This is the code for the Interactive Multi-Task Learning (iMTL) Framework for Next POI Recommendation with Uncertain Check-ins. iMTL exploits the interplay between activity and location preference through the temporal-aware activity encoder and spatial-aware location preference encoder.
- python 3.6.5
- numpy 1.16.2
- pandas 1.0.1
- progressbar 2.5
- tensorflow 1.13.1
- tensorflow-estimator 1.13.0
- Foursquare: Dingqi Yang, Daqing Zhang, and Bingqing Qu. Participatory cultural mapping based on collective behavior data in location-based social networks.
- Yelp: www.yelp.com/dataset/challenge
This module is in charge of preprocessing the raw data to formulate iMLT framework inputs. The processed data would be the five sequences recording information about time, POI, category, type and distance.
This module is the main body of the training and testing model. The model consists of three tasks:
- category prediction
- type prediction
- POI prediction
This module is a library providing useful functions for the previous two modules.