Domain-Specific Databases

The paper assigments are here.

Description:
Starting from the goals of business data processing, the design of database management systems have traditionally been guided by the need to ensure consistency and persistency of structured, scalar data in the presence of short-lived data processing tasks. While this goal led to a design that has served business applications extremely well for more than two decades, many data-intensive applications have different goals and requirements, and thus are not satisfactorily supported by the prevailing model.

As such, the past decade have seen proposals for new database architectures and languages designed to better meet the demands of various domains, such as scientific computing and web databases. Some of these proposals take the evolutionary approach and basically extend the traditional database model with additional capabilities, whereas others argue for specialized from-scratch designs. Despite their differences, the common underlying theme is to design solutions around fundamental data management principles, such as data independence, declarative programming, and top-down designs.

The goal is to review the recent trends in data management, revisit and critically reevaluate traditonal database assumptions and goals, and eventually make projections for the future of data management. To this end, the course will cover domain-specific data management solutions in several non-traditional application domains, including scientific computing, web computing, event stream processing, text and web databases, sensor and peer-to-peer networks, and personal information management.

The course will primarily consist of technical paper presentations and discussions by the participants. The course will also include a project component.