Statement – Alan Halverson

We live in an increasingly connected world. Many people have high speed, always-on Internet access at their homes, and Google is a verb in nearly every language on the planet. Our access to multiple streams of information in digital formats has the potential to keep us in touch with world events and far-away loved ones. We have access to this information on multiple devices, such as wired desktop computers at home and work, and wireless laptop computers, PDAs and cell phones. The promise of these connectivity options is simple enough – to enable access to information when and where we need it. While this may be true, I see two additional trends developing. First, high-speed access to data and information has not reduced how much time people spend using the Internet, but rather has allowed them to view more information in the same amount of time. Second, tools for helping users manage interconnected streams of information have lagged behind access to the information itself. A modern web browser provides a history of sites visited, but the list is flat and provides only a day-by-day grouping of sites. Missing is the hierarchical discovery of the information and the information itself. The Internet is a hierarchy of links and information, and our minds work in a similar fashion. While Google is a good model for cutting across hierarchy to find information quickly, the flat result list model does not match how we discover and remember information. In fact, hierachical data objects are pervasive in information management – a customer and their orders, a grouping of sales figures by continent, country, and region, a computer filesystem, and so on. Developing techniques for efficiently managing information hierarchies is critical to bridging this gap. My research interests concern optimizations for storage and query processing of hierarchically organized data. I have worked on three projects related to my interests while a graduate student: a novel mixture of structureand navigation-based approaches for evaluating path expressions in a native XML query system, a system for processing relational queries over XML data, and a comparison of storage optimization techniques for relational schemas in the context of row-oriented and column-oriented storage for a read-mostly query workload. The next three sections briefly describe each project. I conclude with a description of nearand long-term plans for future work.