Following the TRAIL: Predicting and Explaining Tomorrow's Hits with a Fine-Tuned LLM

2026-02-04

AI Summary

Traditional recommender systems struggle to accurately model user preferences under sparse interactions, often lack trustworthy explanations, and incur high computational costs for full-scale real-time ranking. To address these challenges, this work proposes TRAIL—a fine-tuned large language model (LLM) framework that uniquely unifies short-term item popularity prediction with natural language explanation generation. By aligning structured trend signals with textual explanations through contrastive learning and leveraging carefully constructed positive-negative sample pairs, TRAIL simultaneously enhances prediction accuracy and produces coherent, evidence-based explanations. Extensive experiments demonstrate that the proposed method significantly outperforms strong baselines across multiple metrics, achieving both high accuracy and strong interpretability.

Technology Category

• Machine Learning: Learning Preferences or Rankings
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Humans and AI: Explainable AI (XAI) for Human Understanding

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Large Language Models (LLMs) have been widely applied across multiple domains for their broad knowledge and strong reasoning capabilities. However, applying them to recommendation systems is challenging since it is hard for LLMs to extract user preferences from large, sparse user-item logs, and real-time per-user ranking over the full catalog is too time-consuming to be practical. Moreover, many existing recommender systems focus solely on ranking items while overlooking explanations, which could help improve predictive accuracy and make recommendations more convincing to users. Inspired by recent works that achieve strong recommendation performance by forecasting near-term item popularity, we propose TRAIL (TRend and explAnation Integrated Learner). TRAIL is a fine-tuned LLM that jointly predicts short-term item popularity and generates faithful natural-language explanations. It employs contrastive learning with positive and negative pairs to align its scores and explanations with structured trend signals, yielding accurate and explainable popularity predictions. Extensive experiments show that TRAIL outperforms strong baselines and produces coherent, well-grounded explanations.

Problem

• recommendation systems
• Large Language Models
• explainability
• short-term popularity prediction
• user-item logs

Innovation

• Large Language Models
• Explainable Recommendation
• Contrastive Learning
• Popularity Prediction
• Fine-tuned LLM