A Data Model for Semi-structured Data

Database Management Systems (DBMS) are software systems that implement different data structuring methods in practice. Formally, a Data Model defines data structuring facilities and operations that can be applied to such structured data. Potentially, databases can operate with well-structured data or semi-structured data. In the case of well-structured data, database structure can be separated from actual information content. Such a generic database structure is defined as a so-called database schema. Thus, the database can be seen as just instances of data types predefined in the database schema. A typical sample of well-structured data is the relational data model. Semi-structured data are kept as software repositories that contain both data structures and actual content. A typical sample of such semi-structured data repository is a collection of WWW documents along with navigable links. This paper proposes a method of structuring data repositories suitable for the development of semi-structured databases. The method is based on the concept of so-called structured collection (S-Collections). An S-Collection is an object that encapsulates other data objects and relationships between them. Database may be seen as a set of S-Collections. S-Collections may contain other S-Collections even recursively. Developing a database can be seen as creating individual data objects, combining the objects into S-Collections and inclusion of S-Collections into other S-Collections. Editing operations are always applied to a particular data object (S-collection), what presents a mechanism for supporting the logical integrity of complex semi-structured databases.