W3C home > Mailing lists > Public > semantic-web@w3.org > November 2005

Four schema creation questions

From: David Pratt <fairwinds@eastlink.ca>
Date: Mon, 21 Nov 2005 00:33:28 -0400
To: Semantic Web <semantic-web@w3.org>
Message-id: <F68D1DDD-5A47-11DA-ADD9-000A27B3B070@eastlink.ca>

Hi.  I have been working on a schema for a while. I think the biggest 
challenge is really distilling what  I want to have in the schema and 
in my records and how explicit to be. I had started to put things into 
my triples store I had put together based on rdflb - but based on what 
I have so far but am realizing that there are some things that could 
likely be better. The organization of data could also be better by 
using bag, seq or alt which I am not using at this point. I had heard 
they were a bit difficult to work with so had initially stayed away 
from these structures. My questions follow:

1. Combining schemas:
I have tried to create a schema that is pretty stand-alone (except for 
rdf namespaces) but there are some properties that are available in 
dublin core. Should I drop these properties in favor of dublin core and 
add its namespace or create a reference to dublin core, adding its 
namespace, since these properties are pretty much equivalent. What are 
my alternatives?

2. When to group
I am hoping someone can offer constructive comment on when to group rdf 
data with bag / seq or alt.
As an example, I am using an author property and it is significant that 
the primary author is listed first if there is more than one. It seems 
like a good candidate for a seq but often there is only a single 
author. Another example is keywords. I don't think dublin core puts 
these in a bag for example but it would seem like a good use for a bag 
with list element.  So the question is really when to use a bag and 
when to use a property on its own where it can exist as more than one 
instance in a record.

3. Use of Language attribute
I am planning on having multilingual data but likely 80% of all data in 
the store will likely be in english.  Almost all the rdf I have looked 
at thus far, I do not see very high use of language attributes in rdf. 
Maybe it is simply because english is used much of the time. Can 
someone provide a practical recommendation on use of language 
attribute. I was thinking I ought to be explicit with pretty much any 
literal that is unicode.

4. Datatype attribute
I am looking for any recommendations based on experience with the 
datatype attribute.  My intention is to be more (as opposed to less) 
specific about the type of data I will expect from the records since 
records will be validated before being inserted into the triples store. 
I have been looking at Sparta since it seems to try and match python 
data types with rdf types. Just looking for any kind of advise from 
anyone that has worked with a programming language and rdf types to 
offer any general recommendations based on their experience.

Many thanks
David
Received on Monday, 21 November 2005 04:37:51 UTC

This archive was generated by hypermail 2.4.0 : Thursday, 24 March 2022 20:40:57 UTC