[论文解读] UniShare: A Unified Framework for Joint Video and Receiver Recommendation in Social Sharing

Caimeng Wang, Li Chong+3 more|arXiv (Cornell University)|Feb 10, 2026

Recommender Systems and Techniques被引用 0

一句话总结

UniShare 提出一个统一的、端到端的联合模型，用于预测社交分享中的分享概率和接收者选择，利用双向兴趣、关系-内容对齐以及多任务训练，在真实世界的 Kuaishou 数据集上取得提升。 在线部署显示在分享、接收者满意度等方面的收益。

ABSTRACT

Sharing behavior on short-video platforms constitutes a complex ternary interaction among the user (sharer), the video (content), and the receiver. Traditional industrial solutions often decouple this into two independent tasks: video recommendation (predicting share probability) and receiver recommendation (predicting whom to share with), leading to suboptimal performance due to isolated modeling and inadequate information utilization. To address this, we propose UniShare, a novel unified framework for joint sharing prediction on both video and receiver recommendation. UniShare models the share probability through an enhanced representation learning module that incorporates pre-trained GNN and multi-modal embeddings, alongside explicit bilateral interest and relationship matching. A key innovation is our joint training paradigm, which leverages signals from both tasks to mutually enhance each other, mitigating data sparsity and improving bilateral satisfaction. We also introduce K-Share, a large-scale real-world dataset constructed from Kuaishou platform logs to support research in this domain. Extensive offline experiments demonstrate that UniShare significantly outperforms strong baselines on both tasks. Furthermore, online A/B testing on the Kuaishou platform confirms its effectiveness, achieving significant improvements in key metrics including the number of shares (+1.95%) and receiver reply rate (+0.482%).

研究动机与目标

• 识别分离的视频与接收者推荐在社交分享中的局限性。
• 提出一个统一框架，联合建模三元组（用户，视频，接收者）的分享概率。
• 通过双向兴趣建模、关系-内容对齐以及多模态信号，提升表征能力。
• 通过分层负采样与联合训练缓解数据稀疏性。
• 提供一个新的大规模数据集（K-Share）用于基准测试分享预测模型。

提出的方法

• 将分享概率建模为联合函数 P(S=1|U,I,V) 的统一架构。
• 结合预训练GNN嵌入和多模态视频/用户嵌入以缓解稀疏性。
• 使用定向注意力的双向兴趣建模捕捉对 sharer 与 receiver 的内容对齐。
• 使用LLM基于语义匹配的关系-内容对齐以及GNN派生的社交信号。
• 采用分层负采样与共享嵌入对视频与接收者任务进行联合训练，以实现相互提升。
• 在K-Share数据集上进行评估，指标包括AUC、GAUC、NDCG、Recall和MRR；并进行消融以量化各组件贡献。

Figure 1. A screenshot illustrating the sharing flow on the Kuaishou platform. The process consists of three stages: (1) watching a video and generating sharing intent and clicking the share button to invoke the sharing panel, and (2) selecting a receiver to complete the sharing action, followed by

实验结果

研究问题

• RQ1 统一模型是否能在社交流任务中超越独立的视频和接收者推荐基线？
• RQ2 双向兴趣与关系-内容对齐如何影响对分享者与接收者两端的分享质量？
• RQ3 联合训练是否缓解数据稀疏性并改善长尾视频与接收者的表现？
• RQ4 分层负采样与参数共享对模型有效性有何影响？
• RQ5 在真实平台上的在线部署中，统一方法的表现如何？

主要发现

• UniShare 在 K-Share 的离线评估中在两个任务上均优于分离的基线。
• 视频分享的 AUC 从 0.7512（PLE）提升到 0.7588（UniShare）(+1.01%)；GAUC 从 0.6919 提升到 0.6976 (+0.82%)。
• 接收者推荐的 AUC 从 0.9124（DCN）提升到 0.9307（UniShare）(+2.01%)；GAUC 从 0.9140 提升到 0.9282 (+1.55%)。
• 视频的 Recall@10 提升到 0.2913（原始 0.2869，提升 +1.53%）。
• 消融实验表明组件贡献；移除 BIM、RCA 或 HNS 会降低性能，HNS 对接收者 AUC 的影响尤为显著。
• 在 KuaiShou 的在线 A/B 测试显示：分享量提升 +1.95%，独立分享者提升 +0.805%，分享按钮点击率（CTR）提升 +1.12%，面板 CTR 提升 +1.14%，接收者回复率提升 +0.482%。

Figure 2. Overview of the UniShare Framework. UniShare jointly trains the video recommendation and receiver recommendation tasks, enhancing performance on both through underlying information sharing. Pre-computed GNN embeddings, multi-modal embeddings, and relationship alignment features are incorpo