Homepage of Wentian Lu


Wentian Lu

Database and Information Management Lab
School of Computer Science
University of Massachusetts Amherst

Computer Science Building

Now at Google Inc.

I got my Ph.D degree from the School of Computer Science at the University of Massachusetts Amherst. I was a member of the Database and Information Management Lab. My advisor is Prof. Gerome Miklau. Prior to joining UMass, I received my bachelor's and master's degrees in computer science from Nanjing University.

I am generally interested in data management and database systems, including but not limited to, large scale data processing, privacy in big data, social network analysis and cloud computing.


The effective analysis of social networks and graph-structured data is often limited by the privacy concerns of individuals whose data make up these networks. Differential privacy is a rigorous privacy model that offers individuals an appealing guarantee of privacy. While differentially private algorithms for computing basic graph properties have been proposed, graph modeling tasks that are common to the community can not yet be carried out privately.

In this project, we present algorithms and tools for modeling social network privately, such as the exponential random graph model.


Evaluating the performance of database systems is crucial when database vendors or researchers are developing new technologies. This applies broadly to new storage architectures, new query optimization strategies, new physical or logical designs, new algorithms for automated index selection, etc. Unfortunately, the actual data is often unavailable to the evaluator because privacy, security, and competitiveness concerns prevent the enterprise from releasing their data.

In this project, we present algorithms for the release of a synthetic database which accurately models performance properties of the original database, under differential privacy.


The tension between audit analyses and retention restrictions is present in a broad range of industries where sensitive records are managed, including financial services, healthcare, insurance, technology, education, telecommunications and and others. Currently, data owners have to carefully balance the need for policy compliance with the goal of accurate auditing.

In this project, we present a framework and tools for auditing the changes to a database system in the presence of retention restrictions, which includes a historical data model that supports flexible audit queries, along with a language for retention policies specification.


In provenance system, both data and their traces are recorded. When designing a protection mechanism for such systems, traditional access control is not enough. As traces are essentially metadata of data, which further define the relationship among data, they need a unified solution with abilities of handling such correlations.

In this project, we model provenance as a graph-based structure and propose a framework allowing users to define sensitive provenance and then transform provenance graph. The transformation would serve as a view of original graph where the query on original provenance system will be evaluated against the view.

CMPSCI 445: Information Systems. UMass Amherst.

CMPSCI 645: Database Design & Implementation. UMass Amherst.

Auditing a database under retention policies
Wentian Lu, Gerome Miklau, Neil Immerman.
VLDB Journal

Exponential Random Graph Estimation under Differential Privacy
Wentian Lu, Gerome Miklau.
ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD)

Generating Private Synthetic Databases for Untrusted System Evaluation
Wentian Lu, Gerome Miklau, Vani Gupta.
International Conference on Data Engineering (ICDE)

Auditing a Database under Retention Restrictions
Wentian Lu, Gerome Miklau.
International Conference on Data Engineering (ICDE)

AuditGuard: a system for database auditing under retention restrictions
Wentian Lu, Gerome Miklau.
Proceedings of the VLDB (PVLDB)

Log Sanitization: Auditing a Database Under Retention Restrictions
Wentian Lu, Gerome Miklau.
University of Massachusetts, Amherst. Technical Report

How and how much should we disclose?
Kai Dong, Wentian Lu, Zhen Qin, Yu Huang, Xianping Tao, Jian Lu
3rd International Conference on Pervasive Computing and Applications(ICPCA)

Shadow: A Middleware in Pervasive Computing Environment
for User Controllable Privacy Protection

Wentian Lu, Jun Li, XianPing Tao, Xiaoxing Ma, Jian Lu.
Smart Sensing and Context. First European Conference, EuroSSC

Componentized Software Service and its Application in ARTEMIS-ARC
Wentian Lu, Ping Yu, Xiaoxing Ma, Xianping Tao, Jian Lu.
Application Research of Computers. (abstract)

On Research and Development of Privacy Protection in Pervasive Computing
Wentian Lu, Xianping Tao, Jian Lu.
Proceedings of the 2nd Chinese Conference on Pervasive Computing. (abstract)