互联网实时竞价广告的若干关键问题研究

Real-time bidding (RTB) is an emerging business model for online targeted advertising following the integration of computational advertising and big data research. It has delivered disruptive and transformative innovation for both targeting technology and selling model of online advertising. To date, RTB markets are still experiencing continuing trail-and-error and optimization, and lack of effective business models and decision-making measures. As a result, RTB markets are exposed to severe problems and challenges in macroscopic mechanism design and microscopic decision-making. In this project, we will conduct research on behavioral analysis, decision-making and trading mechanism design, focusing on such significant issues as bidding behavior analysis and large-population games, optimization of market segmentation granularity, multi-channel inventory management, market mechanism design and evaluation. We will also build an agent-based computational-experimental platform, aiming at validating and evaluating our research findings. The proposed project will not only enrich and develop the existing research on E-commerce and computational advertising, but also help promote the healthy and sustainable development of RTB business model and its market ecosystems. Our research findings are expected to provide useful guidance and reference for the emerging and fast-growing RTB markets.

实时竞价(Real-Time Bidding, RTB)是随着计算广告学和大数据的深度融合而诞生的精准广告模式，为互联网广告的投放和售卖模式均带来了颠覆性变革。目前，RTB市场尚处于持续试错和优化的探索阶段，缺乏行之有效的商业模式与决策手段，导致RTB实践中在宏观市场机制设计和微观参与者决策支持两个层面均面临着极大的问题与挑战。面向市场实际问题与需求，本项目拟从参与者行为分析、决策支持和市场交易机制设计三条主线开展研究，重点研究参与者的投标行为模式与大群体博弈、市场细分粒度优化、跨渠道广告库存管理、交易机制设计与评估等关键问题，并建立基于智能体仿真的RTB计算实验平台来验证和评估项目的理论研究成果。本项目不仅对于电子商务和计算广告学基础理论研究具有重要意义，同时有助于促进RTB商业模式及其市场"生态系统"的健康和可持续发展，可望为萌芽和高速发展期的新兴RTB市场提供有益的理论指导与借鉴。

本课题的研究工作始终按计划进行，进展顺利，通过四年系统深入的研究，取得了一批具有原创性的研究成果，圆满完成了项目的预期研究目标。项目执行期间，项目负责人及其团队研究成果如下：发表学术论文66篇（含已接收），均已标注基金委项目资助。其中国际期刊论文20篇，国际会议论文26篇，国内重要学术期刊论文11篇，国内会议文章9篇；其中，SCI检索论文11篇，ESCI检索论文8篇，EI检索论文47篇；项目负责人获得2018年度“中国自动化学会青年科学家奖”，其第一作者论文获得“领跑者5000-中国精品科技期刊顶尖学术论文奖”、《自动化学报》2017“年度优秀论文奖”等奖项；建设完成基于RTB的DSP平台“北斗广告伺服系统”和“互联网RTB广告实验平台”；参加国际或国内重要学术会议10余次。项目产生了良好的学术影响力，在Web of Science数据库的“Real-Time Bidding”主题检索结果中，项目负责人的领域研究国际排名为第一位，项目组所在团队占据前三位。