What Makes an Ideal Quote? Recommending "Unexpected yet Rational" Quotations via Novelty

2025-12-15

arXiv.org

AI Summary

This work addresses the limitation of existing quotation recommendation systems, which predominantly rely on surface-level topic matching and fail to capture users’ preference for quotations that are “surprising yet fitting.” Drawing on defamiliarization theory, the authors propose NovelQR, a framework that formalizes quotation recommendation as a contextual selection problem balancing semantic coherence and novelty. The approach introduces a generative label proxy to construct multidimensional deep semantic tags for enhanced retrieval and incorporates a token-level novelty estimator coupled with a de-autoregressive bias mechanism for reranking. Extensive experiments on multilingual, multidomain datasets demonstrate that NovelQR significantly outperforms baseline methods in human evaluations, achieving higher scores in appropriateness, novelty, and appeal, while also matching or exceeding state-of-the-art performance on automatic novelty metrics.

Abstract

Quotation recommendation aims to enrich writing by suggesting quotes that complement a given context, yet existing systems mostly optimize surface-level topical relevance and ignore the deeper semantic and aesthetic properties that make quotations memorable. We start from two empirical observations. First, a systematic user study shows that people consistently prefer quotations that are ``unexpected yet rational''in context, identifying novelty as a key desideratum. Second, we find that strong existing models struggle to fully understand the deep meanings of quotations. Inspired by defamiliarization theory, we therefore formalize quote recommendation as choosing contextually novel but semantically coherent quotations. We operationalize this objective with NovelQR, a novelty-driven quotation recommendation framework. A generative label agent first interprets each quotation and its surrounding context into multi-dimensional deep-meaning labels, enabling label-enhanced retrieval. A token-level novelty estimator then reranks candidates while mitigating auto-regressive continuation bias. Experiments on bilingual datasets spanning diverse real-world domains show that our system recommends quotations that human judges rate as more appropriate, more novel, and more engaging than other baselines, while matching or surpassing existing methods in novelty estimation.

Problem

• quotation recommendation
• novelty
• semantic coherence
• unexpected yet rational
• defamiliarization

Innovation

• novelty-driven recommendation
• defamiliarization theory
• deep-meaning labels
• token-level novelty estimation
• quotation recommendation