crystal22 / apoir

Adversarial Point-of-Interest Recommendation

Adversarial Point-of-Interest Recommendation (APOIR)

This is the python implementation -- a POI recommendation model using adversarial training.

---

Requirements

• python2.7
• tensorflow >= 1.3.0
• numpy >= 1.14.0

---

Datasets

Datasets	Users	POIs	Check-ins
Gowalla	18,737	32,510	1,278,274
Foursquare	24,941	28,593	1,196,248
Yelp	30,887	18,995	860,888

For the Gowalla dataset, we filter out those users with fewer than 15 check-in POIs and those POIs with fewer than 10 visitors. For Foursquare and Yelp, we discard those users with fewer than 10 check-in POIs and those POIs with fewer than 10 visitors. We partition each dataset into training set and test set. For each user, we use the earlier 75% check-ins as the training data and the most recent 25% check-ins as the test data. All datasets are very sparse (the frequency of most POIs being visited is extremely low).

---

Baseline

• USG
• MGMPFM
• LFBCA
• iGSLR
• LORE
• IRenMF
• GeoMF
• RankGeoFM You can find all the above code at http://spatialkeyword.sce.ntu.edu.sg/eval-vldb17/
• GeoTeaser You can find this code at https://github.com/shenglin1987/geo_teaser
• PACE You can find this code at https://github.com/yangji9181/PACE2017

---

Usage

• To run APOIR, please first clone the code to your python IDE (eg:Pycharm), then run the command: python APOIR.py.

• To customize the code,

• • you can change the Embedding_size in line 9
• • You need to set your data set folder in line 14
• • Then change the user number and POI number to fit the given data set in line 10 and line 11
• • You can also set the learning_rate_value to be a different value in line 17

• You can use your own data sets in the source code