Abstract

In airport terminals, commercial conversion is constrained by boarding deadlines, fixed shop locations, and passengers’ limited spatio-temporal accessibility. Conventional One ID profiling is difficult to apply because passenger records and transaction logs are anonymized and disconnected, while individual preferences may shift across trips. This paper proposes Similar One, a behavioral equivalence framework that infers recurrent passenger-flow patterns under comparable time pressure, gate location, and feasible exposure sets. The framework formulates anonymous airport retail attribution as an Unbalanced Optimal Transport problem, aligning source-side candidate trajectories with target-side POS absorption events while allowing unexplained mass to flow into a dummy sink. Based on this alignment, Reverse Recommendation is defined as a managerial intervention that reshapes passengers’ interaction windows and redirects exposure toward fixed commercial nodes. Empirical analysis suggests a feasible-set contraction boundary around 102.6 minutes, with 59.2% of passenger-side mass affected, and reveals heterogeneous boundary shifts across transaction days and passenger segments.

Download

Share

COinS