Research on Recommendation Algorithm of Joint Light Graph Convolution Network and DropEdge

Table 2

Performance comparison of LG-DropEdge and LightGCN.

Dataset	Method	Precision@20	Recall@20	ndcg@20
Gowalla	LightGCN	0.0558	0.1821	0.1545
	LG-DropEdge	0.0575 (+3.04%)	0.1867 (+2.53%)	0.1582 (+2.39%)
Yelp2018	LightGCN	0.0283	0.0632	0.0520
	LG-DropEdge	0.0301 (+6.36%)	0.0671 (+6.17%)	0.0549 (+5.58%)
Amazon-book	LightGCN	0.0171	0.0415	0.0321
	LG-DropEdge	0.0180 (+5.26%)	0.0435 (+4.82%)	0.0336 (+4.67%)

The bold values represent the experimental results of the algorithm proposed in this paper and the improvement effect on the classical experiment.