| Summary: | 碩士 === 義守大學 === 資訊工程學系 === 89 === The goal of this research is to study how to improve the incremental discoveries of functional dependencies (FD) from crisp and fuzzy relational databases. Functional dependencies are relationship between attributes of a database relation. Discovery of FD’s from relational databases can be applied to database design, database reverse engineering, query optimization, and database security. It has been identified as an important database analysis technique and received considerable research interests in recent years. However, most studies emphasize on non-incremental searching techniques of FD’s from static databases. Few have considered the incremental searching techniques that are required when the database is updated incrementally. More researches are needed to improve current incremental searching techniques to deal with dynamic database behaviors in the real world.
In this research, we proposed three efficient incremental searching algorithms of functional dependencies for crisp relational databases and similarity-based fuzzy relational databases, when a set of tuples is added to the database. Algorithm I is a partition-based incremental searching algorithm on crisp relational database. Algorithm II is a pairwise-comparison-based incremental searching algorithm on similarity-based fuzzy relational database. Algorithm III is a partition-based incremental searching algorithm on similarity-based fuzzy relational database. We also present the complexity analysis and experimental results of these three algorithms. The results show that the proposed algorithms could find the minimal cover of functional dependencies efficiently.
|
|---|
