A hybrid network–fuzzy framework for modelling sustainable tourism development: latent structures from Delphi indicators

The elicitation of expert consensus in future studies often results in complex, high-dimensional datasets characterized by ordinal variables andnon-linear dependencies. Traditional hard clustering techniques frequentlyfail to capture the intrinsic ambiguity and the overlapping nature of futurescenarios derived from Delphi studies. This paper introduces a novel hybridmethodological framework that integrates correlation network analysis withfuzzy clustering to detect latent structures in ordinal Delphi data. We modelthe consensus space as a weighted graph where edge weights represent rankcorrelations between projections. A topological modularity maximization isfirst applied to identify the backbone of the community structure, followedby a Fuzzy C-Means algorithm on the topological feature space to assignmembership degrees to transitional variables. The methodology is appliedto a dataset of 57 items regarding the future of the tourism ecosystem in the Apulia region, evaluated by a panel of experts. Results reveal a tripartitelatent structure comprising structural assets, sustainability governance, andeconomic risks. The validity of the extracted partitions is confirmed via aMonte Carlo permutation test, demonstrating that the detected modularitysignificantly exceeds that of random networks. This approach offers a robustquantitative tool for policy planning, capable of distinguishing between coredrivers and ambiguous bridge factors in scenario building