Social network analysis is a popular field of research and most commonly supported by graph theory-based methods. But there is an up-and-coming approach that has spilled over from the area of artificial intelligence – so-called entropy-driven social network analysis. This approach uses probabilistic conditionals to express relationships between actors or groups of actors rather than merely edges or arrows. The new method allows for calculating all actors’ importance in the net. However, if an analyst imprudently assigns probabilities to conditionals in the social network, the whole structure can be inconsistent, and hence the above-mentioned indices then are meaningless. Therefore, we propose a new interactive algorithm that helps the analyst to detect and revise such inconsistencies. The applicability of the algorithm will be exemplified by a mid-sized family network.