Traditionally, databases are reluctant to accept external user specified constraints to avoid security breaches such as SQL injections and unwanted disclosures. Web interfaces to deep web databases thus do not allow open ended queries, and often hide the underlying database schema as a disclosure prevention mechanism. However, an increasing number of applications are being designed by directly composing web interfaces to extract contents from multiple deep web databases. The pre-fabricated nature of web interfaces forces an all-or-nothing paradigm in which users are required to supply all query constraints and accept all returned responses. Even if a subset of the computed response is of interest, the interfaces may return large tables, wasting computational resources and transmission time.
In this presentation, we propose the concept of interface views as a mechanism to accept user constraints that can be pushed inside the query evaluation process in databases to avoid computing redundant answers without violating the "intention" of the query. We show that under our approach, accepting user constraints is safe and does not unintentionally divulge information by enforcing constraints "monotonically." We thereby provide the framework for constructing web interfaces to deep web databases that are far more flexible than current practice, and also potentially more efficient.
Biography:
Hasan Jamil is an Associate Professor in the Department of Computer Science, University of Idaho. He is a member of IEEE Computer Science, ACM SIGMOD and International Society for Computational Biology. He received his PhD in Computer Science from Concordia University, Canada, and his MS and BS in Applied Physics and Electronics from the University of Dhaka, Bangladesh. Before joining University of Idaho, he also was a faculty in the Department of Computer Science, Wayne State University, Mississippi State University, and Macquarie University, Australia. His current research interests are in the areas of databases, bioinformatics and knowledge representation. He is developing database technologies to change the way we access, trust, analyze and manage large distributed data and the science that depends on it. In particular, he is interested in the management and querying of interaction and gene expression data, and their application in disease gene prioritization and unfolded protein response in oligodendrocyte. His research in the US has been supported by grants from National Science Foundation, National Institutes of Health, and US Department of Agriculture, and in Australia by Australian Research Council.