Incremental database systems: databases from the ground up

This paper discusses a new approach to database management systems that is better suited to a wide class of new applications such as scientific, hypermedia, and financial applications. These applications are characterized by their need to store large amounts of raw, unstructured data. Our premise is that, in these situations, database systems need a way to store data without imposing a schema, and a way to provide a schema incrementally as we process the data. This requires that the raw data be mapped in complex ways to an evolving schema.