## Relative aggression index - what's the best way to compose it?

I study the international conflicts, and there is a dataset where all the official actions of one state towards another are registered. They are of two types: cooperation and conflict. The sum of two types gives a total amount of actions of country A towards country B. For example (randomly picked countries with randomly assigned numbers):

Which means that Belgium made 3 hostile actions to Congo (out of 8 in total), and Congo made 4 actions towards Belgium (out of 9).
What I am looking for is an index that would measure relative aggression of a country.
There are two issues.
1) It should be weighted by total amount of actions of A-to-B.
2) It should take into account a reciprocity of A-to-B and B-to-A actions. In the real data, the correlation between these two is 0.9.
What I thought up right now is something like this. If we call matrix of conflict C, and matrix of total events T, then relative asymmetry of conflictedness (sorry for an ugly word) of a country i towards country j is:
A_i=C_i_j/T_i_j – C_j_i/T_j_i
So, for instance for Belgium towards Congo A=3/8-4/9 = -0.069, which means that Belgium (in this imaginative case) is proportional 7% less hostile to Congo than vice versa.
In your opinion, does it all make sense and if not, give me a hint what should be done?
Thank you!