| There are several limitations of the standard relational database model in its ability to handle uncertainty and incompleteness in data. Rough set theory, usually associated with knowledge discovery in databases, can be used to overcome some of these difficulties. This paper introduces several techniques for rough set uncertainty management in relational databases and discusses relationships between the various methods.; One technique involves the incorporation of rough sets directly into the relational database model, resulting in an extended model. Rough relational operators are defined and the properties of these operators discussed. They are compared with operators of the relational model and with rough sets. Fuzzy sets are then used to define fuzzy rough sets and a similar database model based on fuzzy rough relations.; Rough set techniques are also used for the querying of crisp data in an ordinary relational database. This approach is different from the rough and fuzzy rough relational database models in that the underlying data and relational model are unchanged. RSQL, a rough structured query language is developed for the rough querying of crisp data and the semantics of all the operations defined. Differences between the approaches in this research and related approaches are then discussed.; Finally, because rough sets are typically associated with the discovery of dependencies and "knowledge" in data, rough functional dependencies are defined for the rough data models and relationship of the approaches of this research to knowledge discovery in databases is discussed. |
