UniGD：面向工业检索的统一生成-判别框架
UniGD: A Unified Generative-Discriminative Framework for Industrial Retrieval
浏览论文内容
中文总结 AI 辅助
UniGD是整合检索与相关性评分的统一框架，通过CAGE、CAM、HAM解决工业检索问题，在快手平台及NQ320K、MS300K数据集上均取得显著性能提升。
中文摘要 AI 辅助
生成式检索（GR）是工业搜索广告领域极具潜力的范式，但其部署受限于严格的相关性与延迟要求。现有系统将GR与独立的相关性模型级联，使生成式似然目标与查询-广告相关性判别相分离，这既损害了有效性，又增加了服务成本。本文提出统一生成-判别框架（UniGD），将检索与相关性评分整合至单一模型中。为缓解联合优化中的梯度干扰，UniGD引入冲突感知梯度增强（CAGE），自适应协调两个目标；还设计了码本锚定表示模块（CAM），将物品表示锚定至从多模态预训练模型蒸馏而来的冻结分层码本，从而赋予其丰富且可泛化的语义先验。针对异构短视频、商品及直播广告，UniGD提出异构广告素材建模（HAM），在共享主干上捕捉跨类型语义共性，同时保留类型特定的建模能力。在快手搜索广告平台的在线AB测试显示，UniGD使广告收入提升5.78%，推理延迟降低33%，并优化了判别式相关性估计；在NQ320K和MS300K数据集上，UniGD较最强复现的GR基线的Recall@10分别提升8.44%和3.19%。
英文摘要
Generative retrieval (GR) is a promising paradigm for industrial search advertising, yet its deployment is constrained by strict relevance and latency requirements. Existing systems cascade GR with an independent relevance model, decoupling the generative likelihood objective from query-ad relevance discrimination, which compromises effectiveness and increases serving costs. We propose a Unified Generative-Discriminative framework (UniGD) that integrates retrieval and relevance scoring within a single model. To mitigate gradient interference in joint optimization, UniGD introduces Conflict-Aware Gradient Enhancement (CAGE) to adaptively coordinate the two objectives. UniGD further designs a Codebook-Anchored Representation Module (CAM) that anchors item representations to frozen hierarchical codebooks distilled from a multimodal pretrained model, thereby endowing them with rich and generalizable semantic priors. For heterogeneous short-video, product, and live-stream ads, UniGD proposes Heterogeneous Ad-material Modeling (HAM), which captures cross-type semantic commonality over a shared backbone while preserving type-specific modeling capacity. Online AB tests on Kuaishou search advertising platform show that UniGD raises ad revenue by 5.78%, reduces inference latency by 33%, and improves discriminative relevance estimation. On NQ320K and MS300K, UniGD improves Recall@10 over the strongest reproduced GR baseline by 8.44% and 3.19%, respectively.
