Parametric BIM–BPS automation loops enabling automated model translation and tighter BIM–BEM interoperabilityML/surrogate-based acceleration supporting rapid early-design performance prediction and flexible surrogate updatingCloud/web platforms providing interactive, platform-based analytical feedback

DEEPA[17] — BIM-linked Revit–IES workflow via gbXML; medium interactivity, with no uncertainty handling and low transparency. Autodesk Insight[18] — schematic–DD support with native Revit/cloud integration and high interactivity; parametric ranges are supported, but no formal MCDA or probabilistic uncertainty handling. ZEBO[19] — early energy-design configurator with manual gbXML exchange; deterministic evaluation with sensitivity analysis only.

Major gap — no reviewed DSS achieves fully automated, non-disruptive analytical workflows across design stages.

BIM–LCA/LCSA–MCDA integration combining environmental, economic, and multi-criteria decision layersMulti-objective optimisation workflows generating and comparing performance-oriented alternativesKPI aggregation & visualization dashboards synthesising multiple performance dimensions into interpretable decision views

Jrade et al. (2015)[34] — BIM-linked sustainable-material DSS using TOPSIS; static MCDA with no uncertainty handling. Majer et al. (2022)[35] — envelope-selection DSS using AHP + WSM with partial BIM integration and single-run MCDA. Nik-Bakht et al.[36] — energy–cost DSS linking Revit and EnergyPlus, with Pareto/scenario-based trade-off support. Autodesk Insight[18] — native Revit/cloud workflow with parametric performance comparison; no formal MCDA.

Moderate to major gap — trade-offs are partially supported but lack real-time, explicit, BIM-integrated reasoning.

Sensitivity & scenario-based analysis identifying influential variables and testing early-design assumptionsStochastic/probabilistic optimisation embedding uncertainty directly in early façade/material decisionsUncertainty-aware ML/surrogates using Bayesian or probabilistic learning to communicate predictive reliability

ZEBO[19] — deterministic early-design evaluation with sensitivity analysis; no explicit uncertainty modelling. MOOSAS[49] — simulation-accelerated instant-feedback workflow with sensitivity analysis only; no formal MCDA. pEnergy Analysis[50] — early-stage ML-surrogate DSS with probabilistic inference and sensitivity analysis; parametric BIM link.

Major gap — uncertainty is rarely represented explicitly or used to guide early-stage decision-making.

BIM–IoT–GIS–POE digital-twin pathways linking operational/sensor evidence to building representationsEnvironmental & LCA data integration embedding life-cycle evidence in early BIM-based decisionsMarket/cost/policy-linked data layers bringing contextual material and cost information into decision supportKnowledge-graph / ontology pathways semantically integrating heterogeneous building information

Jrade et al. (2015)[34] — BIM-linked material-selection DSS using MCDA and structured sustainability/material information. Majer et al. (2022)[35] — envelope MCDA using AHP + WSM with market-price data; no learning mechanism. Cove.tool[16] — early-design multi-metric DSS with Revit/Rhino plugins and real market/carbon databases; no learning mechanism.

Moderate to major gap — partial support for real-world data integration (C13) but persistent absence of learning and adaptivity mechanisms (C14).

Typology/archetype-driven defaults using representative building types to inform early assumptionsMulti-LOD and information-need frameworks managing progressive and uncertain early-stage informationRule/template-based completion using BIM libraries and structured property data to reduce missing-input burdenSurrogate estimation from limited early inputs providing fast performance support when detailed data are not yet available

MOOSAS[49] — massing-stage instant-feedback tool using accelerated simulation and an optional ANN module. pEnergy[50] — early-stage ML-surrogate prediction with a parametric BIM link.

Moderate gap — partial support through surrogates and defaults, but no integrated BIM-based inference framework.

Data-driven knowledge discovery extracting reusable decision knowledge from disparate building dataKnowledge graphs & ontologies encoding domain concepts and relationships for structured reasoningKnowledge-/rule-based DSS modules formalising expert rules and structured decision knowledgeCase-based reasoning & precedent retrieval reusing contextualised project knowledge through BIM-linked retrieval

Solibri[83] — IFC-native rule-based compliance checking with explicit rule traceability. Jrade et al. (2015)[34] — BIM-linked material-selection DSS using TOPSIS with fully explainable ranking. TestFit[84] — procedural rule-engine feasibility/yield tool with visible constraints but weak rationale transparency.

Major gap — expert knowledge remains largely tacit and is not systematically encoded or reused in DSSs.

Decision-logic unpacking & traceability making recommendations, criteria, and reasoning paths more inspectableXAI-enhanced feedback using interpretable AI mechanisms to expose prediction drivers and limitationsHuman-controlled AI assistance preserving designer participation and agency within AI-supported conceptual design

Autodesk Insight[18] — highly interactive Revit/cloud performance workflow with limited assumption transparency and partial traceability. Cove.tool[16] — highly interactive early-design DSS with limited transparency of ML internals. VENIS[14] — ML-surrogate DSS combining sensitivity-guided MCDA, evolutionary optimisation, explicit probabilistic uncertainty, and strong interpretability; no adaptive learning.

Critical gap — isolated advances in explainability exist, but BIM-integrated transparency largely absent.