Matching many records by score?

[vavra7]

Hello, I work on project aiming to aggregate data of PC games from multiple sources. Those can be for our example steam, gog and metacritic. The process includes step of matching corresponding games obtained from given sources.

For that matter I use algorithm that calculates score of mutual match. Basically input are two games and output is number representing probability of match. Using this technique I am able to calculate score of mutual match for each with each record of two sources and gain required pairs.

The problem is that goal is not to have match of two sources but multiple sources. For illustration I have data like below representing {src:gameId} with scores.

steam - gog scores

	steam:1	steam:2	steam:3
gog:1	0.1	0.2	0.1
gog:2	0.6	0.1	0.8
gog:3	0.9	0.4	0.1

gog - metacritic scores

	gog:1	gog:2	gog:3
metacritic:1	0.7	0.2	0.4
metacritic:2	0.6	0.8	0.3
metacritic:3	0.1	0.4	0.1

metacritic - steam scores

	metacritic:1	metacritic:2	metacritic:3
steam:1	0.6	0.2	0.1
steam:2	0.7	0.4	0.9
steam:3	0.1	0.4	0.5

Expected output is matched gog - metacritic - steam game in case it exists. Or only gog - steam in case it exists and metacritic is missing. Etc...

I am in search for a solution of this problem when I hit into this library. Could you please put some light in if this library offers any tools to deal with my problem and point me the right direction? Thank you for any help.