Contractual standardisation and AI readiness in maritime governance: evidence from maritime contract-type analytics

Charterparty markets differ not only in cargo type but in the institutional structures governing how contracts are drafted, standardised, and reproduced. Drawing on institutional theory, this study treats maritime contract types—from standardised instruments such as waybills and bunkers to bespoke voyage charters and offshore agreements—as distinct contractual environments, and asks whether they differ systematically in readiness for AI-assisted document analytics. A corpus of 1,783 maritime legal documents was classified using a Legal-BERT multi-label classifier, evaluated through five-fold cross-validation repeated across three seeds, with contract-type-level area under the precision–recall curve (AUPRC) as the primary indicator of AI readiness. Performance varied substantially, from 0.281 for voyage charters to 0.916 for contracts of affreightment (global micro-F1 = 0.569), confirming that maritime document environments are not equally ready for AI-assisted analytics. The expected mediation pathway through textual regularity was not supported (Spearman ρ = 0.027, p = 0.924), indicating that the operative mechanism extends beyond textual homogeneity to label specificity, vocabulary exclusivity, and inter-document scope coherence. The findings carry practical implications for maritime governance institutions, standard-form bodies, and organisations deploying AI tools in contract review.