Panoramic Interests: Stylistic-Content Aware Personalized Headline Generation

2025-01-21

AI Summary

Existing news headline generation methods model only content preferences, neglecting stylistic preferences, thereby failing to achieve genuine personalization. To address this, we propose SCAPE—the first holistic user interest modeling framework that jointly captures both content and style preferences. SCAPE innovatively integrates LLM-coordinated feature extraction with a contrastive learning–driven hierarchical fusion mechanism for short- and long-term interests, and further incorporates a personalized sequence generation module. Evaluated on the real-world dataset PENS, SCAPE significantly outperforms state-of-the-art methods: it improves headline click-through rate by 12.7% and user satisfaction by 9.3%. Notably, it is the first approach to enable dual-dimensional personalization—simultaneously optimizing for both content relevance and stylistic alignment—in headline generation. This work establishes a novel paradigm at the intersection of news recommendation and generative modeling.

Technology Category

• Natural Language Processing: Generation
• Humans and AI: Game Design — Procedural Content Generation & Storytelling
• Machine Learning: Learning Preferences or Rankings

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Web Mining and Content Analysis: Large pretrained models with web data
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search

Abstract

Personalized news headline generation aims to provide users with attention-grabbing headlines that are tailored to their preferences. Prevailing methods focus on user-oriented content preferences, but most of them overlook the fact that diverse stylistic preferences are integral to users' panoramic interests, leading to suboptimal personalization. In view of this, we propose a novel Stylistic-Content Aware Personalized Headline Generation (SCAPE) framework. SCAPE extracts both content and stylistic features from headlines with the aid of large language model (LLM) collaboration. It further adaptively integrates users' long- and short-term interests through a contrastive learning-based hierarchical fusion network. By incorporating the panoramic interests into the headline generator, SCAPE reflects users' stylistic-content preferences during the generation process. Extensive experiments on the real-world dataset PENS demonstrate the superiority of SCAPE over baselines.

Problem

• News Title Generation
• Personalization
• Style Preferences

Innovation

• SCAPE
• Personalized News Headlines
• Style Preference Integration