My research is contextually situated in two complementary arenas: (1) databases in the context of the internet; and (2) data management for biomedicine. What happens when you can connect many different databases that are autonomously organized and managed? How can you query and integrate and analyze all the information that you potentially have access to? How can one represent, and account for, variations in information structure and in information quality? How can a non-technical user be supported in performing these tasks?

In prior research at Michigan, we have built a native XML store -- a hierarchical database from the ground up for storing and querying XML data. We call this project TIMBER . We used such an infrastructure in a variety of application domains with particular data management needs not easily met through traditional databases, including e-commerce, (automotive) engineering design, product lifecycle management (PLM), biological information, and health care.

As a demonstration of our ability to integrate a wide variety of information, we have constructed the Michigan Molecular Interactions Database (MiMI). MiMI is part of the the National Center for Integrative Biomedical Informatics, one of 7 national centers for Biomedical IT set up by the NIH. I am a Senior Scientific Director of this center, which develops techniques to compose information and software tools to facilitate biomedical research into diseases such as cancer and diabetes.

I co-organized a workshop on data management issues for (molecular and cell) biology. Associated with this, I guest edited a special issue, devoted to data management for biology, of the OMICS journal on integrative biology. I was (co-)Program Chair for the ISMB Conference 2005.

A crucial question that runs through all of my research described above is how to build database systems and query models so that they are truly usable? Usability, for me, is not just a question of having a well-designed user interface: it has to be designed into the system from the beginning. For instance, a precise representation with a very complex schema is not useful because most users will not master the schema enough to ask the precise questions they could have asked. In this context, I am thinking about topics such as data modeling, schema design, schema summarization, form generation, and natural language querying. I gave a keynote speech at SIGMOD on database usability .

I have been concerned about how academic scholarship is demonstrated, and have been involved with several efforts in this direction. I established the ACM SIGMOD Digital Review, which provides online reviews of published articles, now migrated to Pubzone. As a complementary effort, I serve as the editor for the datbase section of the Computing Research Repository (CoRR) , which encourages publication prior to review. I am the founder of the Proceedings of the Very Large Database Endowment (PVLDB), which is an effort to bring journal-style reviewing to a prestigious conference.