Reasoning with Evidence, Not Merely Rationales: Verifiable Preference Proofs for LLM-Based Recommendation

2026-10-02

AI Summary

This study addresses the disconnection between rationales and evidence, as well as the weak influence of explanations, in large language model-based recommendation. To this end, it proposes PROVE-REC, a novel framework that introduces the concept of a "grounding-influence" gap. By generating verifiable preference proofs through a dual-channel architecture, the method integrates masked contrastive learning with ranking-preserving objectives to enable end-to-end optimization. This design compels the model to reason strictly from evidence while quantifying its actual causal impact on item rankings. Experimental results demonstrate that PROVE-REC outperforms the strongest baselines by up to 7.45% on real-world datasets, substantially enhancing both the evidence-groundedness of recommendations and their ranking effectiveness.

Abstract

Large language models (LLMs) can infer user preferences from interaction histories and reviews, yet the rationales they generate may not reflect the information actually used for recommendation. A preference claim may be weakly supported by its selected evidence, or may have little effect on the final ranking. We refer to these two failures as the grounding-influence gap. We introduce PROVE-REC, a general framework for verifiable preference reasoning in LLM-based recommendation. Pass A converts the complete pre-target history into a compact preference proof consisting of positive and avoidance claims linked to selected evidence entries. Pass B predicts the next item using only the proof and its selected evidence, preventing the recommender from bypassing the reasoning path. To verify evidence-to-proof grounding, we compare the effect of masking selected evidence with masking a comparable control entry. To verify proof-to-recommendation influence, we remove a preference claim and measure the resulting decrease in the target item's ranking margin. A ranking-preservation objective further retains useful information from the complete history. Comprehensive experiments on wide-ranging real-world datasets demonstrate that PROVE-REC consistently outperforms strong sequential, generative, and LLM-enhanced baselines, with improvements of up to 7.45%. Controlled ablations confirm the effectiveness of the two-pass architecture and verification objectives. Moreover, PROVE-REC produces claims that are more strongly grounded in historical evidence and more influential to recommendation while preserving ranking quality.

Problem

• Large Language Models
• Recommendation Systems
• Preference Reasoning
• Grounding-Influence Gap
• Verifiability

Innovation

• Verifiable Preference Reasoning
• Two-Pass Architecture
• Grounding-Influence Gap
• LLM-Based Recommendation
• Preference Proof