Table 5.

precision@K scores for network reconstruction on the 7 reference multiplex networks, for the network embedding methods combined with different embeddings operators (Hadamard, Weighted-L1, Weighted-L2, Average, and Cosine).

Operators	Method	CKM (95%)	LAZEGA (95%)	C.ELE (95%)	ARXIV (5%)	DIS (2,5%)	HOMO (2,5%)	MOL (2.5%)
Hadamard	node2vec-av	0.6764	0.9174	0.4526	0.8207	0.5578	0.7599	0.2989
	deepwalk-av	0.6564	0.9351	0.4416	0.7886	0.5486	0.7636	0.3164
	LINE-av	1.0	0.9924	0.8924	0.8204	0.4955	0.5191	0.4006
	Ohmnet	0.7842	0.8334	0.5329	0.9156	0.4811	0.6979	0.2591
	MNE	0.9505	0.9094	0.2728	0.7891	0.4218	0.3641	0.1316
	Multi-node2vec	0.8352	0.8811	0.6875	0.8605	0.6063	0.7584	0.3123
	MultiVERSE	0.9687	0.9695	0.7436	0.9015	0.6734	0.8729	0.3674
Weighted-L1	node2vec-av	0.5923	0.9494	0.5129	0.6922	0.5859	0.8123	0.3194
	deepwalk-av	0.5791	0.9784	0.4896	0.6878	0.5921	0.7984	0.3206
	LINE-av	0.9985	0.9953	0.9229	0.7837	0.4921	0.6839	0.3586
	Ohmnet	0.7355	0.8581	0.5785	0.8771	0.6025	0.8019	0.3769
	MNE	0.9926	0.975	0.4722	0.8593	0.4377	0.5241	0.1861
	Multi-node2vec	0.8636	0.9235	0.7379	0.7684	0.6356	0.7649	0.2671
	MultiVERSE	0.8545	0.9638	0.7444	0.8705	0.6678	0.7913	0.3559
Weighted-L2	node2vec-av	0.5886	0.9436	0.5097	0.6983	0.5953	0.8193	0.352
	deepwalk-av	0.5829	0.9672	0.5146	0.6877	0.5857	0.805	0.3233
	LINE-av	1.0	0.9962	0.9367	0.7749	0.4945	0.6697	0.392
	Ohmnet	0.7418	0.8687	0.5724	0.8694	0.6209	0.8143	0.3701
	MNE	0.9926	0.9764	0.4646	0.8818	0.4351	0.5529	0.176
	Multi-node2vec	0.8644	0.93	0.7568	0.7548	0.6361	0.7896	0.2922
	MultiVERSE	0.8653	0.969	0.754	0.8776	0.6784	0.7876	0.3701
Average	node2vec-av	0.8408	0.917	0.4817	0.889	0.5587	0.6809	0.2686
	deepwalk-av	0.8331	0.9379	0.501	0.8853	0.5318	0.6714	0.2795
	LINE-av	0.9855	0.9382	0.7103	0.8725	0.5093	0.5677	0.3244
	Ohmnet	0.9412	0.8287	0.5825	0.906	0.4989	0.6551	0.2887
	MNE	0.9179	0.9151	0.2966	0.7146	0.4175	0.352	0.1444
	Multi-node2vec	0.9767	0.8937	0.6726	0.9498	0.6243	0.6216	0.2901
	MultiVERSE	0.978	0.9059	0.5326	0.9758	0.6316	0.7204	0.4143
Cosine	node2vec-av	0.5103	0.4936	0.18	0.2537	0.1825	0.116	0.0441
	deepwalk-av	0.4807	0.4776	0.1741	0.2835	0.1854	0.1036	0.0462
	LINE-av	0.3291	0.4974	0.1867	0.2638	0.2384	0.1476	0.0454
	Ohmnet	0.5696	0.509	0.1718	0.2655	0.1984	0.1311	0.044
	MNE	0.3169	0.4536	0.1768	0.2445	0.1957	0.1667	0.044
	Multi-node2vec	0.5127	0.52	0.186	0.273	0.195	0.1032	0.0461
	MultiVERSE	0.6395	0.5026	0.1818	0.254	0.1983	0.1522	0.0474

For each multiplex network, the best score is in bold; for each operator, the best score is underlined. The percentage of edges used for the reconstruction is indicated in parenthesis under the name of the network. In the case of large networks (DIS, ARXIV, HOMO and MOL) MultiVERSE achieves the best performance in combination with different operators.