Private Rank Aggregation in Central and Local Models

In social choice theory, (Kemeny) rank aggregation is a well-studied problem where the goal is to combine rankings from multiple voters into a single ranking on the same set of items. Since rankings can reveal preferences of voters (which a voter might like to keep private), it is important to aggregate preferences in such a way to preserve privacy. In this work, we present differentially private algorithms for rank aggregation in the pure and approximate settings along with distribution-independent utility upper and lower bounds. In addition to bounds in the central model, we also present utility bounds for the local model of differential privacy. ∗Work done while an intern at Google Research. 1

[1]  Aaron Roth,et al.  The Algorithmic Foundations of Differential Privacy , 2014, Found. Trends Theor. Comput. Sci..

[2]  Yoram Singer,et al.  Learning to Order Things , 1997, NIPS.

[3]  Thomas Steinke,et al.  Interactive fingerprinting codes and the hardness of preventing false discovery , 2014, 2016 Information Theory and Applications Workshop (ITA).

[4]  H. Young,et al.  A Consistent Extension of Condorcet’s Election Principle , 1978 .

[5]  Thomas Steinke,et al.  Tight Lower Bounds for Differentially Private Selection , 2017, 2017 IEEE 58th Annual Symposium on Foundations of Computer Science (FOCS).

[6]  V. Feldman,et al.  Interaction is necessary for distributed learning with privacy or communication constraints , 2019, STOC.

[7]  R. Graham,et al.  Spearman's Footrule as a Measure of Disarray , 1977 .

[8]  Moni Naor,et al.  Our Data, Ourselves: Privacy Via Distributed Noise Generation , 2006, EUROCRYPT.

[9]  M. Fligner,et al.  Distance Based Ranking Models , 1986 .

[10]  L. A. Goodman,et al.  Social Choice and Individual Values , 1951 .

[11]  Kunal Talwar,et al.  On the geometry of differential privacy , 2009, STOC '10.

[12]  Uri Stemmer,et al.  Heavy Hitters and the Structure of Local Privacy , 2017, PODS.

[13]  Nir Ailon,et al.  Aggregating inconsistent information: Ranking and clustering , 2008 .

[14]  Douglas W. Oard,et al.  Learning to Rank for Mathematical Formula Retrieval , 2021, SIGIR.

[15]  S L Warner,et al.  Randomized response: a survey technique for eliminating evasive answer bias. , 1965, Journal of the American Statistical Association.

[16]  Eric Horvitz,et al.  Social Choice Theory and Recommender Systems: Analysis of the Axiomatic Foundations of Collaborative Filtering , 2000, AAAI/IAAI.

[17]  Thomas Steinke,et al.  Robust Traceability from Trace Amounts , 2015, 2015 IEEE 56th Annual Symposium on Foundations of Computer Science.

[18]  Janardhan Kulkarni,et al.  Differentially Private Correlation Clustering , 2021, ICML.

[19]  Vassilis Zikas,et al.  How Private Are Commonly-Used Voting Rules? , 2020, UAI.

[20]  Moni Naor,et al.  Rank aggregation methods for the Web , 2001, WWW '01.

[21]  Jonathan Ullman,et al.  Fingerprinting Codes and the Price of Approximate Differential Privacy , 2018, SIAM J. Comput..

[22]  Claire Mathieu,et al.  Electronic Colloquium on Computational Complexity, Report No. 144 (2006) How to rank with few errors A PTAS for Weighted Feedback Arc Set on Tournaments , 2006 .

[23]  Martin Wistuba,et al.  Learning to Rank Learning Curves , 2020, ICML.

[24]  Raef Bassily,et al.  Linear Queries Estimation with Local Differential Privacy , 2018, AISTATS.

[25]  C. L. Mallows.  NON-NULL RANKING MODELS. I , 1957 .

[26]  Anke van Zuylen,et al.  Rank Aggregation: Together We're Strong , 2009, ALENEX.

[27]  Jeff A. Bilmes,et al.  Consensus ranking under the exponential model , 2007, UAI.

[28]  M. Trick,et al.  Voting schemes for which it can be difficult to tell who won the election , 1989 .

[29]  Vincent Conitzer,et al.  Improved Bounds for Computing Kemeny Rankings , 2006, AAAI.

[30]  Marina Meila,et al.  Tractable Search for Learning Exponential Models of Rankings , 2009, AISTATS.

[31]  Sofya Raskhodnikova,et al.  What Can We Learn Privately? , 2008, 2008 49th Annual IEEE Symposium on Foundations of Computer Science.

[32]  Martin J. Wainwright,et al.  Local Privacy, Data Processing Inequalities, and Statistical Minimax Rates , 2013, 1302.3203.

[33]  Thomas Steinke,et al.  Between Pure and Approximate Differential Privacy , 2015, J. Priv. Confidentiality.

[34]  Seth Neel,et al.  The Role of Interactivity in Local Differential Privacy , 2019, 2019 IEEE 60th Annual Symposium on Foundations of Computer Science (FOCS).

[35]  Alex Gershkov,et al.  Optimal Voting Rules , 2013 .

[36]  Anthony Wirth,et al.  Correlation Clustering , 2010, Encyclopedia of Machine Learning and Data Mining.

[37]  Gang Li,et al.  Private Rank Aggregation under Local Differential Privacy , 2019, Int. J. Intell. Syst..

[38]  John Duchi,et al.  Lower Bounds for Locally Private Estimation via Communication Complexity , 2019, COLT.

[39]  Gerome Miklau,et al.  Differentially Private Rank Aggregation , 2017, SDM.

[40]  Cynthia Dwork,et al.  Calibrating Noise to Sensitivity in Private Data Analysis , 2006, TCC.

[41]  Kunal Talwar,et al.  Mechanism Design via Differential Privacy , 2007, 48th Annual IEEE Symposium on Foundations of Computer Science (FOCS'07).

[42]  John G. Kemeny,et al.  Mathematical models in the social sciences , 1964 .