[open-literature] Open Correspondence - changes and developments

Thu Nov 4 16:49:29 UTC 2010

Evening,

I've been thinking about the Open Correspondence project and just 
wondered if there were any comments, especially on the proposed changes 
to the letters ontology so that better data can be defined from it.

*Ontology changes*

One of the reasons behing setting up the ontology was defining the 
various book relationships so that you could see how a particullar title 
was being discussed in the letters and to whom. For various reasons, the 
initial idea was simplified to just textReferred but I'm seriously 
considering resplitting the field into the original two fields. I'm 
suggesting that the element becomes something along the lines of:
_authorTextReferred_ - refer to a work that the author wrote, such as 
Dickens talking about "David Copperfield" to a correspondent.
_textReferred_ - referring to a work that the author didn't write but is 
writing about, such as when Dickens is talking about Leigh Hunt's work 
to George Cattermole.

There are also other groups mentioned in the text such as the "Reform 
League" which are political entities so does any one have the best idea 
as to how to represent this in RDF - perhaps another element is needed 
to letter along the lines of politicalEntity?

One of the new pieces that I'd like to develop is a new set of links to 
do with the works of Dickens so that the authorTextReferred would refer 
to an entity with the publication details of the book and an abstract 
should one exist (or can be gleaned from dbpedia). I know the Dublin 
Core would take care of most of this but one query is how to best model 
the two publication dates that exist for Dickens's work. Does it matter 
that both magazine and book publication dates would be modelled as 
dc:date but would that cause a issue when querying the endpoints when 
both are in operation since the magazine edition came out before the 
book but does have a bearing on publication (or should this also be 
another entity?).

The next major piece of parsing (and this feeds into the XML) is getting 
the geographical data from the letter headings such as "Gad's Hill" and 
representing those using a geographical ontology to open the data up and 
start playing with it in maps as per the discussion on the okfn-help list.

The final one would be getting the people referred in the letters 
(whether they are fictional or real but I think that that is a much 
larger job and would require another extension to the ontology but 
personReferred and Character might be useful at the moment) and the 
above paragraphs will take some time to complete.

*Storage*

As per the okfn-discussion and earlier emails, I think that the site 
needs to store the endpoint data in something like Couch. Rufus, I know 
that we've mooted it but are there any issues for the server doing this? 
It might help with getting the timeline loading a little more quickly. 
Could we use something like GeoCouch should it be suitable for the 
geographical data? (I've yet to play around with it or the data so just 
mooting at the moment.)

I'm sure that any bugs or issues might be found whilst using the data 
but if anyone notices one, please do let me know (if we haven't popped 
it up on the trac instance 
(http://knowledgeforge.net/shakespeare/trac/)). I know that there were 
mentions of work being done on the Open Correspondence stuff - before I 
really start getting back into the parsing, is anyone doing anything to 
it so I know what to avoid or if we need to give any one access to 
commit changes that we don't know about?

Just trying to tidy up the current state of the project and get any 
feedback if there is any one using the site. I'm probably starting the 
work on Saturday but I'd be grateful for thoughts, comments.

Thanks,

Iain

Iain Emsley

mob: 0772 399 2492
skype: iainemsley
blog: austgate.co.uk
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.okfn.org/pipermail/open-literature/attachments/20101104/6ff82436/attachment.html>