Fuzzy Ontology Reasoning over Sustainable Development Data
TL;DR
This work presents a fully reproducible reasoning pipeline, including slice extraction, OWL 2-to-fuzzyDL translation, and mixed-integer linear programming (MILP) inference using Gurobi, that provides an expressive, declarative, and reproducible framework for analysing sustainable-development indicators while preserving the semantics of linguistic concepts.
Abstract
Sustainable-development indicators are inherently graded: a territory is not simply poor or resilient, but exhibits these characteristics to varying degrees. Consequently, policy questions are naturally expressed through vague concepts such as high unemployment, stressed neighbourhood, or green-transition readiness. This work shows how such questions can be answered through reasoning over a fuzzy Web Ontology Language (OWL 2) ontology, rather than by ad hoc data-processing scripts. Our case study is the ontology of the Sustainability Decision Framework (SDF) Project, a five-module fuzzy ontology integrating Sustainable Development Goal (SDG) statistics for the United Nations Economic Commission for Europe (UNECE) region with geopolitical knowledge and a fuzzy semantic layer comprising 35 linguistic datatypes and 21 composite concepts defined through weighted sums and ordered weighted averaging (OWA). The ontology contains almost 400,000 statistical observations, and its complete export into a fuzzyDL Knowledge Base comprises 2.96 million axioms; for this reason, reasoning is performed on the slices of the ontology that are relevant to each query. Reasoning is performed using fuzzy-dl-owl2, our Python re-engineering of the fuzzyDL reasoner and the Fuzzy OWL 2 framework. We present a fully reproducible reasoning pipeline, including slice extraction, OWL 2-to-fuzzyDL translation, and mixed-integer linear programming (MILP) inference using Gurobi. We demonstrate the approach through five applications covering fuzzy territorial profiling, cross-border stress analysis, reasoning over economic groupings, analyst-defined concepts with linguistic hedges and coherence checks of the vocabulary, and temporal trend analysis scenarios. All experiments are conducted on real-world data, and the source code, configuration files, and complete fuzzyDL export of the ontology are publicly available. The results show that fuzzy ontological reasoning provides an expressive, declarative, and reproducible framework for analysing sustainable-development indicators while preserving the semantics of linguistic concepts.