Analyzing and Mitigating Repetitions in Trip Recommendation

2025-07-26

AI Summary

In trip recommendation, severe POI repetition in model outputs critically undermines both diversity and accuracy. To address this, we propose AR-Trip, the first framework to integrate logits perturbation into both training and decoding phases of trip recommendation, coupled with a recurrence-aware predictor that dynamically suppresses the re-generation probability of previously recommended POIs via historical sequence modeling. AR-Trip introduces three key innovations: (1) attention-based repetition sensitivity modeling; (2) adaptive logits masking during decoding; and (3) a cycle-consistency constraint loss. Extensive experiments on four public benchmarks demonstrate that AR-Trip significantly reduces repetition rate by 38.7% on average, while simultaneously improving Recall@10 and MRR by 5.2% and 4.8%, respectively—effectively balancing recommendation diversity and accuracy.

Technology Category

• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Knowledge Representation and Reasoning: Preferences
• Natural Language Processing: Generation

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Trip recommendation has emerged as a highly sought-after service over the past decade. Although current studies significantly understand human intention consistency, they struggle with undesired repetitive outcomes that need resolution. We make two pivotal discoveries using statistical analyses and experimental designs: (1) The occurrence of repetitions is intricately linked to the models and decoding strategies. (2) During training and decoding, adding perturbations to logits can reduce repetition. Motivated by these observations, we introduce AR-Trip (Anti Repetition for Trip Recommendation), which incorporates a cycle-aware predictor comprising three mechanisms to avoid duplicate Points-of-Interest (POIs) and demonstrates their effectiveness in alleviating repetition. Experiments on four public datasets illustrate that AR-Trip successfully mitigates repetition issues while enhancing precision.

Problem

• Analyzing repetition issues in trip recommendation models
• Mitigating undesired repetitive outcomes in POI suggestions
• Improving precision while reducing duplicate recommendations

Innovation

• Cycle-aware predictor prevents duplicate POIs
• Perturbations added to logits reduce repetition
• AR-Trip enhances precision and mitigates repetitions