Database systems have become to require to support time dimension for its new application areas which need not only current data but also the previous instances of the data (history data).
This thesis presents the design considerations to incorporate time dimension support into DBMS as a normal database function. They include how to store history data and how to manage its state. The implementation structure for existing DBMS to support time dimension is proposed and the semantics of database operations in its structure is discussed. And this time support is implemented on PERD (Prolog-based Extended Relational DBMS) in C-prolog language.
