Database Reference
In-Depth Information
a set of data, as well as the active and dynamic data structure of the legacy data models,
with the constraints structures to resolve their synonyms and homonyms confl icts. These
constraints include integrity constraint enforcement, derived data maintenance, triggers,
protection, version control, etc. The frame model unifi es data and rules, allowing these
advanced features to be implemented effectively.
The frame model performs in the object-oriented paradigm. All the conceptual entities
are modeled as objects. The same attribute and behavior objects are classifi ed as a class.
Besides, both facts and rules are viewed as objects in the frame model design. The frame
model logical schema in a class format is shown in Figure 3 (Fong & Huang, 1997).
The frame model consists of two classes: static classes and active (dynamic) classes.
Static classes represent factual data entities and active classes represent rule entities. An
active class is event driven, obtaining data from the database when it is invoked by a certain
event. The static class stores data in its own database. The two classes use the same structure.
Combining these two types of objects within the inheritance hierarchy structure enables the
frame model to represent hybrid knowledge.
Fong and Huang (1997) translated existing data models into a frame model of the
universal database. The structure of the frame model consisted of several classes such as
Header, Attributes, Methods, and Constraints classes. According to the frame model, a
universal database could be formed. Therefore, old and new database systems could coexist
to form a data warehouse for a decision support system.
Fong and Huang (1999) investigated architecture of universal data warehousing for the
connectivity of relational and OO data model using an ORDBMS. A frame model metadata
was chosen to represent the conceptual and logical schema of the universal data warehouse,
which structures an application domain into classes, and its data in relational tables. The
universal data warehouse, using an ORDBMS, offers a relational and an OO view for the
data warehouse to accommodate different types of queries effi ciently. Fong & Pang (1999)
proposed a frame metadata model approach to integrate existing databases and evolve them
to support new database applications. This facilitates an evolutionary approach to integrating
existing databases to support new applications.
Data Warehousing and Star Schema
A data warehouse is a database specifi cally created to facilitate decision-making. A
data warehouse retrieves data from operational and Online Transaction Processing (OLTP)
system, but the data are transformed and optimized for analysis.
Nowadays, the demand for information continues to increase as companies realize that
information generates revenues, reduces cost and enlarges market shares. Keen competition
in rapidly changing business environments is expected and these conditions will generate
increasing demand for reliable, easy-to-access decision-making information.
A star schema is a simple structure with relatively few tables and well-defi ned join
paths. This design provides fast query response time and a simple schema that is understood
by the analysts and end users. A star schema contains two types of tables: fact tables and
dimension tables. Fact tables contain the quantitative or factual data about a business, the
information being queried. This information is often numerical measurements and consists
of many columns and millions of rows. Dimension tables are usually small verse fact tables
and contain more descriptive information. Dimension tables contain the data needed to place
transactions along a particular dimension.
