ElementMindX: Offline Supplier-Substitution Ranking for Natural-Language Trade-Shock Decision Support

Trade disruptions, whether from tariffs, export bans, or production shocks force analysts to make high-stakes sourcing decisions under deep uncertainty. We present \method, an interactive decision-support tool that translates natural-language scenario descriptions into counterfactual simulations over global trade graphs. While the deployed system supports materials ranging from silicon and tantalum to energy products, this paper evaluates the supplier-ranking module on copper. We approach supplier substitution as an offline learning-to-rank task: using historical exposure events in observational trade panels, we train models to rank plausible alternative suppliers based on observed next-period market-share gains. \method then pairs these learned rankings with a transparent heuristic allocator to visualize volume rerouting and estimate shortages. In a pilot study of 1,277 copper importer-shock groups from 1992--2022, the learned rankers improve over global-share and incumbent-share baselines on Hit@1 and NDCG@10 across three copper HS6 markets.