Augmented Intention Model for Next-Location Prediction from Graphical Trajectory Context

Algorithm 2

Location prediction algorithm.

Input: trajectory: , candidate nodes: , window size:
Output: trained model
(1)	for each user do
(2)	for each traj. do
(3)	strategy 1: random selection
(4)	strategy 2: based on angle
(5)	strategy 3: based on conditional probability
(6)	//Use one of the above strategies applied on each traj.
(7)	Initialize: parameters θ
(8)	for each batch nodes do
(9)
(10)	Neural Network (GCN ({Points: , Adj_Points: }))
(11)	end for
(12)	end for
(13)	end for