From semistructured data to XML: Migrating the Lore data model and query language
Roy Goldman, Jason McHugh, Jennifer Widom · Markup Languages Theory and Practice · 2000
Research on semistructured data over the last several years has focused on data models, query languages, and systems where the database is modeled as some form of labeled, directed graph [Abi97, Bun97]. The recent emergence of eXtensible Markup Language (XML) as a new standard for data representation and exchange on the World-Wide Web has drawn significant attention [BPSM98]. Researchers have casually observed a striking similarity between semistructured data models and XML. While similarities do abound, some key differences dictate changes to any existing data model, query language, or DBMS for semistructured data in order to fully support XML. This paper describes our experiences migrating the Lore database management system for semistructured data [MAG+97] to work with XML. We present our modified data model, whose definition was a subtly challenging task given that XML itself is just a textual language. Based on this model, we describe changes to Lorel, Lore’s query language. We also briefly discuss changes to Lore’s dynamic structural summaries (DataGuides [GW97]) and the relationship of DataGuides to XML’s Document Type Definitions (DTDs). 1 OEM and Lorel Lore is a complete database management system designed specifically to handle semistructured data [MAG+97]. Lore’s original data model, OEM (for Object Exchange Model), is a simple, self-describing, nested object model that can intuitively be thought of as a labeled, directed graph [PGMW95]. In OEM all entities are objects that can either be atomic or complex. Each object has a unique object identifier (oid). Atomic objects contain a value from one of the atomic types, e.g., integer,