arXiv：cs.AI· Yan Fang, Jialin Chen, Chun Gan, Hang Yu, Mingjun Nie, Yeyu Zhang, Fengxiang He, Ching Law·· 6 小时前
LLM-OSDA：面向多轮 LLM 对话原生广告的最优停止动态拍卖
LLM-OSDA: An Optimal-Stopping Dynamic Auction for Native Advertising in Multi-Turn LLM Conversations
AI 导读
研究者提出 LLM-OSDA，一种融合 Bellman 最优停止、赢家分配与包络定价的动态按点击付费拍卖，用于多轮 LLM 对话中的原生广告投放。在精确 Bellman oracle 下，其期望折扣点击分配对每家广告主的出价单调，包络支付使诚实出价在期望意义上弱占优；实际部署中用学习到的 StopNet 近似 Bellman 动作价值。
正文
Abstract:LLM-native advertising embeds sponsored content directly into model-generated responses, shifting the unit of sale from a fixed slot to a moment within an evolving conversation. Existing LLM ad-auction mechanisms primarily operate within a single response, settling the winner but not the timing. The extension is nontrivial: with one native insertion opportunity per session, the stopping time depends on bids, coupling timing with allocation, so static truthfulness arguments no longer apply. We propose the LLM-based Optimal Stopping Dynamic Auction (LLM-OSDA), a dynamic cost-per-click auction that integrates Bellman optimal stopping, winner allocation, and envelope pricing. A bid-independent LLM layer estimates contextual click quality and seamlessly renders the winning ad, while bids enter only the committed auction mechanism. Under an exact Bellman oracle, the expected discounted-click allocation is monotone in each advertiser's bid, and the corresponding envelope payment makes truthful bidding weakly dominant in expectation. For practical deployment, a learned StopNet approximates the Bellman action values. We show that its decisions differ from the optimal policy only near the stopping boundary and bound the resulting incentive loss in terms of its approximation error. Experiments on a simulated conversational advertising corpus show that LLM-OSDA improves net revenue by 11 percent over the strongest fixed-timing baseline while maintaining comparable user retention. Code is at this https URL.
| Comments: | 21 pages, 9 figures. Submitted to ACL Rolling Review (ARR), October 2026 cycle |
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computer Science and Game Theory (cs.GT); Machine Learning (cs.LG) |
| MSC classes: | 68T07 |
| ACM classes: | I.2.6; I.2.7 |
| Cite as: | arXiv:2608.00123 [cs.CL] |
| (or arXiv:2608.00123v3 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2608.00123 arXiv-issued DOI via DataCite |
Submission history
From: Yan Fang [view email]
[v1]
Fri, 31 Jul 2026 11:47:54 UTC (1,650 KB)
[v2]
Tue, 4 Aug 2026 07:53:57 UTC (1,650 KB)
[v3]
Tue, 13 Oct 2026 08:26:25 UTC (1,628 KB)
来源：arXiv：cs.AI · arxiv.org