Hard Rules, Soft Preferences: Bridging Reasoning, Learning, and Optimization for Personalized Packing Checklist Generation

2026-07-16

AI Summary

This work addresses the challenge of generating personalized packing lists for air travel that simultaneously satisfy hard constraints—including aviation safety regulations, item dependencies, and baggage limitations—while aligning with user preferences. The authors propose a three-stage framework integrating symbolic reasoning, preference learning, and constraint optimization: a symbolic engine first produces a compliant initial list; a two-stage preference model (combining gradient-boosted trees and LambdaMART) then infers item utilities and priorities from sparse user interactions; finally, a CP-SAT solver selects a compact, feasible subset. Evaluated on 604 real-world itineraries, the approach achieves 99.7% compliance recall, AUC-ROC of 0.943, NDCG@5 of 0.923, and perfect constraint satisfaction. Upon deployment, it doubled in-app list completion rates and significantly reduced both editing and completion time, effectively overcoming the limitations of template-based or unconstrained recommendation systems.

Abstract

Packing for air travel is recurring and error-prone: the checklist must be personal and context-aware, yet feasible under safety rules, item dependencies, and luggage limits. Existing packing assistants are template-driven and generic, or recommendation-driven but unconstrained, leaving users to manually patch regulatory and capacity violations. We propose a reasoning-guided learning framework with three stages: (1) a symbolic engine that generates a regulation-aware seed checklist with explicit dependency structure, (2) a two-stage preference learner that estimates inclusion and priority utilities from user add and remove actions while mitigating survivorship bias, and (3) a CP-SAT optimizer that selects a compact, compliant subset. The architecture instantiates a general pattern for constrained personalization, applicable wherever hard feasibility coexists with sparse preference signals. On 604 labeled trip scenarios, comprising 29K inclusion labels and 343K pairwise comparisons, the symbolic engine attains 99.7% recall and 0.96 rubric validity, compared with 0.78 to 0.81 for frontier LLMs. Gradient-boosted trees and LambdaMART reach an AUC-ROC of 0.943 and an NDCG@5 of 0.923. CP-SAT attains 100% constraint satisfaction, compared with 28% for greedy selection and 10% for random selection. Deployment in FlyEnJoy, a production iOS travel app, doubled checklist completions and reduced editing and completion time.

Problem

• personalized packing checklist
• constraint satisfaction
• preference learning
• travel planning
• regulatory compliance

Innovation

• constrained personalization
• symbolic reasoning
• preference learning
• CP-SAT optimization
• survivorship bias mitigation