A Static-Dynamic Hypergraph Neural Network Framework Based on Residual Learning for Stock Recommendation

Algorithm 1

Dynamic hypergraph construction.

	Input: Input embedding X; Nearest neighbor parameter k
	Output: Hyperedge set E; The set of vertices possessed by a hyperedge e pos(e)
	Function: K-nearest neighborhood selection knn; Distance function dis; Smallest distance index selection topK
(1)	for u in range (len (X)) do
(2)	D = dis (X (u), X)
(3)	D = sort (D)
(4)	ind = topK (D, 2)-topK (D, 1)
(5)	E. insert (ind)
(6)	pos (ind). insert (u)
(7)	end for
(8)	for e in E do
(9)	= knn (X[e], X, k)
(10)	pos (e). insert ()
(11)	end for