Browsing by Author "Qi HE"

Now showing 1 - 4 of 4
  • No Thumbnail Available
    Item
    Extending and Inferring Functional Dependencies in Schema Transformations: Extended Version
    (2004-03-01T00:00:00Z) Qi HE; Tok Wang LING
    Functional dependency (FD) plays an important role in individual databases. In this paper, we study FDs in the context of multi-database interoperability. A major challenge in the integration of heterogeneous database schemas is of schematic discrepancies, when the data (values) of one database correspond to metadata (schema labels) of another. We first study the schematic discrepant transformations, i.e., transformations between schematic discrepant schemas. We then define "restricted FD", an extension to conventional FD, to formalize some class of constraints in schematic discrepant databases, and give a complete set of inference rules of restricted FDs. Then we study the propagation of restricted FDs during schematic discrepant transformations. Algorithms are proposed to derive all the restricted FDs in transformed schemas from restricted FDs in original schemas. At last, we show some applications of restricted FDs in the context of multi-database interoperability: (1) use FDs to verify whether a SchemaSQL view is well-defined, (2) use FDs to normalize transformed(integrated) schemas, and so on.
  • No Thumbnail Available
    Item
    Inferring and Applying Functional Dependencies in Schematic Discrepant Transformations
    (2003-06-01T00:00:00Z) Tok Wang LING; Qi HE
    In relational model, schematic discrepancy occurs when the same information is modeled differently as attribute values, attribute names, relation names or database names in different schemas. Originally raised in schema integration, people have identified many applications of it recently. This paper focuses on the inference and use of functional dependency (FD) constraints in the transformations among schematic discrepant schemas. We first study restructur-ing operators which are used to implement schematic discrepant transformations. Specifically, we study the reconstructibility and commutativity of restructuring operators, which can be used to simplify a transformation. Then to infer FDs in transformed relations, we propose restricted FDs to represent integrity con-straints in original relations. We also study the properties on how restricted FDs change when applying each kind of restructuring operators. Then we give an al-gorithm to infer FDs in schematic discrepant transformations. Our algorithm can compute all FDs in transformed relations which can be inferred from restricted FDs in original relations. At last, we identify the use of FDs in 3 scenarios: (1) use FDs to judge the correctness of SchemaSQL views; (2) use FDs to normalize a transformed relation; (3) use FD constraints to verify the integrity of data from different sources.
  • No Thumbnail Available
    Item
    Resolving Schematic Discrepancy in the Integration of ER Schemas
    (2003-06-01T00:00:00Z) Qi HE; Tok Wang LING
    n this work, we address the schematic discrepancy problem in the integration of ER schemas. Schematic discrepancy occurs when same information is modeled as attribute values, object type (entity type or relationship set) names, or attribute names in different ER schemas of databases. We first define the concepts of local/global attributes and identifiers of entity types and relationship sets resp in an ER schema, which provide essential semantics to resolve schematic discrepancy. Then based on those concepts, we give algorithms to resolve schematic discrepancy among ER schemas to be integrated.
  • No Thumbnail Available
    Item
    A semantic approach to XML schema integration
    (2004-09-01T00:00:00Z) Qi HE; Tok Wang LING
    In this paper, we adopt a semantic rich model, Object-Relationship-Attribute model for Semi-Structured data (or ORASS) to represent XML schemas. A challenge in XML schema integration comes from the hierarchical structure of XML. For example, two sets of XML elements from two sources may constitute the same relationship type, but in different hierarchies. Then in the integrated schema, we need decide a "good" hierarchy of these elements. In general, we require an integrated schema preserves the semantics of source schemas, has minimum redundancy and leads to low cost data transformation. Guided by these criteria, we developed algorithms to merge equivalent elements and equivalent relationship types among elements from source schemas, and proposed a top-down approach to integrate ORASS schemas, meeting challenges caused by the hierarchical structure of XML.