The code has been tested running under Python 2.7. The required packages are as follows:
- tensorflow-gpu == 1.14.0
- keras == 2.2.4
- keras_self_attention == 0.49.0
https://pypi.org/project/keras-self-attention/ - keras_pos_embd == 0.8.0
https://pypi.org/project/keras-pos-embd/0.8.0/
-
/model
- CDAPM.py #The implementation of CDAPM
- EmbeddingMatrix.py #The Embedding for text-embedding for a 3-d vector to a 2-d vector
- mode_normalization.py *# The implementation of Mode Normalization *
-
/data # Path of New York Foursquare and Los Angelos Geo-Tweets datasets
- la_tweets1100000_cikm.txt # Los Angelos Geo-Tweets
- tweets-cikm.txt # New York Foursquare check-ins
- venues.txt # New York POIs
-
/word_vec
- glove.twitter.27B.50d.txt # Pretrained Glove word embeddings
-
/features # The processed trajectory samples and features by geo_data_decoder.py
- features&index_seg_gride_fs # New York
- features&index_seg_gride_la # Los Angelos
-
config.py # It contains all the configurations in CDAPM
-
eval_tools.py # It contains helper functions and evaluation tools
-
geo_data_decoder.py # Preprocessing for both New York Foursquare and Los Angelos Geo-Tweets
-
read_features.py # Get the features and samples that model need without running geo_data_decoder.py
-
train.py # Training procedure of CDAPM. This is the entrace of CDAPM model.
we provide the processed features --/features, so we don't need the original dataset you can train and evaluate model :
python train.py
model | code |
---|---|
MF | The implementation of author |
HMM | The implementation of author |
ST-RNN | The implementation of author |
PoVDSSA | The implementation of author |
SERM | CDAPM.py-serm |
DeepMove | The implementation of author |
CDAPM | CDAPM.py-cdapm |
Some large files are not available in this repository.Please download on link: https://pan.baidu.com/s/1YsonkFBk2pN-PUJOC-Bbyg password: ki82
- /data # Path of New York Foursquare and Los Angelos Geo-Tweets datasets
- la_tweets1100000_cikm.txt # Los Angelos Geo-Tweets
- tweets-cikm.txt # New York Foursquare check-ins
- venues.txt # New York POIs
- /word_vec
- glove.twitter.27B.50d.txt # Pretrained Glove word embeddings