W3C home > Mailing lists > Public > public-rdb2rdf-wg@w3.org > May 2010

Review of UC Document

From: Juan Sequeda <juanfederico@gmail.com>
Date: Sun, 9 May 2010 23:42:44 -0500
Message-ID: <m2tf914914c1005092142u126e3b16h6e613398e16f70ca@mail.gmail.com>
To: public-rdb2rdf-wg@w3.org
Hi all

Here are my comments

Use Cases

   - UC1 presents 6 tables and the equivalent RDF. Do we really need all of
   that. Each additional table is not bringing anything new to the table IMO. I
   would suggest to eliminate some of the tables.
   - I don't think that we should show any SQL, as part of UC1. This will
   through people off and start thinking how the SQL was produced and why it
   was produced that way. That is not the intention of this document. I suggest
   to eliminate any SQL that is shown
   - UC3 has several titles (responsible, goal, problem, req, use case
   description, example). I honestly like this break down and would advocate
   that each use case be presented this way. However, I don't know if this is
   normal for W3C use cases. Anyways, I believe UC3 needs to be cleaned up a
   bit.
   - I believe that UC5 and UC6 as they stand are not usecases, but they are
   important because they show a requirement that I like (UC6). I recommend
   that we combine these two give the use case by showing a relational schema
   of a school shown in the image that I drew and how it would be mapped to a
   domain ontology [1]. Something like: "SemantEducaTrix, the most recent
   Semantic Web company to burst into the educational software market, is
   mapping the school system's relational database to RDF / SPARQL and their
   own School Domain Ontology".

Requirements

As explained in my previous email responding to Lee, I'm presenting the 3
different RDB2RDF options that I understand exist [1]

   - 3.1.1 represents Option 1. I strongly suggest that we change the title
   of this section to Direct Mapping and use the word Local Ontology when
   possible. We can show that a Relational Database can be directly mapped to
   RDF by generating a Local Ontology from the relational schema, and the
   relational data will become RDF instances of the Local Ontology that was
   generated. This is completely isomorphic.
   - 3.1.2 is a bit confusion. The title says graph transformations but in
   the first sentence it talks about ontologies. I strongly suggest that we use
   the word Domain Ontology in this section. Using the word "Shape" makes no
   sense to me. If we used the word non-isomorphic in the title here, shouldn't
   we also use it as the title for 3.1.1. In other words, the wording is very
   confusing. I would simply call this RDB to Ontology Mapping (or something
   like this). Option 2 and 3 of my images kind of come together in this
   requirement. Option 2 shows the case when there is no local ontology
   created, therefore the req 3.1.1 is not a subset of this one. On the other
   hand, Option 3 builds on the case that we need req 3.1.1... but then it
   would mean that we are mapping two ontologies: the local and domain.... and
   this is out of the scope. Souri was against this, and I also believe that we
   shouldn't get in this terrain. However, I put this out there because this is
   what I believe are our current options.

Do people agree with me on these options? Are there more? Am I wrong? Which
options are we considering? Only 1 and 2?

We don't need Section 3.3, do we?


[1] http://userweb.cs.utexas.edu/~jsequeda/rdb2rdf/RDB2RDF_Option_2.jpg
[2] http://userweb.cs.utexas.edu/~jsequeda/rdb2rdf/

Juan Sequeda
+1-575-SEQ-UEDA
www.juansequeda.com
Received on Monday, 10 May 2010 04:43:20 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 10 May 2010 04:43:20 GMT