Consider "Pain". There are 2 cliques:
EFO:0003843 and HP:0012531.
The EFO is returned first but the HP merges a bunch of stuff (EFO is by itself).
Should the clique size be used in scoring? all else being equal, I'd rather get the larger clique... And it might be a way to naturally prefer better-acting curies to a bunch of random UMLS dross.
Consider "Pain". There are 2 cliques: EFO:0003843 and HP:0012531.
The EFO is returned first but the HP merges a bunch of stuff (EFO is by itself).
Should the clique size be used in scoring? all else being equal, I'd rather get the larger clique... And it might be a way to naturally prefer better-acting curies to a bunch of random UMLS dross.