Ch. 7
Building Data-Driven Decision Support Systems

In a Data-Driven DSS it is important to develop and maintain metadata about the DSS data. We have data dictionaries for transaction systems, but because DSS data may come from many sources, creating a dictionary and metadata is especially important for a DSS. Also, we need integration of DSS data that comes from different sources. The data dictionary provides a reference about how we have combined data from various data sources.

Metadata is defined as "data about the data" in a DSS database. It provides a directory to help the Decision Support System’s Database Management System locate the contents in a data warehouse or data store. Metadata is a guide to mapping data as it is transformed from the operating environment to the data warehouse environment; and it serves as a guide to the algorithms used for summarization of current detailed data. Metadata is semantic information associated with a given variable. This type of data must include business definitions of the data and clear, accurate descriptions of data types, potential values, the original source system, data formats, and other characteristics.

Metadata resources include database catalogs and data dictionaries. It also includes the names of variables, length of fields, valid values, and descriptions of data elements. The semantic data is often stored in a data dictionary. Metadata insulates a data warehouse or database from changes in the schema of source systems.

Data-Driven DSS must have high quality data; inaccurate data can result in bad decisions. High quality data is accurate, timely, meaningful, and complete. Assessing or measuring the quality of source is a preliminary task associated with evaluating the feasibility of a Data-Driven DSS project.

The above comparison of DSS data and operating data suggests some architectural issues related to building a Data-Driven DSS. Let’s try to address the needed DSS software architecture more systematically.