum)
- depending on your platform the location string of libFM might be corrected (variable
libFM_location) - run the python script
Exec_GroupFM.py