Large-scale retrieval systems must operate under strict latency constraints while maintaining high recall. Sparse retrieval offers efficiency and interpretability, whereas dense retrieval provides stronger semantic matching. Although hybrid approaches combine both signals, their interaction is often limited, especially under intersection-based retrieval. We introduce HybridSparse, an end-to-end hybrid retrieval framework that strengthens sparse--dense interaction across modeling, training, and serving. It adopts a unified encoder with a shared backbone and jointly optimizes lexical and semantic representations through co-training. To further improve alignment, we incorporate hybrid score regularization and consistency distillation, enabling more stable and effective hybrid scoring. Experiments on public benchmarks demonstrate consistent improvements over strong sparse, dense, and hybrid baselines. In large-scale production deployment for Bing advertisement retrieval, HybridSparse delivers a +1.30% RPM gain, highlighting its practical impact.
Haotong Bao, Jianjin Zhang, Weihao Han et al.· Annual International ACM SIG...· 0 citations
Tumor protein p53-regulated apoptosis-inducing protein 1 (TP53AIP1) has been implicated in tumor suppression, but its role in breast cancer remains unclear. This study evaluated the expression pattern, prognostic value, immune infiltration association, methylation status, and biological function of TP53AIP1 in breast cancer using TCGA transcriptomic data, public methylation datasets, immunohistochemistry, and in vitro experiments. TP53AIP1 was significantly downregulated in breast cancer tissues and was identified as an independent prognostic factor for poor survival. TP53AIP1 expression was positively associated with transcriptome-estimated infiltration of natural killer cells, mast cells, and plasmacytoid dendritic cells. Methylation analysis showed that TP53AIP1 promoter hypermethylation was associated with reduced TP53AIP1 expression, suggesting a potential epigenetic silencing mechanism. Functionally, TP53AIP1 overexpression suppressed breast cancer cell proliferation, migration, invasion, and epithelial-mesenchymal transition and, promoted apoptosis and cell-cycle arrest. Mechanistically, TP53AIP1 overexpression reduced MEK/ERK phosphorylation, whereas MAPK pathway reactivation partially reversed its inhibitory effects on malignant phenotypes. These findings suggest that TP53AIP1 may serve as a potential prognostic biomarker and tumor-suppressive candidate in breast cancer, with its effects at least partly associated with MAPK pathway attenuation.
Meihai Deng, Xueting Wu, J. Bai et al.· Discover Oncology· 0 citations