Transcript BoeingW3CPresentation

Enabling Semantic Access to
Enterprise RDB Data
Jun YUAN
David H. JONES
Mathematics & Computing Technology
Boeing Phantom Works
BOEING is a trademark of Boeing Management Company.
Copyright © 2007 Boeing. All rights reserved.
Motivation
Boeing Technology | Phantom Works
Semantic
Mappings
Semantic
Query
Generator
Information
consumer
E&IT | Mathematics and Computing Technology
•
Advanced knowledge about database
technology is needed to query a DB.
•
•
•
Semantic Query
Translator
•
Post-Query
Processing
What if a database schema changes?
•
Query Interface
•
Database
•
Schema 1.0
Schema 2.0
Schema 1.1
Schema 2.1
Schema N
Copyright © 2007 Boeing. All rights reserved.
•
Understanding schemata.
Knowing query language syntax.
…
Many changes are not driven by
semantics.
–
–
–
–
–
Normalization or De-normalization.
Using different DBMS products.
Change of data types.
Stored procedures vs. Views.
…
Query statements usually have to be
modified accordingly.
About query answers
•
•
•
A table of rows.
Ever-growing information sharing.
Is retrieved information semantically
understandable?
What can we do better?
Semantic Query Layer
Boeing Technology | Phantom Works
Semantic
Mappings
Semantic
Query
Generator
Information
consumer
Semantic Query
Translator
Post-Query
Processing
Query Interface
Database
Schema 1.0
Schema 2.0
Schema 1.1
Schema 2.1
Schema N
Copyright © 2007 Boeing. All rights reserved.
E&IT | Mathematics and Computing Technology
• Semantic Independence.
• Isolate semantics from the underlying nonsemantics things in DB technology.
• Present information consumer with the semantic
part only.
• Benefits
• Semantics of data is explicitly presented by
ontology.
• Query against ontology directly.
• Semantics of query answers is explicitly
presented by ontology.
• Non-semantic changes are hidden from end users.
• …
Semantics Models in Database Technology
Boeing Technology | Phantom Works
E&IT | Mathematics and Computing Technology
• Semantic model is commonly used in DB design.
• Three-level DB design
• Conceptual level model (ER Diagram) is a kind of semantic
model.
• When and why explicit semantics gets embedded?
•
•
•
•
From conceptual model to logical model.
Schema normalization.
No explicit relationships.
…
• Explicit semantic representation was there, but
disappeared later in the process.
• Schema re-engineering is a practice to recover such
explicit semantics.
Copyright © 2007 Boeing. All rights reserved.
Some Related Work in the Past
Boeing Technology | Phantom Works
E&IT | Mathematics and Computing Technology
• Object-oriented query language, e.g. OQL.
• A navigational query paradigm
• Exciting features: complex objects, polymorphism, operation
invocation, late binding, …
• Rule based query language, e.g. Datalog.
• Powerful expressiveness, but performance is an issue.
• Solid research results on mappings between FOL and
Relational calculus.
• Solid research results on query optimization
• Not trying to replace RDB, but leverage with it
Copyright © 2007 Boeing. All rights reserved.
More Challenges
Boeing Technology | Phantom Works
E&IT | Mathematics and Computing Technology
• Expressiveness of query specification.
• Trade-off between the power of expressiveness and
performance.
• Mappings
• Ontology & underlying data model.
• Semantic query primitives & relational query primitives.
• Query answer transformation.
• How to formulate URIs -- Entity Resolution
• A lot of intermediate data needs to be preserved.
Performance
• Maximize the efficiency of relational query engine
– Push down query as much as possible.
• Efficient post-query processing.
Copyright © 2007 Boeing. All rights reserved.