Critical Issues of XML Schema & RDF & Suggested Solutions Peter P. Chen Bootstrap Institute & Louisiana State University
OUTLINE General Comments Critical Issues of RDF Critical Issues of XML Schema Important Issues to be Studied How We Might Solve the Concerns of Microsoft and Others
General Comments Some of the arguments are –Not valid (sometimes) –Divert our attention to our important goal: “Semantic Web” Example: –“Data vs. Meta Data”: Actually, a data model can be used to model both “data” and “meta data.” The distinction may be useful for short-term marketing purposes,but we need to study the long-term implications
Critical Issues of RDF Strength: It can be viewed as –A binary directional Entity-Relationship (ER) model –More powerful than pure XML “tree structures” Weaknesses of Current “Specifications”: –The use of the concept of “URI” as an “entity” is sometimes confusing and not always consistent –Multi-level specifications seems to be not very clean –“Bag” and several other concepts are not well defined –Multiple “graphs” for same “semantics/statement,” and, possibly multiple “semantics/statements” for the same graph. –The “RDF schema” specifications are unnecessarily complicated
Critical Issues of XML Schema In general, the “syntax” part of XML Schema is simpler than that of RDF. Weaknesses/Needs of Current XML Schema specifications: –Mixtures of document world terminology and data world terminology (perhaps, too much on some of the unnecessary legacy) –Needs a growth path to full ER modeling Capabilities including facilities to capture RDF semantics
Important Issues to be Studied Data Model for the Semantic Web –An architectural plan of data models (including Modifications of existing models A new XML data model? An ERM for XLink? Multi-level of data models –Cleaning up theoretical problems of current RDF model –Identification of mapping between ERM’s (for the near term, RDF) and XML tree structures –Design Mechanisms to capture missing semantic information from ERM’s (RDF) to XML tree structures Schema Issues –Simpler –Unified Syntax –Semantics Rich –Growth Plans for Full ERM capabilities
How the Suggested Solution Might Solve the Concerns of Microsoft and Others (A) Microsoft seems to recognize the value of using an ER model as RDF model is fine Suggested Solutions: –An architectural plan of data models including: Modifications of existing models A new XML data model? An ERM for XLink? Multi-level of data models –Cleaning up theoretical problems of current RDF model
How the Suggested Solution Might Solve the Concerns of Microsoft and Others (B) Microsoft seems to object to the RDF schema specifications that did not rely on existing XML infrastructures Suggested solutions: –simplified, unified, & semantically richer syntax –Specify mappings between RDF and XML Schema –Design Mechanisms to capture missing semantic information from ERM’s (RDF) to XML tree structures –Growth Plans for Full ERM capabilities