This is Jian Xu's web page at Computer Science Dept. UBC Information about my academic life is updated here. Currently you can find my resume and publications

I am a final year PhD student in CS-UBC and will graduate Fall 2011, I am currently working in Data Management and mining group. My supervisor is Prof Rachel Pottinger. Before I came to UBC, I studied in CS&E,University of New South Wales and got my master's degree.

I come from Suzhou China which is famous for its classical gardens. She is a very beautiful small town and you can find some pictures of her in the Gallary.

I am interested in many aspects of Computer Science. Currently I am focusing on topics of data integration. When I was in Australia, I worked on stream processing. Click on them to get more detailed information.

I worked in the JIIRP project where I helped researchers to manage and integrate data for disaster management and response. The project also motivates my research on semantic integration of domain heterogenous databases and especially query answering issues involved.

Take a look at the publications part for my writings and papers on these topics.

Data integration is the process to bring data managed in heterogeneous databases together and enable query answering over them. It distinguishes itself from the traditional database merging approach that one creates a schema and load data from various sources. Data integration systems avoid duplicating and reformatting data and allow data managed in different schemas remain in their forms of being managed. It builds schema mappings between likely distributed schemas and translate queries using these mappings. A query to the data integration system is written in one schema and is translated to queries written in other schemas in the system. Translated queries are passed to the corresponding data sources to be processed and answers are collected and returned.
My research area is relational peer data management (RPDMS). It is a very flexible data integration model in the big family of data integration architectures. Highlights for an RPDMS are : (1) the peer-to-peer network as physical networking layer allows flexible organization of data sources; (2) the system uses only pair-wise schema mappings for query rewriting, thus eliminates burdens of creating and maintaining federated schemas; (3) all peers in the system are created equally and so do they publish data sources on the semantic overlay network of the PDMS, which no centralized/crucial node or "hub" like node ensures maximal robustness.
The PDMS integration model is an especially good candidate for disaster management in which an integration system needs to be set up and deployed very quickly using very limited computational resources, yet allowing adding data sources at any time during disaster response missions.
I am facing new challenges not yet handled by existing approaches in relational PDMS research and aim to develop solutions to them. The concrete problems to solve are : (1) To optimize the topology of the semantic overlay network of a PDMS so that data sources belonging to different domains co-exist in one system minimally affect the query answering performance and ensure good scalability w.r.t. the number of data sources; (2) to support processing of object decomposition aggregates which is required in the affiliated JIIRP project. (3) to support continuous query processing and reasoning with source uncertainty. The prospective data integration system will extend existing PDMS systems and make it applicable for a much wider class of data integration scenarios, potentially making database technology more widely used and data sharing and reusing much easier.

This is for Stream processing. To be added.

• J.Xu, R. Potinger, The Decomposition Aggregation Query for Semantic Integration submitted to EDBT 2010
• J.Xu Thesis Proposal: On PDMS support of extended heterogeneity, aggregation and advanced query processingin CS dept. UBCpdf
• J.Xu, R. Potinger, Optimizing Acquaintance Selection in a PDMS in CS, UBC Technical report: TR-2007-18. September 2007 pdf
• X. LIN, J. Xu, Q. Zhang, H. Lu, J.X. Yu, X. Zhou, Y. Yuan, Approximate Processing of Massive Continuous Quantile Queries over High Speed Data Streams in TKDE Vol 18, 5 May 2006
• Y. Zhang, X. Lin, J. Xu, F. Korn, W. Wang Space-efficient Relative Error Order Sketch over Data Streams in ICDE2006 pdf
• Lin, H. Lu, J. Xu, and J.X. Yu Continuously Maintaining Quantile Summaries of the Most Recent N Elements over a Data Stream in ICDE, 2004 pdf
• J. Xu, X. Lin, Xf. Zhou. Space Efficient Quantile Summary for Constrained Sliding Windows on a Data Stream WAIM 2004