A Model Game Pricing Method Based on Data Purity

doi:10.11871/jfdc.issn.2096-742X.2023.05.006

Abstract

Abstract:

[Objective] In the big data era, data circulation and sharing has become the trend. Hence data pricing and trading methods have received a lot of attention. As an important part of data pricing, model pricing is the top priority in data pricing research. In model pricing, it is necessary to solve the problem of compensation for data owners and maximization of benefits for all participants. Furthermore, to make sure that data owners participate in data trading with confidence, and their privacy concerns must be satisfied. At the same time, data quality, which has a direct influence on the accuracy of learning models, is also the focus of this research. [Methods] In this paper, we propose a model game pricing method based on data purity, using data quality and noise as the compensation basis on the data acquisition side and game theory on the model selling side. [Results] The proposed method can give fair and convenient compensation to data owners and maximize the revenue of data platforms and model buyers. [Limitations] However, the gaming process in complex trading environments still needs to be improved accordingly. [Conclusions] The effectiveness of the methods is demonstrated by experiments and new ideas are provided for the development of model pricing and data marketplaces.

Key words: data quality, game theory, data marketplace, model pricing, privacy protection

JIANG Dong, ZHANG Xiaowei, YUAN Ye. A Model Game Pricing Method Based on Data Purity[J]. Frontiers of Data and Computing, 2023, 5(5): 63-73, https://cstr.cn/32002.14.jfdc.CN10-1649/TP.2023.05.006.

Figures/Tables 6

Fig.1

Table 1

Key dimensions and corresponding indicators"

符号	描述
$D i$	到数据交易市场中参与交易数据拥有者
$z i$	数据拥有者i提供的数据集
$ε i$	数据拥有者i提供的隐私敏感度，由差分隐私来衡量
$P o$	数据平台对数据拥有者的补偿
$P c$	模型买家的单位精度预算
$a c c$	模型买家提供的模型精准度
$Q i$	数据拥有者i所贡献数据集的质量分数
$C o m i$	数据拥有者i所贡献数据集的完整度分数
$R e d i$	数据拥有者i所贡献数据集的冗余度分数
$F r e i$	数据拥有者i所贡献数据集的新鲜度分数
$D P i$	数据拥有者i所贡献数据集的数据纯度
$U i$	在某种数据纯度条件下，该数据集训练出模型能够达到的最大精度
$V$	模型购买者在精度acc条件下可以得到的效用
$Φ$	模型购买者的收益函数
$Ψ$	数据平台的收益函数
$C$	数据平台的训练代价

Table 1

Fig.2

Fig.3

Fig.4

Fig.5

References 23

[1]	THE ECONOMISTS. The world’s most valuable resource is no longer oil, but data[EB/OL]. 2017. https://www.economist.com/leaders/2017/05/06/the-worlds-most-valuable-resource-is-no-longer-oil-but-data.
[2]	Data.gov[EB/OL]. https://www.data.gov.
[3]	UBALDI B. Open government data: Towards empirical analysis of open government data initiatives[J]. OECD Working Papers on Public Governance, 2013, 4: 947-959.
[4]	CHEN L, KOUTRIS P, KUMAR A. Towards model-based pricing for machine learning in a data marketplace[J]. Proceedings of the 2019 ACM International Conference on Management of Data, 2019: 1535-1552.
[5]	JIA R, DAO D, WANG B, et al. Efficient task-specific data valuation for nearest neighbor algorithms[J]. Proceedings of the VLDB Endowment, 2019, 12(11): 1610-1623. doi: 10.14778/3342263.3342637
[6]	SHAPLEY L S. A value for n-person games[J]. Classics in game theory, 1997, 69(1): 307-317.
[7]	张小伟, 江东, 袁野. 基于博弈论和拍卖的数据定价综述[J]. 大数据, 2021, 7(4):61-79. doi: 10.11959/issn.2096-0271.2021039
[8]	AGARWAL A, DAHLEH M A, SARKAR T. A marketplace for data: an algorithmic solution[J]. Proceedings of the 2019 ACM Conference on Economics and Computation, 2019: 701-726.
[9]	HADDADI S, GHASEMI A. Pricing-based stackelberg game for spectrum trading in self-organised heterogeneous networks[J]. IET Communications, 2016, 10(11): 1374-1383. doi: 10.1049/cmu2.v10.11
[10]	LV X Y, ZHANG R T, YUE J. Competition and Cooperation between participants of the Internet of Things Industry Value Chain[J]. Advances in Information Sciences & Service Sciences, 2012. 4(11): 406-412.
[11]	LIU J, LOU J, LIU J, et al. Dealer: an end-to-end model marketplace with differential privacy[J]. Proceedings of the VLDB Endowment, 2021, 14(6): 957-969. doi: 10.14778/3447689.3447700
[12]	STAHL F, VOSSEN G. Data quality scores for pricing on data marketplaces[C]. Proceedings of the Intelligent Information and Database Systems, 2016, Part I: 215-224.
[13]	AN B, XIAO M, LIU A, et al. Crowdsensing Data Trading based on Combinatorial Multi-Armed Bandit and Stackel-berg Game[C]. 2021 IEEE 37th International Conference on Data Engineering, 2021: 253-264.
[14]	GU B, YANG X, LIN Z, et al. Multiagent actor-critic net-work-based incentive mechanism for mobile crowdsensing in industrial systems[J]. IEEE Transactions on Industrial Informatics, 2020, 17(9): 6182-6191. doi: 10.1109/TII.2020.3024611
[15]	DUAN X, ZHAO C, HE S, et al. Distributed algorithms to compute Walrasian equilibrium in mobile crowdsensing[J]. IEEE Transactions on Industrial Electronics, 2016, 64(5): 4048-4057. doi: 10.1109/TIE.2016.2645138
[16]	ZHAN Y, LIU C H, ZHAO Y, et al. Free market of multi-leader multi-follower mobile crowdsensing: An incentive mechanism design by deep reinforcement learning[J]. IEEE Transactions on Mobile Computing, 2019, 19(10): 2316-2329. doi: 10.1109/TMC.7755
[17]	Bank-Marketing[EB/OL]. https://www.kaggle.com/datasets/henriqueyamahata/bank-marketing.
[18]	WANG R Y, STRONG D M. Beyond accuracy: what data quality means to data consumers[J]. Journal of management information systems, 1996, 12(4): 5-33.
[19]	YU H, ZHANG M. Data pricing strategy based on data quality[J]. Computers and Industrial Engineering, 2017, 112: 1-10. doi: 10.1016/j.cie.2017.08.008
[20]	YANG J, ZHAO C, XING C. Big data market optimization pricing model based on data quality[J]. Complexity, 2019(1): 1-11.
[21]	GHOSH A, ROTH A. Selling privacy at auction[J]. Proceedings of the 12th ACM conference on Electronic commerce, 2011: 199-207.
[22]	ZHANG M, BELTRAN F, LIU J. Selling data at an auction under privacy constraints[C]. Conference on Uncertainty in Artificial Intelligence, 2020, 124: 689-698.
[23]	SONG T, TONG Y, WEI S. Profit allocation for federated learning[C]. In 2019 IEEE International Conference on Big Data, 2019: 2577-2586.