High levels of uncertainty faced by decision makers can be alleviated by characterizing multiple possible ways in which the future might unfold with scenario narratives. Aiming at describing alternative plausible chains of outcomes of key uncertainty factors, scenario narratives are often associated with graphical networks describing the relationships between the outcomes of the factors. We present a participatory framework for bottom-up development of such networks, the PACNAP (PArticipatory development of Consensual narratives through Network Aggregation and Pruning) framework. In this framework, relationships of influence between factor outcomes are judged by a group of scenario process participants. We develop an optimization model for pruning an aggregated graph based on these judgments. The model selects those edges of the aggregate graph that the participants most agree upon and can be tailored to identify compact graphs of varying degrees of cyclicity. As a result, a variety of graphical representations of varying structural richness can be explored to arrive at a succinct representation of a consensus view on the structure of a joint narrative. To this end, the main formal results are the representation of the participants’ agreement lexicographically in a linear objective function of a 0-1 program, and the translation of the requisites of the compactness and cyclicity of the resulting pruned graphs into a set of network flow constraints. The problem of identifying a consensus graphical representation is a general one and our graph pruning method has application potential outside the specific domain of narrative development as well.

Publisher Copyright: © 2025 The Authors

Peer reviewed