Leveraging Reviews: Learning to Price with Buyer and Seller Uncertainty
Leveraging Reviews: Learning to Price with Buyer and Seller Uncertainty
复制标题
利用评论：学习在买家和卖家的不确定性下定价
DOI：
10.1145/3580507.3597663
复制
发表时间：
2023
期刊：
影响因子：
--
通讯作者：
Vitercik, Ellen
中科院分区：
文献类型：
--
作者：
Guo, Wenshuo;Haghtalab, Nika;Kandasamy, Kirthevasan;Vitercik, Ellen
客户可以访问在线市场上单个产品的数百条评论。买家经常使用来自其他客户的评论，这些评论与他们的类型相同-例如衣服的身高或护肤产品的皮肤类型-来估计他们的价值，他们可能事先不知道。只有很少相关评论的客户可能会犹豫购买，除非价格低，所以对于卖家来说，在设定高价和确保有足够的评论之间存在紧张关系，以便买家可以自信地估计他们的价值。同时，卖家可以使用评论来衡量他们希望出售的商品的需求。在这项工作中，我们研究这个定价问题，在一个在线设置中，卖方与一组买家的许多类型，一个接一个，在一系列ofTrounds。在每一轮中，卖方首先设定一个价格。然后，买家到达并检查具有相同类型的先前买家的评论，这些评论揭示了这些买家的事后价值。根据评论，买家决定购买，如果他们有充分的理由相信他们的事前效用是积极的。至关重要的是，卖方在定价时不知道买方的类型，甚至不知道类型的分布。我们提供了一个不后悔的算法，卖家可以使用它来获得高收入。当有d个类型时，经过循环，我们的算法达到了一个问题无关的（T2/3d 1/3）后悔界。然而，当任何给定类型出现的最小概率qmin很大时，特别是当qmin ∈ Ω（d−2/3 T −1/3）时，同样的算法达到了一个后悔界。我们补充这些上限与匹配的下限在这两个政权，表明我们的算法是最小最大最优的低阶terms.This is a summary of work that won theExemplary AI Track Paper Award at EC'24.
Customers can access hundreds of reviews for a single product in online marketplaces. Buyers often use reviews from other customers that share their type---such as height for clothing or skin type for skincare products---to estimate their values, which they may not know a priori. Customers with few relevant reviews may hesitate to purchase except at a low price, so for the seller, there is a tension between setting high prices and ensuring that there are enough reviews so buyers can confidently estimate their values. Simultaneously, sellers may use reviews to gauge the demand for items they wish to sell. In this work, we study this pricing problem in an online setting where the seller interacts with a set of buyers of finitely many types, one by one, over a series ofTrounds. At each round, the seller first sets a price. Then, a buyer arrives and examines the reviews of the previous buyers with the same type, which reveal those buyers' ex-post values. Based on the reviews, the buyer decides to purchase if they have good reason to believe their ex-ante utility is positive. Crucially, the seller does not know the buyer's type when setting the price, nor even the distribution over types. We provide a no-regret algorithm that the seller can use to obtain high revenue. When there aredtypes, afterTrounds, our algorithm achieves a problem-independentÕ(T2/3d1/3) regret bound. However, when the smallest probabilityqminthat any given type appears is large, specifically whenqmin∈ Ω(d−2/3T−1/3), the same algorithm achieves a [EQUATION] regret bound. We complement these upper bounds with matching lower bounds in both regimes, showing that our algorithm is minimax optimal up to lower-order terms.This is a summary of work that won theExemplary AI Track Paper Awardat EC'24.
登录
查看更多内容
产品市场中的异构学习
DOI：
10.2139/ssrn.3961223
发表时间：
2021
期刊：
SSRN Electronic Journal
影响因子：
--
作者：
A. Kakhbod;Giacomo Lanzani;Hao Xing
通讯作者：
Hao Xing
在线评论中的客户动机和反应偏差
影响因子：
3.5
作者：
Saram Han;C. Anderson
通讯作者：
C. Anderson
从评论中快速和缓慢地学习
DOI：
10.2139/ssrn.3072495
发表时间：
2017
期刊：
NBER Working Paper Series
影响因子：
--
作者：
D. Acemoglu;A. Makhdoumi;Azarakhsh Malekian;A. Ozdaglar
通讯作者：
A. Ozdaglar
随机单边全信息老虎机
DOI：
10.1007/978-3-030-46133-1_10
发表时间：
2019
期刊：
ArXiv
影响因子：
--
作者：
Haoyu Zhao;Wei Chen
通讯作者：
Wei Chen
DOI：
10.1145/3219166.3219233
发表时间：
2017-11
期刊：
Proceedings of the 2018 ACM Conference on Economics and Computation
影响因子：
--
作者：
M. Braverman;Jieming Mao;Jon Schneider;Matt Weinberg
通讯作者：
M. Braverman;Jieming Mao;Jon Schneider;Matt Weinberg