- From: Eric Prud'hommeaux <eric@w3.org>
- Date: Tue, 6 Apr 2010 10:46:42 -0400
- To: Daniel Miranker <miranker@cs.utexas.edu>
- Cc: public-rdb2rdf-wg@w3.org
* Daniel Miranker <miranker@cs.utexas.edu> [2010-04-05 21:08-0500] > Dear Gang, > > Attached you will find material I'm hoping will serve as a starting > point for structuring and enabling > our case discussion. I view this as brain storming mode. There > are pieces I'm not happy with and similarly > choices of terminology. I'm eager to get the ball rolling. > > (Yes, this should all be on the wiki, but I don't have write > privileges yet.) > > You will find: > > 1) A word file, Taxonomy of Use Cases.doc. > > In addition to the obvious content from its name, is also an outline > of what we might include in the > description of use cases. I've included that in-line for email discussion: Use Case Content 1.Classified per the taxonomy 2.Descriptive text 3.Logical Model (UML) 4.Physical Model (DDL) 5.Sample database content 6.Intended RDF, RDFS and OWL results 7.Pointer to domain ontology (as appropriate) Taxonomy of Use Cases 1.Role of Ontology a.Domain Ontology Mapping problem includes, a priori, a federating ontology. (To discuss: Are there other roles an existing domain ontology might play?) b.Putative Mapping problem is not particular about an a posteri ontology. Ontology is created from the RDB. (To discuss, this may include, or we may list as a different subcase, “easy” renaming/assignment of labels to the output. Either way, I advocate a distinct case when there is no existing ontology) c.Federating Mapping problem includes forming the federating ontology in the course of RDF enabling the databases. i.e. schema integration of existing databases is handled in the course of completing the use case; similar to creating an Enterprise Schema at the beginning of a data warehousing effort. (of similar text for data marts) 2.Integration Goals a.Source RDB structure at issue i.Structured Consider only highly structured database content, and treat string/text fields as atomic data types of secondary interest. ii.Structured + Semistructured Text fields in the database are elemental. iii.Structured + Tagged Text Text fields in the database are parsed and tagged per an existing ontology. b.Other data source i.Structured ii.Structured + Semistructured iii.Structured + Tagged Text iv.Mash-up RDB2RDF output commingled with other semantic data sources, but no detailed federated data operations needed 3.Motivation a.Derived from real-world scenario b.Derived from hypothetical scenario c.Small example intended to focus attention on particular issues. I propose adding: 4.Expressivity a.Node Label Generation Graph node names are synthesized from a function of database attributes b.Datatype expression 1.Simple Relational data (cells) are mapped to rdf datatypes per SQL XSD mapping. 2.Micorparsing Relational data are parsed and mapped to rdf graphs. > 2) An Excel file, UseCaseCategorizationNtodo.xls > > Juan and I have made a first stab at assigning individual use cases > to the taxonomy. Also, a tabulation of what is > already present on the wiki per, what me might include in the > description of use cases. excellent, thanks a million! > 3 - 6) For the four use cases that I could locate DDL, I ran the DDL > through a UML reverse engineer. Here's the DDL for UC1: http://www.w3.org/2008/04/DiabeticPatientsDataSet/CreateDB.sql > Sincerely, > > Dan -- -ericP
Received on Tuesday, 6 April 2010 14:47:17 UTC