| 保量和非保量合同中的曝光资源整合分配优化 |
| Integrated allocation optimization of impression resource in guaranteed and non-guaranteed contracts |
| 摘要点击 32 全文点击 0 投稿时间：2024-05-07 修订日期：2025-03-05 |
| 查看/发表评论 下载PDF阅读器 |
| 中文关键词 展示广告; 保量合同; 非保量合同; 分布鲁棒; 整合优化 |
| 英文关键词 display advertising; guaranteed contracts; non-guaranteed contracts; distributionally robust; integrated optimization |
| 基金项目 国家自然科学基金项目（面上项目，重点项目，重大项目） |
| 作者 | 单位 | 邮编 | | 隋鑫 | 电子科技大学 | 611731 | | 代文强* | 电子科技大学 | 611731 | | 王俊洁 | 电子科技大学 | |
|
| 中文摘要 |
| 展示广告是在线广告中一种重要的投放形式, 在实际投放过程中, 发布商通过售卖曝光资源获得收益, 其中售卖方式可分为保量合同与非保量合同两种, 如何合理地将曝光资源分配给两种售卖方式以达到收益最大是发布商的目标. 然而, 曝光的不确定性给发布商的决策带来了极大挑战. 基于此, 本文研究保量和非保量合同中曝光资源整合分配优化问题, 通过历史数据获取曝光的部分分布信息, 构建不确定集, 首先提出带有联合机会约束的展示广告分布鲁棒分配模型. 其次提出投射定理, 得到保守近似的二阶锥规划(Second-Order Cone Programming, SOCP)模型并进行仿真分析. 最后在大规模样本外测试中, 通过与抽样平均近似(Sample Average Approximation, SAA)方法进行对比, 验证了本文模型和求解方法的有效性以及鲁棒性. 参数的灵敏度分析为发布商收集数据信息, 优化调整策略提供了启示和建议. |
| 英文摘要 |
| Display advertising is a crucial role in online advertisement, and the publisher obtains revenue by selling impression resources through guaranteed contracts and non-guaranteed contracts in the real advertising delivery process. The purpose of publisher is maximizing revenue by optimally allocating impression resource between the two contracts. However, the uncertainty in impression supply results is a challenge. This study focuses on integrated allocation optimization of impression resource in both guaranteed contracts and non-guaranteed contracts. By utilizing historical data to capture partial distribution information and construct an ambiguous set, a distributionally robust model with joint chance constraints for display ad allocation is proposed. Furthermore, the model is conservatively approximated as a Second-Order Cone Programming (SOCP) model by a proposed projection theorem and subjected to simulation analysis. Finally, in large-scale out-of-sample experiments, the effectiveness and robustness of th |
| 关闭 |
|
|
|
|
|
