Point-of-Interest Recommendation Model Based on Graph Convolutional Neural Network

Algorithm 1

Model Training.

Input User check-in set , regularization coefficient , user-POI graph , time-POI graph , geographical threshold

Output Model parameters

1. Initialize model parameters;

2. Calculate the feature vector according to Equation (3);

3. Calculate the geographical factor coefficient according to Equation (4);

4. Calculate the user preference scores for the POI according to Equation (6);

5. Minimize Equation (7) and optimize model parameters with a stochastic gradient descent algorithm, repeating steps 2, 3, 4, and 5 before model convergence.