The Crusader Kings series simulates relationship scores not just between countries but also between any relevant people in those countries, which can easily get to tens of thousands of characters. While they did quite a lot of historic research, they of course did not go so far as to find out what the Duke of Wessex thought about some minor noble in India in 1066. They modeled the relationship system in a way which scales far better.
The gist is that each character received a bunch of character traits like "Zealous", "Cynic", "Brave", "Craven", "Arbitrary", "Just" etc. Some of them based on how historic evidence described well-known historic people, but most of them randomly assigned. Now the relationship between any two characters is simply a function taking their personality traits into account. "Zealous" vs. people of other religions: -20 opinion. "Brave" vs. "Craven": -10 opinion. "Both Just": +10 opinion, and so on. This allows to calculate a hypothetical relationship score between any two characters anywhere in the world, even if they would probably not even have known of each others existence.
Opinion modifiers between any two characters might also be affected by temporary modifiers resulting from interactions between those characters ("insulted me during a banquet: -20 opinion for 5 years") but those are rather the exception than the norm when it comes to quantifying the relationship between any two characters.
The advantage of this system is that this system scales linearly instead of quadratically. You still need to research and apply properties for
n entities, but you do not need to do the same for
n * n relationships.
The same system could be used for countries in your game:
- Think of some inherent properties countries could have which could affect their relationship with other countries: Political system, culture, language, values, etc.
- Make up plusses and minuses for opinions based on those properties. Like "Democratic" vs. "Autocratic": -20. "Same official language": +10. "Both socially progressive": +20.
- Think of any other game systems you have which might feed into the opinion system. For example, when your game simulates exports and imports (a very important factor for real-world political relationships), then you might add opinion penalties for countries which are competitors for the same product or opinion bonuses for countries which trade a lot with each other.
A word of warning, though: When you are trying to model the world of 2022, then you are stepping into a minefield of controversy. Your data and the simulation resulting from it will necessary contain a couple things with the potential to offend a lot of people. So make sure that it is clear that your product is supposed to be first and foremost a game and not meant to be a political statement. But even then you can expect that the PR and community management of your game will become a pretty nerve-wracking job. Oh, and if your game is sufficiently successful, then it will probably get banned in a couple countries for contradicting the local government's narrative of world politics. For example: Good luck modeling the middle-east in a way that is not going to offend the censors of any authoritarian government in the area.