Uncovering the Bigger Picture: Comprehensive Event Understanding Via Diverse News Retrieval
This work addresses the need for diverse perspectives in event understanding for news consumers, though it is incremental as it builds on existing retrieval methods with a novel diversity-focused approach.
The paper tackles the problem of redundant and limited viewpoint exposure in news retrieval systems by proposing NEWSCOPE, a two-stage framework that enhances event coverage through sentence-level semantic variation modeling, achieving significantly higher diversity without compromising relevance.
Access to diverse perspectives is essential for understanding real-world events, yet most news retrieval systems prioritize textual relevance, leading to redundant results and limited viewpoint exposure. We propose NEWSCOPE, a two-stage framework for diverse news retrieval that enhances event coverage by explicitly modeling semantic variation at the sentence level. The first stage retrieves topically relevant content using dense retrieval, while the second stage applies sentence-level clustering and diversity-aware re-ranking to surface complementary information. To evaluate retrieval diversity, we introduce three interpretable metrics, namely Average Pairwise Distance, Positive Cluster Coverage, and Information Density Ratio, and construct two paragraph-level benchmarks: LocalNews and DSGlobal. Experiments show that NEWSCOPE consistently outperforms strong baselines, achieving significantly higher diversity without compromising relevance. Our results demonstrate the effectiveness of fine-grained, interpretable modeling in mitigating redundancy and promoting comprehensive event understanding. The data and code are available at https://github.com/tangyixuan/NEWSCOPE.