Adaptive User Modeling with Long and Short-Term Preference for Personalized Recommendation

This code provides an implementation of the SLi-Rec network for sequential recommendation in this paper.
To see more, please visit:
https://github.com/microsoft/recommenders/tree/master/reco_utils/recommender/deeprec/models/sequential .

Data Preparation

sh data_preparing.sh

In data/, it will generate these files:

• reviews_information
• meta_information
• train_data
• test_data
• user_vocab.pkl
• item_vocab.pkl
• category_vocab.pkl

Model Implementation

The model is implemented in model.py. Training model:

python sli_rec/train.py

After training, run the following code to evaluate the model:

python sli_rec/test.py

The model below had been supported:

Baselines:

• ASVD
• DIN
• LSTM
• LSTMPP
• NARM
• CARNN
• Time1LSTM
• Time2LSTM
• Time3LSTM
• DIEN

Our models:

• A2SVD
• T_SeqRec
• TC_SeqRec_I
• TC_SeqRec_G
• TC_SeqRec
• SLi_Rec_Fixed
• SLi_Rec_Adaptive

Dependencies (other versions may also work):

• python==2.7
• tensorflow==1.4.1
• keras==2.1.5
• numpy==1.15.4