RE: [jena-dev] Newbie Design Question....

Hi,
I hope you don't mind, I'm forwarding your query to the Semantic Web Europe
list, as although Bali is a long way away, they're looking at use
cases/scenarios etc. My own comments below.


>-----Original Message-----
>From: Cokorda Raka Angga Jananuraga [mailto:don_raka@yahoo.com]
>Sent: 04 November 2002 08:19
>To: jena-dev@yahoogroups.com
>Subject: [jena-dev] Newbie Design Question....
>
>
>Hi,
>
>Allright, I've been able to use Jena for some simple stuffs in the
>tutorial :), MemModel, RDBModel (mysql, mm_generic).
>
>Now, I'm facing a design question. I have the following case:
>
>---
>I'm a journalist for "Fortunately" magazine, responsible for column
>"Life Science Advances". I'm now in the middle of preparing an article
>about "Biotechnology". So, from my home in Bali I browse the web,
>doing some desk research. I am reading a good article from "Scientific
>Earth" magazine, titled "Sheep Cloning in Ireland", when suddenly a
>memory about the term/idea/concept "genetic engineering" come up to
>surface. It leads me to another memory of an old friend of mine who is
>now working as lead researcher in "Bali Biotech Center"...., Prof.
>Wayan. He is widely known as an expert in "genetic engineering".
>So..., I decide to have an interview with him, next week.
>---
>
>I want to model that complex statement in RDF. In that model, I need
>to capture the background why the decision to interview Prof. Wayan is
>made. (In the above case: the reason to interview Prof. Wayan is
>because I know he's an expert in genetic engineering, a term which is
>[according to me] very relevant to the article which I'm reading now.
>Going further: I'm reading that article as part of my job to complete
>my next article for my magazine).
>
>So I break down that complex statement this way:
>
>[1] Statement1 = Prof. Wayan ----expertise-----> Genetic Engineering
>[2] Statement2 = Article "Sheep Cloning in Ireland" ----concept----->
>Genetic Engineering
>[3] Statement3 = Statement1 -----evoker-----> Statement2
>[4] Statement4 = Statement3 -----action-----> Appointment for
>Interview
>
>Well, I feel there's a good chance that the way I break down that
>complex statement is dead wrong. That's why I'm asking to this
>forum... for some enlightment ;) Where can I get article / guideline /
>tutorial / design patterns about modelling that kind of complex
>statement (or information modelling in RDF)?
>
>Thanks in advance,
>Cokorda Raka
>PS:
>[1] There's actually no such thing as "Bali Biotech Center". It's
>purely fictional :)
>[2] To Brian: I'll try to get BdbModel to work on my laptop tommorow,
>and write the instruction as I go.

The way you've approached looks reasonable, though I think the implicit
stage you've skipped here is probably pretty important, the identification
of the players in the statements and the identification of the roles. e.g.
Prof. Wayan is presumably a person, so finding an existing vocabulary that
includes this so you can qualify this part of the whole statement is
probably a useful first step. The same follows with the other entities -
e.g. genetic engineering might well be an Open Directory category. But the
process I'm applying there depends on my prior knowledge of the
vocabularies - in practice a repository like the MEG Registry, the DAML
ontology lists or the SWAG Dictionary should be helpful for tracking these
down. I think the abstraction technique you used on the properties is valid.
I guess if a search for terms matching 'evoker', 'action' etc proves
unrewarding then you will have to generalise this back some more (tricky
with 'action'!), or identify the domain ('processes' in this case?) and do a
search on that. If that still doesn't get you anywhere, then it's time to
refer to the (mostly vapourware) guides on creating your own terms/vocabs.

I suppose the modelling process is essentially the same as that found in
other fields like object modelling, E-R or RDBMS schema creation. With a bit
of luck large chunks of the numerous guidelines for those could be
transmogrified into RDF design patterns.

Cheers,
Danny.

Received on Tuesday, 5 November 2002 08:45:19 UTC