Table 2:

Plausibility and completeness of the UMLS knowledge embeddings. _SN indicates the incorporation of the Semantic Networkin the embeddings, which otherwise were learned only from the Metathesaurus.

Model	UMLS Metathesaurus					UMLS Semantic Network
	RTC-P	RTC-R	PPA	H@10	MRR	RTC-P	RTC-R	PPA	H@10	MRR
TRANSE	0.7712	0.6479	0.9340	0.2161	0.1400	–	–	–	–	–
TRANSD	0.9080	0.8895	0.9734	0.2780	0.1674	–	–	–	–	–
TRANSE_SN	0.8649	0.8019	0.9746	0.2240	0.1425	0.5105	0.7790	0.9150	0.7125	0.4882
TRANSD_SN	0.9188	0.8915	0.9729	0.2775	0.1670	0.6840	0.8771	0.9017	0.7300	0.4680
GAN (TRANSE_SN+DISTMULT)	0.8959	0.8424	0.9833	0.2727	0.1650	0.6109	0.7898	0.9367	0.7883	0.5373
GAN (TRANSD_SN+DISTMULT)	0.9311	0.9130	0.9803	0.3164	0.1886	0.8419	0.8546	0.9200	0.7867	0.5236