minkorange / mggcf

MGGCF

Multi-Graph Group Collaborative Filtering

This is our Tensorflow implementation for the paper:Multi-Graph Group Collaborative Filtering. We also provide two baselines code.

Introduction

---

Multi-Graph Group Collaborative Filtering is a new group recommendation method based on graph neural network.

Environment Requirement

---

The MGGCF code has been tested running under Python 3.6. The required packages are as follows:

• tensorflow == 1.8.0
• numpy == 1.14.3
• scipy == 1.1.0
• sklearn == 0.19.1

The AGREE and NeuMF baselines code are based on pytorch framework.

• pytorch version: '0.3.0'
• python version: '3.5'

Example to Run the Codes

Copy data file folder to the corresponding method file folder like MGGCF or AGREE. Then run AGREE or NeuMF:

python main.py

run MGGCF:

python MGGCF.py

For MGGCF method,

the value of HR and NDCG in the test dataset will be printed in command window after each optimization iteration.Like:

Epoch 0 [51.9s]: train==[285.65265=81.19891 + 204.45381]
Epoch 0 [51.9s + 216.7s]: train==[285.65265=81.19891 + 204.45381], g_hit=[0.92033], g_ndcg=[0.58268], u_hit=[0.53179], u_ndcg=[0.29446]
Epoch 1 [42.9s]: train==[190.82164=33.28221 + 157.53944]
Epoch 1 [42.9s + 205.2s]: train==[190.82164=33.28221 + 157.53944], g_hit=[0.93200], g_ndcg=[0.59525], u_hit=[0.56358], u_ndcg=[0.31497]

After trainging process, the best result will be printed in command window like this:

Best Iter=[8]@[7223.4]	g_hit=[0.98000], g_ndcg=[0.66206], u_hit=[0.68825], u_ndcg=[0.39949]

Parameter Tuning

we put all the parameters in the config.py for AGREE and NeuMF. and we put all the parameters in the parser.py for MGGCF.

Dataset

We provide two processed datasets: CAMRa2011 and Movielens-Simi. group(user)RatingTrain.:

• Train file.
• Each Line is a training instance: groupID(userID)\t itemID\t rating\t timestamp (if have)

group(user)RatingTest:

• group(user) Test file (positive instances).
• Each Line is a testing instance: groupID(userID)\t itemID\t rating\t timestamp (if have)

group(user)RatingNegative

• group(user) Test file (negative instances).
• Each line corresponds to the line of test.rating, containing 100 negative samples.
• Each line is in the format: (groupID(userID),itemID)\t negativeItemID1\t negativeItemID2 ...

groupMember

• group member file
• Each line is a group information: groupID\t userID1 \t userID2 ...