Joint Modeling of User Behaviors Based on Variable-Order Additive Markov Chain for POI Recommendation

Table 1

Performance of SONG_N and SONG on four datasets.

Data	Metric	SONG_N	SONG	Inc
BJ	Rec@1	0.2049	0.2327	14%
	Rec@5	0.3208	0.3662	14%
	Rec@10	0.3577	0.4055	13%
	Rec@15	0.3710	0.4270	15%
SH	Rec@1	0.1488	0.1581	6%
	Rec@5	0.2622	0.2800	7%
	Rec@10	0.3094	0.3297	7%
	Rec@15	0.3291	0.3555	8%
NYC	Rec@1	0.2337	0.2740	17%
	Rec@5	0.4361	0.5042	16%
	Rec@10	0.4800	0.5523	15%
	Rec@15	0.4909	0.5639	15%
TKY	Rec@1	0.1769	0.1994	13%
	Rec@5	0.3364	0.3724	11%
	Rec@10	0.3909	0.4318	10%
	Rec@15	0.4192	0.4623	10%
BJ	NDCG@1	0.3256	0.3564	9%
	NDCG@5	0.1341	0.1440	7%
	NDCG@10	0.0959	0.1030	7%
	NDCG@15	0.0771	0.0832	8%
SH	NDCG@1	0.2978	0.3084	4%
	NDCG@5	0.1522	0.1580	4%
	NDCG@10	0.1170	0.1209	3%
	NDCG@15	0.0972	0.1008	4%
NYC	NDCG@1	0.3580	0.4056	13%
	NDCG@5	0.1785	0.1973	11%
	NDCG@10	0.1280	0.1407	10%
	NDCG@15	0.1024	0.1124	10%
TKY	NDCG@1	0.3248	0.3506	8%
	NDCG@5	0.1649	0.1765	7%
	NDCG@10	0.1199	0.1283	7%
	NDCG@15	0.0977	0.1044	7%