To answer my question about the prediction and determination of conflicts, we need a clear definition of the term. As far as I know is the term conflict often mistaken with the term loss of separation.
Let me start with the definition of conflict:
Converging of aircraft in space and time which constitutes a predicted violation of a given set of separation minima. (emphasis mine)
Whereas the definition of a loss of separation is:
A defined loss of separation between airborne aircraft occurs whenever specified separation minima in Controlled Airspace are breached. Minimum separation standards for airspace are specified by ATS authorities, based on ICAO standards. (emphasis mine)
To summarize the matter, a conflict between two aircraft occurs, not when there is a loss of separation, but when there is an impending loss of separation.
Seems pretty straightforward, right?
The challenge is that little word predicted. Who predicts the converging, when and by which measurements? When does a conflict occur in the eyes of an ATC-controller or a network manager? A related question is At what distance between two aircraft will the air traffic controller get an alarm?
I think this is a rather gray area, where probabilities come into play.